\'

THE JOURNAL

OF

ANIMAL BEHAVIOR

VOLUME 7, 1917

EDITORIAL BOARD

Madison Bentley Edward L. Thorndike

University of Illinois Teachers College, Columbia University

Gilbert V. Hamilton Margaret F. Washburn

Frazysburg, Ohio Vassar College

Samuel J. Holmes John B. Watson

The University of California The Johns Hopkins University

Walter S. Hunter William M. Wheeler

The University of Kansas Harvard University

Harvey A. Carr, The University of Chicago
Editor of Reviews

Robert M. Yerkes, University of Minnesota
Managing Editor

Published Bi-monthly

at Cambridge, Boston, Mass.

HENRY HOLT & COMPANY

34 West 33d Street, New York

G. E. STECHERT & CO., London, Paris and Leipzig, Foreign Agents

Entered as second-ciass matter March 7, 1911, at the post-office at Cambridge, Boston,
Massachusetts, under the act of March 3, 1879

CONTENTS OF VOLUME 7, 1917

Number 1, January-February pages

Hess, Carl von. New experiments on the light reactions of

plants and animals 1-10

Yerkes, Robert M. Methods of exhibiting reactive tenden-
cies characteristic of ontogenetic and phylogenetic stages. . 11-28

Reese, A. M. Light reactions of the crimson-spotted newt,

Diemyctylus z'iridcsccns 29-48

Hunter, Walter S., assisted by Yarbrough, Jos. U. The in-
terference of auditory habits in the white rat 49-65

Lashley, K. S. The criterion of learning in experiments with

the maze 66-70

Cole, William H. The reactions of Drosophila ampelophila

Loew to gravity, centrif ligation, and air currents 71-80

Olmsted, J. M. D. Geotropism in Planaria maculata 81-86

Financial statement for 1916.

Number 2, March-April

Yarbrough, Joseph U. The delayed reaction with sound and

light in cats 87-1 10

Utsurikawa, Nenozo. Temperamental differences between

outbred and inbred strains of the albino rat 111-129

Hubbert, Helen B., and Lashley, K. S. Retroactive asso-
ciation and the elimination of errors in the maze 130-138

Lashley, K. S. A causal factor in the relation of the distribu-
tion of practice to the rate of learning 139-142

Rau, Phil. The courtship of Pieris protodice 143-144

Number 3, May-June

Carr, Harvey. The distribution and elimination of errors

in the maze 145-159

Reeves, Cora D. Moving and still lights as stimuli in a

discrimination experiment with white rats 160-168

/3/73

CONTENTS iii

PAGES

Pearce, Binnie D. A note on the interference of visual

habits in the white rat 169-177

Lashley, K. S. Modifiability of the preferential use of the

hands in the rhesus monkey 178-186

Baldwin, Francis Marsh. Diurnal activity of the earthworm 187-190

Number 4, July-August

Stephens, T. C. The feeding of nestling birds 191-206

Hussey, Roland F. A study of the reactions of certain birds

to sound stimuli 207-219

Schaeffer, A. A. Choice of food in ameba 220-258

Carr, Harvey. Maze studies with the white rat. I. Normal

animals 259-275

Number 5, September-October

Carr, Harvey. Maze studies with the white rat. II. Blind

animals 277-294

Carr, Harvey. Maze studies with the white rat. III. Anos-
mic animals 295-306

McColloch, James W. and Yuasa, H. Notes on the migra-
tion of the Hessian fly larvae 307-323

Shadall, Elsa. Reactions of.Opalina ranarum 324-333

Yoakum, C. S. Similar behavior in cow and man with a note

on emotion 334-337

Peterson, Joseph. Frequency and recency factors in maze

learning by white rats 338-364

Carr, Harvey. The alternation problem. A preliminary
study 365-384

Announcement to subscribers 385

Number 6, November-December

Wells, Morris M. The behavior of limpets with particular

reference to the homing instinct 387-395

Taliaferro, W. H. Literature for 1916 on the behavior of

the lower invertebrates 396-404

Turner, C. H. Literature for 1916 on the behavior of

spiders and insects other than ants 405-419

iv CONTEXTS

PAGES

Wells, Morris M. Literature for 1916 on ants and Myrme-

cophils 420-434

Vincent, Stella B. Literature for 1916 on the behavior of

vertebrates 435-443

Craig, Wallace. On the ability of animals to keep time with

an external rhythm 444-448

Carr, Harvey. Smith's " Mind in Animals " 449-450

Strong, R. M. Wood's " The Fundus Oculi of Birds " 451

Carr, Harvey. Holmes's "Animal Behavior " 452-453

Walcott, Charles D. Story of Granny, the mountain squirrel 454-455
Announcement to subscribers 456

Subject and Author Index

VOLUME 7

Original contributions are marked by an asterisk (*)

Adams, C. C. Ecology of invertebrates,
411, 415.
*Albino rat, temperamental differences

in, 111.
Allard, H. A. Flashing in the firefly,

413, 415, 446, 448.
Allee, W. C. Kheotaxis in Asellus, 396,

402.
Allen, B. M. Behavior in the spiny
lobster, 396, 402.
*Alternation problem, 365.
Amans. Locomotion in the cicadas, 413,
415.
*Ameba, choice of food in, 220.
Amphibians, literature on, 435.
*Animals, reactions of, 1.
*Anosmic animals, maze studies with,
295.
Ant, literature on, 420.
Ant pest, control of, 423.
Arlitt, Ada H. Behavior in the chick,

441, 442.
Ashworth, J. H. Hibernation in the fly,

411, 415.

Back, E. A. Fruit fly, 414, 415.
Bagg, Halsey J. Individual differ-
ences, 441, 442.
Baker. The green apple aphis, 406, 407,

409, 416.
*Baldwin, Francis M. Diurnal activity
of the earthworm 187.
Barber, Ernest R. The Argentine ant,

420, 421, 422, 423, 425, 433.
Barber, H. S. Migration of Myrapods,

412, 416.

Barbey, A. Behavior in the beetle Cer-

ambyx heros, 406, 416.
Beebe, C. W. Fauna, 433.
Belsing, S. W. Behavior in the pecan

twig-girdler, 410, 416.
Bingham, H. C. Reactions of the bird
dog, 439, 442.
*Bird, feeding in the, 191;

^reactions of the, to sound stimuli,

207;
literature on, 436.
Blackman, M. W. Habits of Pityogenes
hoplinsoni Ewaine 416.

Blair, K. G. Luminous insects, 446, 448.

*Blind animals, maze studies with, 277.

Bovie, W. T. Schumann rays, 396, 402.

Brittain, W. E. Feeding habits of

Bepressaria heraclina, 406, 416;
fly as a disease carrier, 413, 416.
Brun, R. Orientation in the ant, 427,

433.
Buddenbrock, W. von. Tropism theory,

396, 402.
Burtt, Harold E. Behavior in the white

rat, 440, 442.

Cary, L. B. Sense organs of Cassiopea
xamachana, 396, 402.
*Carr, Harvey. Distribution and elimina-
tion of errors in the maze, 145;
*maze studies with normal white rats,

259;
*maze studies with blind white rats,

277;
*maze studies with anosmic white

rats, 295;
*the alternation problem, 365;
*Smith's "Mind in Animals," 449;
*Holmes's "Animal Behavior," 452.
*Cat, delayed reaction in, 87.
Chidester, F. E. The painted turtle,

416.
Churchill, E. P., Jr. Learning in the

goldfish, 440, 442.
Clausen, C. P. Behavior in Coccinelli-

dae, 406, 416.
Coad, B. R. Hibernation in the weevil,

411, 416.
Cockle, J. W. Habits of Lepidoptera,
416.
*Cole, William H. Beactions of Droso-
phila, 71.
Cooke, Mills W. Bird migration, 438,

442.
Cory, E. N. The Columbine leaf miner,

406, 416.
Cosens, A. Hibernation in the lady-bird

beetle, 411, 416.
Cotton, E. C. Life history of the Amer-
ican fever tick, 416.
Coupin, H. Paper-making insects, 416.
*Courtship, in Pieris protodice, 143.

VI

INDEX

Cowan, Edwina A. Behavior in the

chick, 441, 442.
Craig, Wallace. Rhythmic activities of
animals, 437, 442, 446, 448;
*rhythm in animals, 444.
Crawley, W. C. Ants, 421, 425, 427,
428, 433;

notes on Myrmecophily, 424, 433.
Crozier, W. J. Behavior of Holothuria
captiva, 396, 402;
pulsation of the cloaca of Holothur-

ians, 396, 402;
behavior of the barnacle Conchoderma

virgatum, 397, 402;
coloration of nudibranchs, 397, 402 ;
chemical sense in vertebrates, 435,
442.
Cummins, B. F. The louse as a disease

spreader, 413, 416.
Cushman, E. A. Behavior in the apple

red-bug, 406, 416.
Cutaneous sensitivity, literature on, 435.

Davis. Life history of Corpus uni-
punctata, 406, 416.
*Delayed reaction, in the cat, 87.
Demuth. Wintering of the bee, 414,

418.
Dolley, W. L. Reactions to light in

Vanessa antipoa, 416.
Donisthorpe, H. A new species of ant,

426, 433.
Donisthorpe, J. K. Literature on
Myrmecophily, 424, 425, 426, 427.
433.
Dove, W. E. Hibernation in the house-
fly, 411, 416.
Dow, R. P. Insect burrows, 415, 416.
*Drosophila, reactions of, 71.
DuPorte, E. Melville. Death feigning
in Tychius picirostris, 412, 416.

* T_> arthworm, diurnal activity in, 187.
JLv Ecology, literature on, 411.
Essig, E. O. A coccid-feeding moth,

406, 416.

Esterly, Calvin O. Feeding habits of
copepods, 397, 402.

Evans, A. T. Breeding in the house-
fly, 409, 416.

* C* eeding, in nestling birds, 191 ;

literature on, 406.
Felt, E. P. Habits of the codling moth,

407, 416.

Fish, literature on, 435.

Fitzsimmons, F. W. The house-fly and

disease, 413, 416.
Fletcher, John A. Behavior of the

chick, 441, 442.

*Fly larvae, migration in, 307.
Forbes, S. A. Habits of the' Northern

corn root-worm, 416.
Foucher, G. Orthoptera, 416.
Fracker, S. B. Behavior in Lach-

nosterna, 406, 418.
Frison, T. H. Habits of Psithyrus

variablis, 416.
Frost, S. W. Biological notes on Ceuto-

rhync.hus marginatus, 417.
Furness, William H. Mentality of the

monkey, 440, 442.

Gaige, F. M. Swarming in the ant,
428, 434.
*Geotropism, in Planaria maculata, 81.
Gibson, Arthur. Control of ants, 424,

434.
Godderham. Feeding of Depressaria

heraclina, 406, 416.
Good, C. A. The apple maggot parasite,

413, 417.
Goodale, H. D. Behavior of the capon,

439, 442.

Graham-Smith, G. S. Parasites of the

common fly, 413, 417.
Grave, C. Feeding in the oyster, 397,

402.

*TJabit, in the white rat, 49.

*formation, in the white rat, 169.
Hamilton, G. V. Reactions of primates,

440, 442.

Harris, J. A. Habits of the beetle

Bruclius, 409, 417.
Hayes, W. P. Life history of the maize

bill-bug, 406, 409, 417.
Herrera, M. .Intelligence in the insect,

417.
Herrick, G. W. Life history of the

cherry-leaf beetle, 406, 417.
*Hess, Carl von. Reactions of plants and

animals, 1.
Hibernation, literature on, 411.
Hilton, W. A. Reactions of the rare

whip scorpion, 417.
Hindle, Edward. Flies and disease, 413,

417.
Hodge, C. F. Control of flies, 417.
Holloway, T. E. Phototropism experi-
ments, 405, 417.
Holmes, S. J. "Animal Behavior,"

452.
Homing instinct, in limpets, 387.
Horton. Anti-ant bands, 424, 428, 434.
Howat, I. Effect of nicotine on the

frog, 438, 442.
*Hubbert, Helen B. Elimination of

errors in the maze, 130.

INDEX

vn

Hungerford, H. B. Behavior of the
Sciara maggot, 406, 417.
*Hunter, Walter S. Habits in the white
rat, 49.
Hunting behavior, literature on 406.
*Hussey, Boland F. Reactions of birds
to sound stimuli, 207.
Hutchison, E. H. Mating in the house-
fly, 407, 417.
Hyslop, J. A. The host of Zelia verte-
brata, 417;
habits of Horistonotus uhlerii, 417.

* [ nbred rat, compared with outbred, 111.
1 Insect, literature on, 405.
Instinct, homing, 387;

literature on, 438.
Invertebrates, literature on, 396.

Johnson, H. M. Visual discrimination
in vertebrates, 436, 442.
Jordon, H. Nervous system of certain
holothurians, 397, 402.

Kanda, S. Geotropism in animals, 397,
398, 402.
Keilin, D. Studies on Diptera Larva, 417.
Keith, E. D. The ghost moth, 417.
Kellogg, F. M. Response of the earth-
worm, 400, 403.
Kellogg, J. L. Opinions on ciliary

activities, 397, 402.
Kempf, E. J. Behavior of a monkey,

440, 442;
learning in the monkey, 440, 442.
Kenoyer, L. A. Pollination in insects,

414, 417.
King, J. L. Life history of Pterodontia

flavipes, 413, 417.
King, W. V. Anopheles punctipennis

and disease, 413, 417.
Knab, F. Weevil larvae, 417;

behavior of Dermatobia hominis, 409,

417.

Lankester, E. R. Behavior of Fora-
minifera, 398, 402.
*Lashley, K. S. Learning in maze ex-
periments, 66;
*elimination of errors in the maze,

130;
*relation of practice to rate of learn-
ing, 139;
*use of the hands in the rhesus mon-
key, 178;
free-swimming Paramoecia, 399, 403;
color vision in the bird, 436, 443;
homing activities in the bird, 438, 443.
Laurent, Philip. Flashing in the fire-
fly, 446, 448.

*Learning, relation of practice to rate
of, 139.
Legendre, J. The mosquito, 417.
Leng, C. W. Notes on Cychrissi, 417.
Letisimulation, literature on, 412.
Lewis, E. M. Relation of body tempera-
ture to that of an animal's environ-
ment, 400, 404.
*Light, reaction to, in the cat, 87.
Limpet, behavior in, 387;

literature on, 387.
Loeb, J. Heliotropic reactions of ani-
mals and plants, 398, 403.
Lohner, L. Feeding experiments on the

leech, 398, 403.
Lyne, W. H. Life history of the cod-
dling moth, 417.

Mammals, literature on, 435.
Mann, W. M. The Brazilian ant,
421, 425, 426, 432, 433, 434.
Marlatt, C. L. House ant, 420, 424, 434.
Mast, S. O. Feeding of Amoeba on In-
fusoria, 399, 403;
feeding of Amoeba on Rotifers, 399,

403;
free-swimming Paramoecia, 399, 403 ;
orientation in Gonium pectorale, 399,
403.
Maternal behavior, literature on, 409.
Matheson. Life history of the cherry

leaf beetle, 417.
Mating, literature on, 407.
Maupas, E. Copulation of nematodes,

399, 403.
Mayer, A. G. Nerve conduction in Cas-
siopea, 399, 403.
a theory of nerve conduction, 399, 403.
*Maze, experiments with, 66;
*elimination of errors in, 130;
*distribution and elimination of errors
in, 145;
* studies with the white rat, 259, 277,
295;
*learning, in the white rat, 338.
McAfee, W. L. Behavior of tiger beetle,

425, 434.
*McColIoch, James W. Migration of fly

larvae, 307.
MeDermott, F. A. Flashing in the fire-
fly, 445, 446, 448.
McGregor, E. A. The privet mite, 406,

409, 417.
Mclndoo, N. E. The coccinellid beetle,

412, 417.
Memory, literature on, 414.
Mendelssohn, M. Behavior of the leuco-
cyte, 399, 403.
Metalnikov, S. Behavior of Protozoa,
399, 403.

Vlll

INDEX

Meyers, Gr. C. Learning in the pig, 440,
' 443.

Migration, literature on, 412.

Miller, J. M. Behavior of Megastigmus

spermotropus, 409, 418.
•Monkey, use of the hands in, 178.

Montane, L. A Cuban chimpanzee, 439,
443.

Moore, A. R. Response of the earth-
worm, 400, 403;
orientation in Gonium, 400, 403.

Morse, E. S. Flashing in the firefly,
413, 417, 446, 448.

Muller, H. R. Falling reflex in the cat,
437, 443.

Myrmecophils, literature on, 420.

Nesbit, W. Behavior of wild animals,
442, 443.
Xewman, H. H. Behavior of Phalangi-
dae, 444, 448.
*Newt, light reactions of, 29.
Xininger, H. H. Life history of the
bee, 418.

*t \ Imsted, J. M. D. Geotropism in
\J Planaria maculata, 81.
*Opalina ranarum, reactions of, 324.
Orchymont, A. de. Respiration in the

insect, 413, 418.
Orientation in the ant, 427.
Osborn H. Life history of the leaf

hopper, 406, 411, 418.
Osburn, R. C. Migration in the dragon-
fly, 412, 418.
*Outbred rat, compared with inbred, 111.

Packard, C. W. Life history of the
Hessian fly parasite, 413, 415, 418.

Paddock, F. B. Observations on the
turnip louse, 406, 418.

Parker, G. H. Reactions and structure
of the sea anemone, 400, 403 ;
structure of Metridium, 400, 403.

Parker, R. R. Migration of the house-
fly, 413, 418.

Patch, E. M. Ecology of the aphid, 412,
418.

Patten, B. M. Effect of age on the
blowfly larva, 418.

Payne, O. G. M. Life history of Tele-
phones literatus, 413, 418.

Peairs, L. M. Behavior of web-worm

larvae, 445, 448.
*Pearce, Binnie D. Interference of vis-
ual habits in the white rat, 169.

Pellett, F. C. Habits of Polistes metri-
cus, 410, 418.

Pemberton. Effect of temperature on
the fruit fly, 414, 415.
*Peterson, Joseph. Factors in learning
by the white rat, 338;
tone perception in the rat, 435, 443;
completeness of response, 441, 443.
Phillips. Outdoor wintering of the bee,
414, 418.
Pictet, A. Locomotion in the insect,

405, 418.
Pierce, W. D. Habits in the weevil,
409, 41S;
effects of temperature on the insect,
414, 418.
*Pieris protodice, courtship of, 143.
-Planaria maculata, geotropism in, 81.
*Plants, reactions of, 1.

Rabaud, E. Death-feigning reflex in
the insect, 400, 403.
Rasmussen, A. T. Hibernation, 439

443.
Rau, Xellie. Behavior of the solitary
bee, 408, 409, 415, 418;
biology of the mud-dauber wasp,
409, 413, 418;

sleep in the insect, 414, 418.
*Rau, Phil. The courtship of Pieris pro-
todice, 143;
behavior of the solitary bee, 408,

409, 415, 418;
biology of the mud-dauber wasp, 409,

413, 418;
sleep in the insect, 414, 418.
*Reactive tendencies, methods of ex-
hibiting, 11.
*Reese, A. M. Light reactions of the

crimson-spotted newt, 29.
*Reeves, Cora D. Discrimination experi-
ment with the white rat, 160.
Reuter. Effect of sound on animals,

435, 443.
Rhythm, ability of animals to keep time
with, 444;-
literature on, 444.
Richardson, C. H. Attraction of Dip-
tera to ammonia, 405, 418 ;
chemotropie response of the house-
fly, 405, 418.
Roberg, D. N. The family Phoridae and

disease, 413 418.
Robinson, A. Behavorism, 442, 443.
Roeber, J. Vision in insects, 418.
Rogers, C. G. Relation of temperature
of animals to that of their en-
vironment, 400, 404.
Rohwer, S. A. Mating of the saw-fly,
407, 418.

INDEX

IX

Root, F. M. Feeding of animals on
rotifers, 399, 403;
feeding of animals in Infusoria, 399,
403.
Euggles. Insects and disease, 413, 419.
Eunner, G. A. The cigarette beetle,
418;
effects of roentgen rays on the beetle,
405, 418.

Sanders, J. G. Behavior in Lachno-
sterna, 406, 418.
Satterwait. Life hifctory of Corpus

unipunctata, 406, 416.
Sayle, M. H. Eeactions of Necturus,
436, 443.
*Schaeffer, A. A. Choice of food in
Ameba, 220;
feeding in Ameba, 400, 404;
behavior in Ameba, 401, 404.
Schoene, W. J. Feeding in the seed-
corn maggot, 406, 418;
biology of the P. brassicae, 406, 418.
Seurat, L. G. Copulation of nematodes,

399, 403.
Sell, E. A. Notes on the 12-spotted
cucumber beetle, 407, 411, 414, 419;
migration in the beetle, 419;
behavior of a Western flower beetle,
419.
*Shadall, Elsa. Eeactions of Opalina
ranarum, 324.
Shannon, H. J. Migration in insects

and birds, 413, 419.
Sleeping behavior, literature on, 414.
Smith, E. M. < ' Mind in Animals, ' ' 449.
Smith, H. S. Habits of parasitic Hy-

menoptera, 409, 419.
Smith, M. E. South Carolina ants, 425,

, 428, 433, 434.
Snyder, T. E. Behavior of termites,
408, 409, 419, 430, 432, 434;
white ants, 421, 434.
Somes, M. P. Mating in Solarium caro-
linensis L., 407, 419.
*Sound, reaction of the cat to, 87;
*reaction of the bird to, 207;
literature on, 435.
Spider, literature on, 405.
Squirrel, a mountain, 454.
*Stephens, T. C. Feeding of nestling

birds, 191.
*Strong, E. M. Wood's "The Fundus
Oculi of Birds," 451.
Studhalter. Insects and disease, 413,

419.
Swarming, in the ant, 428.
Swift, W. B. Developmental psychology
in lower animals, 442, 443.

Swynnerton, C. F. M. Experiments on
insects, 419.

* r T* aliaf erro, W. H. Behavior of the
1 lower invertebrates, 396.

Technique, literature on, 415.

Theobald, Fred V. Literature on aphi-
didae, 425, 434.

Tillyard, B. J. Behavior in the dragon-
fly, 419.

Titus, E. G. Structure of Metridium,
400, 403.

Torrey, H. B. Physiological analysis of
behavior, 401, 404.

Tower, D. G. Feeding in Cirphis uni-
punctata, 407, 419.

Townsend, C. H. T. Insects and dis-
ease, 413, 419.

Tropisms, literature on, 405.

* Turner, C. H. Behavior of spiders and

insects, 405;
behavior in the spider, 407, 410, 419;
mating in Lasius niger L., 408, 419;
feeding in the false spider, 419.
Turner, C. L. Breeding in Orthoptera,

408, 409, 415, 419.
Turner. Feeding in the green apple
aphis, 406, 407, 409, 416.

Urbahns, T. D. Life history of Haoro-
cytus medicagnis, 409, 419.
*Utsurikawa, Nenozo. Temperamental
differences in albino rats, 111.

Vertebrates, literature on, 435.
*Vincent, Stella B. Behavior of
vertebrates, 435.
*Vision, in the white rat, 169;
literature on, 436.

* 11 7 alcott, Charles D. Story of Granny,
VV the mountain squirrel, 454.

Walton, A. C. Eeactions of Paramoe-
cium caudatum to light, 401, 404.

Warren, A. Feeding in the Hawaiian
dragonfly, 406, 419.

Wasteneys, H. Heliotropic reactions of
animals and plants, 398, 403.

Watson, J. B. Spectral sensitivities of
birds, 437, 443;
the conditioned reflex, 437, 443.
homing activities of birds, 438, 443.

Watson, J. E. Life history of Anti-
carsia gemmatiUs, 406, 407, 419.

Weed, L. H. Falling reflex in the cat,
437, 443.

Wiedman, F. D. Parasitism in the mon-
key, 413, 419.

Welch, P. S. Behavior in Lepidoptera,
411, 412, 419.

X

INDEX

*Wells, Morris M. Behavior of limpets,
387;
*literature on ants and Myrmeeophils,

420.
Wenrich, D. H. Reactions in the mol-

lusk. 401, 404.
"Wheeler, W. M. The tropical ant Pliei-

dole feregrina, 423, 434;
the Indian ant Triglyphothrix stria-

tidens, 423, 434;
the "Western Amazon ant, 428, 430,

434;
marriage flight of the bull-dog ant,

429, 434;
the Australian ant, 434;
a blind worker ant, 434;
a phosphorescent ant, 434;
behavior in Phalangidae, 444, 447,

448.
•White rat, auditory habits in, 49;

•discrimination experiments with,

160;
-vision in, 169;

*maze studies with the normal, 259 ;
*ma'ze studies with the blind, 277;
*maze studies with the anosmic, 295;
*maze learning in, 338.
Whitmarsh, E. D. Life history of Apa-

teticus cynicus, 406, 409, 419.

Williams, F. X. Life history in Meth-
oca stygia, 410, 419.

"Willis, 11. 8. Behavior in Amoeba, 402,
404.

"Winn, A. F. Heliotropism in the but-
terfly, 405, 419.

Wood, Casey A. ' ' The Fundus Oculi
of Birds," 451.

Yarbrough, Joseph U. Auditory habits
in the white rat, 49;
•delayed reaction in cats, 87.
Yerkes, Ada W. Behavior in the albino

rat, 441, 443.
•Yerkes, Robert M. Methods of exhibit-
ing reactive tendencies, 11 ;
mental life of the monkey, 439, 443;
ideational behavior in man and ani-
mals, 440, 443;
provision for the study of moneys and
apes, 440, 443.
•Yoakum, C. S. Similar behavior in cow

and man, 334.
•Yuasa, H. Migration of the fly larvae,
307.

'Vetek, J. Insects and disease, 413, 419.

JOURNAL OF ANIMAL BEHAVIOR

Vol. 7 JANUARY-FEBRUARY No. 1

NEW EXPERIMENTS ON THE LIGHT REACTIONS
OF PLANTS AND ANIMALS'

CARL VON HESS

I

Gentlemen: Allow me, before the order of the day, to give
a brief report of a discovery, which, though it stands only in
loose relation to our theme, seems to me of general interest. I
speak of the accommodation of the alciopids.

The alciopid is, as you know, a nearly transparent pelagic
annelid, whose comparatively highly developed eyes have been
repeatedly the object of histological research. It was believed
that muscular elements could be demonstrated anatomically in
these eyes and on this supposition theories of the act of accom-
modation were grounded. These theories can easily be proved
mistaken, as the following shows; I will not further dwell on them.

On account of the small size of the eyes the largest — which
I was enabled to examine had a diameter of hardly 1 mm., — it
had been considered heretofore impossible to attack the prob-
lem of their accommodative changes experimentally.

However, I was able, by laying the living and carefully iso-
lated eyes on suitable electrodes, under seawater, and by ob-
serving them through binocular lenses in a very strong light
falling from above, to follow the changes caused by electric
irritation, and thus to discover a most remarkable accommo-
dative process, unique in the animal world.

At another time I shall describe this process in detail, tonight

I shall limit my description to the main facts.

1 A lecture before the Morphological Society of Munich, reported and trans-
lated by Miss Hilda Lodeman.

2 CARL VON HESS

If one views a fresh alciopid eye from in front, the surface
surrounding the lens is seen to be threaded over with numerous
fine silvery shining stripes, which have hitherto been mistakenly
interpreted as muscles (Hesse). In fact these are structures
which, like an iris, obstruct the passage of diffuse light into
the eye; besides this, they make the eyes which are turned
forward and downward, as invisible as possible to an enemy
coming from below. They thus have the same effect as that
which I some time ago proved to be the case with the silver
sheen of fish.

Just below the lens there is a spot in the very soft eye-wall
which, one may observe, contracts when the eye is stimulated;
all the other portions of the tegument remain motionless. The
lens, when stimulated, moves forward perceptibly, it approaches
the cornea, as one may perceive most readily by looking at the eye
in profile. Herewith is proved that the alciopids have an active
near accommodation; for, by the above-mentioned contraction
the distance between the lens and the retina is increased, while
the lens remains unchanged in form. The way in which the
change in the location of the lens is brought about is most inter-
esting: The alciopids are distinguished from all other animals
with otherwise similarly constructed eyes, by possessing a double
vitreous body. Directly back of the lens we find a viscous
fluid which is distinctly separated from the posterior space of
the vitreous body and adheres closely to the w T alls of the eye
on all sides. At the lowest point of this front part of the vit-
reous body the latter displays a curious ampulliform knob which
is connected with the eye-water space by a canal and was form-
erly interpreted as an auditory sac by zoologists, and at present
is supposed to be a gland belonging to the vitreous body for the
secretion of its substance. My experiments show the real use
of this protuberance. It occupies exactly the spot in the eye-
wall in which alone contractile elements are found; the muscles,
contracted, press the lump together like a rubber bulb filled with
liquid, thus forcing a part of its contents into the eye, and
slightly pushing forward the lens which rests in a bowl-like
groove in the front surface of the vitreous body.

This is the second accommodative process among the inver-
tebrates with which we have become acquainted ; the mechanism
differs entirely from that which I have proved Cephalopods to

NEW EXPERIMENTS ON LIGHT REACTIONS 3

possess. Our observations teach anew how greatly physiologi-
cal experiment can aid us in the interpretation of histological
discoveries.

II

Among the lightreactions of Echinodermata which I have
newly discovered and upon which I shall make only a brief
report tonight, a certain interest attaches to those of the star-
fish, if for no other reason but that until now almost nothing of
their sensitiveness to light was known. On the ground of
anatomical research it was taken for granted that the familiar
red points at the ends of their five arms were light receiving
apparatus. Attempts to elucidate the question experimentally
led to contradictory results. Some authors assert that those
starfish which have an inclination to move toward the light
cease showing this impulse after the tips of the arms with the
" eyes " are cut off; according to other writers individuals thus
mutilated still crawl to the light.

In the course of systematic experiments I discovered the
surprising fact that the feet of the Astropectinids are highly sensi-
tive to light. If light is flashed on them their little feet, relaxed
in the dark, are instantaneously jerked in and the widely opened
ambulacral groove is closed along the whole of the lighted area,
the flanking white spines shutting over the incurled little feet.
This startling phenomenon, which I was able to record in a
number of snapshots, gave me the opportunity of examining
the differing effects of colored lights. As with all the hitherto
thoroughly examined invertebrates, it was found that colored
lights have similar or identical relative values for our starfish
as for the totally color-blind human eye; red lights remain
almost or quite without effect even when very strong, while
green and blue lights have a much stronger effect than the red
lights, even when the latter seem to our normal eyesight much
darker than the former. I was able also to prove adaptive
changes in these starfish and to carry out exact measurements
during my observations.

New and most remarkable light reactions in sea urchins were
also disclosed. So far it had been known from experiments of
Sarasin and Uexkull that some sea urchins raise their spine
slightly when shaded from the light. More exact observations
of their qualities of sight had not yet been made. I discovered

4 CARL VON HESS

the following interesting phenomenon appertaining to Centro-
stephanus longispinus. The animals have surrounding their abo-
ral pole, 20 or 30 beautiful lilac colored, clublike processes about 3
mm. long, concerning which we knew hitherto only that they some-
times move in rotation, at other times are quiescent. I noticed
that if a specimen at rest was slightly shaded, for example, if
one's hand were passed quickly between window and reservoir,
the little clubs began to rotate in a most lively manner. Further
experiment showed that in order to bring about such agitation
an exceedingly slight lessening of the lighting suffices. If, for
instance, the greater part of the light reaches the animal from
a gray pasteboard held at the proper angle, and I replace this
board with one which is only a little darker in shade, the clubs
begin to rotate quickly. Even with this method it was possible
to a certain degree to make determining measurements, and I
was able by the further use of differently colored boards for the
lighting again to show convincingly that these animals also
behave like totally color-blind human beings brought under cor-
responding conditions. Still more delicate, surprisingly exact
measurements were made by using the method which I shall
now describe.

Ill

Several writers have thought to deduce an argument against
the experiments I have so far made with the qualities of sight
in animals from the idea that I bring the " objective light-reac-
tion " of animals into relation with the " subjective light sensa-
tion " of man. For anyone to whom the science of color is fami-
liar, this argument is easily controverted. Still it is evident
that there is a great advantage in showing that the problem may
be attacked from quite a new direction. Therefore in a new
series of experiments on a large scale, I brought the light sensi-
tiveness of animals into relation, not to the ' ' subjective light
sensation " of human beings, but to the " objective light reac-
tion " in the human eye, to the changes in the size of the pupil
caused by light. This correlation was successful after I had
made extended and rather difficult preparations, as follows:

We know from former experiments of M. Sachs (1893) that
the degree of contraction of the pupil caused by a colored light,
the "motor irritative value" of a colored light, depends on the
strength of luminosity in which the colored light is seen. Until
now we lacked a practical method of comparing the changing

NEW EXPERIMENTS ON LIGHT REACTIONS 5

size of the human pupil and the varying reactions to light in
the lower animals. Here you see an apparatus 2 which I con-
structed for this purpose and which does excellent service also
in examining physiological and pathological changes in the
human pupil. Of this use of the instrument I shall speak else-
where in detail. At present it shall be described only in so far
as it serves in the solution of the problems in comparative physi-
ology now before us. With the aid of a proper system of lenses,
and placed at a certain distance from it, a Nernst lamp illu-
mines very strongly and evenly a circular space. In front of
the first lens there is a movable double frame which by a lever
arrangement enables one to light this circular space first by a
physically exactly determined colored glass light, and imme-
diately thereafter, without intermediate lighting, by a mensur-
able variable light of almost colorless gray, for comparison.
The change in the strength of light in the gray field is caused
by the sliding in opposite directions of two acute-angled prisms
of gray glass. For every position of the latter, the amount of
light which penetrates it from the Nernst lamp is determined;
this amount will be expressed in the following table in per-
centages of the strength of the Nernst light. With this appa-
ratus, which can be used for many purposes, I have made a
large number of measurements; if I give only a brief summary
of these, please do not conclude a correspondingly brief period
of labor on this subject; the table below is alone the result of
over 1,000 separate measurements.

Motor Irritant Values of Colored-glass Lights

The numbers give the amount of light allowed to fall through the gray prisms
in percentages of the whole amount striking these, the motor equation determ-
ining the former amount.

If I I

S o§? 3 5

o ^i* -g o .» & 8 S S

£ ttSja H Q Z $ « U (U

Red 9-11 1.5-2.2 <0.6 7.3-9.3 0.9-1.1 <0.6 <0.6 <0.8 <1.0
Blue 1.5-2.5 2-3 9.9-11.8 0.8-0.9 7.4-8.8 9.3-11.1 8.3-11.1 11.1-14.8 8.3-14.8

2 This apparatus, "Differential Pupilloscope," is manufactured by C. Zeiss.

6 CARL VON HESS

I began with measurements of the normal human eye in order
to determine the average pupillomotor irritant value of the
various colored lights. Further measurements of relatively blue-
seeing, red-green blind (so-called red-blind), showed, as may be
seen in the table, that a very slight irritant value of red, and a
hardly perceptible variation from the normal motor-irritant value
of blue, are characteristic of this disturbance of the sense of
sight. For the sake of brevity I shall limit myself in the fol-
lowing to- the discussion of the red and blue values, these being
of the greatest importance to us. In two cases of totally color-
blind which I have repeatedly examined, red proved to have a
very slight motor-reactive value (<0.6), blue, a comparatively
high value of 9-11.8% (as compared to 1.5-2.5% in the normal
eye). These are the three principal kinds of pupil reactions
which occur among normal and color-blind human beings and
with these we must compare the motor reactive values found
among the different animals.

For the day bird, the sensitive value of red is like our own;
this corresponds to the fact which I had already discovered by
another method, that day birds in most cases see red lights
nearly or quite as we see them. The relatively small values
of blue, — they are the smallest which I have met with in the
animal series — correspond to another fact which I had dis-
covered, namely, that day birds in consequence of red and
yellow oil globules located in front of the light receiving ap-
paratus, are relatively blue blind.

With the help of the apparatus I was enabled, among other
things, to answer the following question, which I raised some
time ago. The beautiful blue of the feathers of many birds is
interpreted by almost all zoologists as decorative color for the
attraction of the other sex: this interpretation assumes that
these birds see blue as we see it, that therefore the oil drops
do not exist. For if these drops are found in the eyes of these
birds as they are found in the hen and the dove, then a blue
which seems to us gorgeous must look to them blue-gray or
colorless gray. So far I have had no opportunity to examine
such birds with the spectrum according to the method described ;
but a short time ago I examined the movements of the pupil
of the Butterfly-finch (Mariposa phoenicotis) with the new ap-
paratus ; the motor values are the same as for chicken and dove ;

NEW EXPERIMENTS ON LIGHT REACTIONS 7

and herewith it is proved that the beautiful blue on breast and
tail of this bird cannot be for adornment.

Among the night birds I found the motor values like those of
the color blind human eye, a fact which corresponds to the
superior number of rods and cones in the retina of these birds.
The relatively slight differences are sufficiently explained by the
fact that in the retina of the night birds, the cones are not en-
tirely lacking as many assume; indeed, I have repeatedly been
able to count in such retinas one to two million cones, with
slightly colored oil balls.

Among the invertebrates, examination with the new appa-
ratus of the movements of the pupils of Cephalopods, which are
particularly well suited to the measuring experiments, shows as
you see striking conformity to the irritative values for the totally
color-blind human eye. By the use of other methods also, I
have been able to show that these invertebrates are totally
color-blind. I cannot here dwell on these new experiments.

A glance at the table will show you further that the motor-
sensitiveness of bees to colored lights, of mollusks (Psammobia)
and of sea urchins (Centrostephanus) is almost identical with
that of a color-blind man, whereas it differs characteristically
from that of red-blind eyes. The reaction of bees I need not
mention again, as the bees as well as fish and crabs may easily
be proved totally color-blind by other methods which I have
developed. The continually repeated mistaken assertions of a
few zoologists, from which a color sense in these animals is sup-
posed to be deducible, need no new refutation after the above
measurements are studied by anyone at all familiar with the
subject.

The advantages of the new methods of research which I have
here briefly indicated consist essentially in the following points:
All the light reactions which I have hitherto carefully investi-
gated in animals, the contraction of the pupils in birds and in-
vertebrates, the swimming of fish and crabs and the flying of
bees toward the light, the phenomena of retraction in Serpula
and Psammobia, the rotations of the little clubs in the Centro-
stephanus, etc., all these manifold movements which are caused
by increasing or lessening the light, we are able by the help of
our apparatus to measure with the identical, physically exactly
determined colored lights, and to express their motor-sensitive

8 CARL VON HESS

values in terms of one and the same measurable variable light
with which each colored light is compared. Besides this, we
are now in a position to bring all these reactions of animals in
relation to the motor-sensitive values which the same colored
lights have for the pupil of the normal, the red-blind, and the
totally color-blind human eye.

That it would be possible to carry out such exact measure-
ments by this new process, I myself could not foresee at the be-
ginning of these tests; as the results obtained coincide in every
detail with those of my former widely differing experiments,
they prove most satisfactorily the accuracy of the statements
I have previously made about the sight qualities of animals.

IV

The long well-known fact that, on plants, red lights have
comparatively slight, blue, on the contrary, strong heliotropic
effect, that therefore in this respect there exists a certain simi-
larity between the effect of colored lights on plants and on
animals, gave J. Loeb occasion to accept the " Identity of ani-
mal and plant heliotropism." Some time ago, referring to older
experiments made by Wiesner and to more recent ones by
Blaauw, I had expressed doubts of this theory; as in spite of this,
Loeb's followers have again energetically taken up the defence
of the identity of the two tropisms, it seemed to me advisable
to attack this interesting question with new methods. In order
to settle it so that every possible objection should be met, both
reactions must be studied under identical conditions, with the
same colored lights, and especially in quantitative experiments,
the same light for measuring must be used for both.

These conditions were fulfilled by the following procedure.
Etiolated seedlings of various kinds, in long narrow boxes, were
exposed on one side to the rays of a suitable Nernst-light spec-
trum and simultaneously from the opposite side, to the light
used for measurement and comparison, the latter being variable,
an electric light placed in a tunnel adapted to the purpose.
Its strength I varied partly by changing its position as required,
to distances nearer to or farther from the plants, and partly
by means of an episkotister. This method proceeds in the same
lines as those developed in my experiments with Artemia and
other animals which shun the light. Starting from a medium

NEW EXPERIMENTS ON LIGHT REACTIONS 9

distance found by preliminary trials, after a very few hours we
find the plants in red, yellow and green bent far over towards
the measuring light, those in green, blue and a part of violet,
towards the spectrum, those in the outer edge of violet and ultra-
violet again bent toward the measuring light. Through this
experiment we have found two lights in the spectrum whose
heliotropic strength is equal to that of the composite light.
The circumstance that the plants bend over on each side of these
two colors in opposite directions make a comparatively exact
spectroscopic determination of their respective wave lengths
possible. By repeating such experiments, taking different dis-
tances of the lamp from the plants, I obtained each time two new
points for the construction of curves. You see here the curve
of the motor irritative values of the different lights of the spec-
trum for the invertebrates, next to it the curve of some among
the plants (Brassica napus) which I have observed and you can
see from these that there can be no question of identity between
the two results ; the curve for animals has its maximum in yellow-
green, with a wave length of about 526/*/*, that for Brassica napus
has its maximum in blue or in the beginning of violet, with a
wave length of about 475/*/*! In yellowish-green, where we find
the maximum for animals, the heliotropic effect on the plants
has already reached nearly its minimum.

A second method for the investigation of certain questions
occupying my attention, I worked out in this way: I have
already shown that one can obtain beautiful and convincing
results if a reservoir is lighted by rays reflected from colored
paper at both ends, and direct light from the window is shut
off by placing shades as required. Animals seeking the light,
without exception hasten to that end which is lightest in the
opinion of a color-blind individual quite irrespective of the way
in which normal sight interprets the values. The heliotropic
movements of plants have hitherto been observed only when
caused by light from the spectrum or through colored glasses;
it had never been attempted to find out whether heliotropic
movements appear also when light from such reflecting surfaces
alone is used. After I had found in a few introductory experi-
ments that such is in fact the case to a quite surprising degree,
I used this method for the solution of the problem before us. It
is one easily adapted to the use of the interested layman.

10 CARL VON HESS

If the tropisms were identical, the plants placed between the
colored papers should behave in relation to these in exactly the
same manner as animals under like conditions. If, however, the
heliotropism of plants differs from that of animals as much as
the curves indicate, then, if we carefully choose a green surface
and a blue, place animals and plants between the two, the
former will go to the green side and the plants will bend toward
the blue in exactly opposite directions. This behavior is indeed
quite marked as you see by the samples set before you. The
plants bend over to the blue often in one to two hours after
being placed in position.

I have taken the liberty of briefly introducing to you two new
methods for the investigation of the heliotropism of plants,
because I believe they may do good service in botanical experi-
ments and elsewhere, especially in quantitative experiments, and
because particularly the second method may easily be handled
by amateurs, and gives marked results, besides being well suited
to use in the lecture room. As to the pertinent scientific ques-
tions, these I have touched upon today only in so far as the
often repeated assertions of Loeb, that animal and plant helio-
tropism is identical, required a final refutation.

V

In conclusion, let me add a word on my discoveries about the
sight qualities of fish and invertebrates. Zoologists and botan-
ists have again and again declared they cannot acknowledge
my ' theories ' (as they call them) because they stand in too
harsh contradiction to the prevailing doctrines. The truth of
the matter is, that I have never set up any theory whatever,
but have made known only facts which every conscientious
observer may easily verify for himself. What Sprengel pro-
mulgated in 1793, and has been taught ever since about the
connection between the coloring of flowers and the visits of
insects, was a theory. This theory is now finally done away
with, for it is built upon demonstrably wrong surmises as to
the sight qualities of bees. Plant biology, for a hundred years
and more under the ban of this doctrine, which even Darwin
believed to be true, will now needs turn to the task of ascer-
taining the real meaning of the splendor of color in blossoms.

METHODS OF EXHIBITING REACTIVE TENDENCIES

CHARACTERISTIC OF ONTOGENETIC AND

PHYLOGENETIC STAGES

ROBERT M. YERKES

From the Harvard Psychological Laboratory

Methods which have contributed importantly to our knowl-
edge of the ontogeny and phylogeny of reactive tendencies, and
more especially to those types of adaptive behavior which we
call ideational, are few and unsatisfactory. Only recently have
experimental devices and procedures been suggested which are
alike suited to reveal the reactive tendencies of ontogenesis and
phylogenesis and to stimulate interest in genetic description of
behavior.

Following a brief historical sketch, I shall describe an appa-
ratus by means of which three of the most recent and promising
of our behavioristic methods may be used.

From the birth of interest in the problems of psychogenesis,
about the middle of the last century, until the end of the cen-
tury, no scientific means of approaching the problems of idea-
tional behavior 1 were developed. Romanes, Brehm, Morgan,
and their psychological contemporaries who happened to be
interested in evolutionary or genetic problems worked either
from anecdotal materials or from observations gathered by the
use of crude and unstandardized methods which may fairly be
characterized as wholly unsuited to scientific inquiry. We re-
gard their contributions to genetic psychology as suggestive of
possibilities of research or as defining problems rather than as
important additions to our knowledge of fact.

With the appearance of Thorndike's mental initiative, the

situation radically changed, for the puzzle-box or problem method

came into existence and began to be used systematically as a

1 1 shall designate as ideational behavior those forms of adaptive response which
in objective characteristics are identical with, or strikingly resemble, what we ap-
propriately and with common consent call ideational behavior in man.

12 ROBERT M. YERKES

means of testing for various types of behavior. Thorndike him-
self devised various forms of apparatus and problem, while at
the same time making them contribute most stirringly to our
knowledge of the psychology of the chick, cat, dog, and monkey.
Kinnaman, Small, Porter, Watson, and a host of other Ameri-
can and European experimentalists followed Thorndike's lead
in the application of experimental devices to the analysis of
problem-solving behavior.

It may not be amiss to point out that the puzzle-box method,
although an important advance scientifically over the casually
or inexactly arranged situations of the earlier period — not to
mention the anecdote — does not adequately fulfill the require-
ments of comparative and statistical method. True, it has
possibilities of adaptation or improvement in these respects
which have never been realized, but the fact is that mostly the
data of response to a puzzle-box problem or situation are so
meager and inexact as to be of scant value for purposes of com-
parison or statistical treatment. Comparative and genetic
psychology alike demand methods which shall yield precise,
varied, and comparable data of reaction from measurements of
various stages, types, and conditions of organization.

L. W. Cole departed from the well-worn path which Thorn-
dike had earlier broken, in originating the serial stimulus method
of testing for imaginal or ideational behavior. This method,
also, was ill-adapted to statistical needs, and like the earlier
procedures, yielded only roughly comparable data. As thus far
used, it is an indicator of problems rather than a scientifically
exact instrument for solving them or of obtaining detailed de-
scriptions of behavior. It has already served an important end
in breaking up the monotonous succession of problem-box
studies.

Simultaneously with Cole's work on raccoons, which really
revived interest in animal ideation, Hamilton, from a very
different direction, attacked the general problem of reactive
tendencies. As a psychiatrist, he had become deeply interested
in applying the comparative method to the problems of psychiatry
and in bringing the facts of animal psychology and genetic psy-
chology to bear upon the practical problems of mental disease
and defect. His first experimental attempt was a study of reac-
tive tendencies in the dog. Over a period of ten years, he has

METHODS OF EXHIBITING REACTIVE TENDENCIES 13

gradually perfected his method, the while applying it to various
ontogenetic stages in man, cat, dog, and monkey, to defective
and deranged human adults, and to many and diverse types
of animal.

The Hamilton method, which, in the opinion of the writer,
is equal in importance to any method of studying behavior yet
proposed, has been almost wholly neglected by comparative
psychologists and its results are very imperfectly known.

While Cole and Hamilton were busy with their new methods,.
Carr and Hunter 2 were perfecting, in the study of the white rat,
what has appropriately been termed the method of delayed
reaction. It is a simple and ingenious way of testing for idea-
tion. Like Hamilton's, Hunter's contribution to our science is'
important methodologically as well as for its factual materials.
But whereas Hamilton's method of quadruple choices is suited
to reveal reactive tendencies and to exhibit their genetic rela-
tions, Hunter's serves primarily as a test of the ability of an
organism to respond to a situation from which the significant
feature (stimulus) has vanished.

For purposes apparently foreign to the interests of both Hamil-
ton and Hunter, the writer a few years ago devised yet another
method of studying ideational and other reactive tendencies.
It has been called the method of multiple choices. It was
planned as a means of gathering strictly comparable data of
reaction from diverse types of organism, stages of development,
and conditions of normality or abnormality. It was the writer's
hope and conviction that most varied scientific materials should
be assembled systematically in the interest of genetic descrip-
tion. The method is therefore appropriate to human psychology
and to infrahuman, to child psychology and to psychopathology.

To sum up: — for reasons which are obvious to every careful
student of behavioristic method and result, Hamilton's method
of quadruple choices is a preeminently valuable means of dis-
playing reactive tendencies; Hunter's is an uniquely serviceable
test of ability to respond appropriately to controllable absent
stimuli; and the writer's is a promising mode of evoking varied
types of response and of reactive tendency for purposes of classi-
fication an d more detailed analysis.

2 The method is hereafter referred to as Hunter's because he alone has pub-
lished concerning it.

14 ROBERT M. YERKES

The three methods differ so much in value, or rather in their
special kinds of serviceableness, that they may not be directly
compared. All are useful in the study of ideational and other
highly adaptive forms of behavior, but each has certain peculiar
advantages, whatever the ideational problem in question. For
this reason, chiefly, it has seemed to the writer important, as
a matter of economy and efficiency of research, to devise a form
of apparatus which should enable the investigator to use at will
any one of the three methods.

It has not been especially difficult to plan such an apparatus,
for the writer has had opportunity to use, and to see used, each
method, and has had full advantage of the published results
of Hamilton and Hunter, as well as personal contact with them.
It may be convenient to refer to the device now to be described
as the convertible ideational or reactive tendency apparatus.
It is called an ideation apparatus, not because its usefulness is
limited to the study of the function of the idea, but because it
was originally devised as a means of discovering those types of
behavior which are either definitely ideational or closely akii
thereto. Objectivists who are offended by the term ideation
may substitute reactive tendency or some other equivalent term.

The three methods for which this apparatus may be employed
are presented, not as the final word in the study of complex
behavior, but rather as the first words concerning a new ap-
proach to genetic problems.

DESCRIPTION OF APPARATUS

The apparatus consists (1) of twelve identical boxes, each
with an entrance door and an exit door that can be raised or
lowered by the experimenter from his observation stand; (2) a
reaction chamber in which the subject responds, as may be, to
a definite experimental situation, which may be described as
a " setting " of the various mechanisms (this setting differs for
the three methods, and also from trial to trial in the Yerkes'
method) ; (3) a release box in which the subject is confined be-
tween trials and from which it is admitted, at the proper mo-
ment, to the reaction chamber; (4) alleys for the passage of the
subject from the rear of the reaction mechanisms or boxes to
the release box; (5) twelve reward mechanisms, one for each box;
(6) a keyboard, or series of levers, (depending upon the size of

METHODS OF EXHIBITING REACTIVE TENDENCIES 15

the apparatus) connected by means of cords or wires with the
various entrance and exit doors of the apparatus, and so ar-
ranged as to enable the experimenter to unlock and open or to
close and lock any given door by a simple movement of a key
or lever; (7) a protected incandescent lamp in each of the boxes,
with the necessary switch and timing mechanisms for its satis-
factory use in connection with the Hunter method of delayed
reaction (lamps need not be installed in the twelve boxes, but
only in those which are to be used for the delayed reaction
method) .

This apparatus may be built in three sizes: small, medium,
and large.

The small apparatus is suitable for experiments with such
organisms as the toad, frog, lizard, tortoise, mouse, rat, spar-
row, canary, and other like-sized amphibians, reptiles, birds, or
mammals. The medium-sized apparatus is suited for experi-
ments with the tortoise (large), snake, dove, crow, domestic
fowl, cat, small dog, raccoon, rabbit, squirrel, marmoset, and
o her medium-sized reptiles, birds, or mammals. The large
apparatus may be used for various types of large-sized lower
vertebrates, and for such mammals as the cat (large), dog, pig,
goat, sheep, bear, monkey, ape, and man.

The several figures indicate the general plan of the apparatus
and certain of the most important points of construction.

Each reaction box, according to figures 1 and 3, and also
according to the measurements of table 1, occupies five degrees
of arc. The width of the box is therefore determined by its
distance from the center X (figures 1 and 3). By making the
boxes intercept six degrees instead of five, the advantage can be
gained of shorter distances between release door and entrance
door, but there results the serious disadvantage that the appa-
ratus is so spread out as to demand a considerable eye movement
for inspection of the twelve reaction boxes. There is the further
disadvantage, in the wider angle, that the large apparatus re-
quires for its installation a floor area of nearly thirty-six by
thirty-six feet. For these and other reasons, it has seemed
desirable to make use of the five degree angle in the designing
of this convertible apparatus.

The alleys are, in each size of apparatus and throughout their
lengths, the same width inside as the reaction boxes are outside.

16

ROBERT M. YERKES

RB

Figure 1. — Left half of medium sized reactive tendency apparatus. (1) 1-6, reac-
tion mechanisms or boxes; En, entrance door; Ex, exit door; (2) RC, reaction
chamber; (3) rb, release box; rd, door between release box and reaction cham-
ber; (4) A, A, alley from reaction boxes to release box; D, door between alley
A and release box; X, center of circle on arc of which reaction boxes are placed.

METHODS OF EXHIBITING REACTIVE TENDENCIES

17

The plan of the medium sized apparatus appears as figure 1,
and in figure 2 there is shown an enlargement of one of the
reaction boxes, with the arrangement of sliding entrance and exit
doors and the concealed reward mechanism. Figure 3 represents
the three sizes of apparatus in their relations. These must, of
course, be built separately and be independent of one another.

Figure 2. — Ground plan of reaction box. En, entrance door; Ex, exit door; s, s,
wooden guides for sliding door; B, wooden block for food cup; R, food cup.

The small apparatus should be made of quarter inch white
wood (poplar), red wood, or pine, according to locality, and
covered with netting made of No. 20 wire, three meshes to the
inch. The medium sized apparatus should be made of half inch
stock, and the wire netting used as a covering, or for other
necessary purposes in connection with it, should be No. 17 wire,
two meshes to the inch. The large apparatus should be made
of seven-eighths inch stock, and the accompanying wire netting
should be made of No. 12 wire, one mesh to the inch.

18

ROBERT M. YERKES

Figure 3. — General plan for three sizes of reactive tendency apparatus. S, small apparatus; M, medium appa-
ratus; L, large apparatus. X, center of circles on arcs of which reaction boxes and outer alley walls are
placed; A, release box for small apparatus; B, release box for medium apparatus; C, release box for large
apparatus; D, E, F, alleys for small, medium, and large apparatus, respectively; 13, release door (the release
doors for the three sizes of apparatus are shown); 14, door between release box C and alley F.

METHODS OF EXHIBITING REACTIVE TENDENCIES

19

All stock should be planed on both sides, and the apparatus
should be given two or three coats of dark gray paint, if it is
to be exposed to the weather. If, instead, it is to be used in-
doors, it should be painted white or gray, according to the
degree of illumination of the experiment room.

The walls of the reaction chamber should be made of wire
netting of the weight indicated above. The outer walls of the
alleys may be made of wood or wire netting. The release box
should be built of wood except for the wire netting cover and
door. The entrance and exit doors should be made of wood. 8

In table 1 are presented the chief dimensions for the three
sizes of apparatus under consideration.

Table 1

Principal Dimensions in Centimeters or Inches of
Convertible Reactive Tendency Apparatus

Measurements Dimensions Dimensions . Dimensions

for for for

Of reaction boxes Small Medium Large

Width outside 10 cm. 30 cm. 60 cm.

Width inside (minimum) 7.5 cm. 25 cm. 51 cm.

Length outside 30 cm. 60 cm. 140 cm.

Length inside 29- cm. 58- cm. 135- cm.

Depth outside 20 cm. 40 cm. 200 cm.

Depth inside 19+ cm. 38+ cm. 198- cm.

Of entrance and exit doors

Width 8.4 cm. 27 cm. 54 cm.

Length 2.0 cm. 40 cm. 200 cm.

Of release box

Width 33+ cm. 99+ cm. 198+ cm.

Length 30 cm. 60 cm. 140 cm.

Depth 20 cm. 40 cm. 200 cm.

Of release box doors

Width 10 cm. 30 cm. 60 cm.

Length. 20 cm. 40 cm. 200 cm.

3 For details see Behavior Monographs, vol. 3, no. 1, p. 14.

4

20 ROBERT M. YERKES

Measurements Dimensions Dimensions Dimensions

Of alleys Small Medium Large

Width inside 10 cm. 30 cm. 60 cm.

Depth 20 cm. 40 cm. 200 cm.

Distance from center X

to entrance doors 114.5 cm. 343.6 cm. 687.1 cm.

Distance from release door

to entrance doors 105.9 cm. 317.6 cm. 635.1 cm.

Of strips for doors to slide in

Thickness 1/4 in. 1/2 in. 7/8 in.

Width 2 cm. 3 . 5 cm. 6 . 5 cm.

Length 20 cm. 40 cm. 200 cm.

Block for reward mechanism

Width 6 cm. 10 cm. 15 cm.

Length 10 cm. 30 cm. 60 cm.

Depth 2 cm. 4 cm. 6 cm.

Hole in block

Diameter 4 + cm. 6 + cm. 7 + cm.

Food cup

Diameter at top 4 cm. 6 cm. 7 cm.

Depth 2 cm. 4 cm. 6 cm.

Cover for food cup

Width 7 cm. 20 cm. 30 cm.

Length 8(2+6) 14(4 + 10) 25(10 + 15)

Space necessary for apparatus in use

Width 10 ft. 20 ft. 30 ft.

Length 12 ft. 20 ft. 36 ft.

Certain suggestions concerning details of construction are of
practical importance. It is desirable, for the sake of uniformity,
to supply each box with a floor. This floor should be cut shorter
than the sides of the box so that the entrance and exit doors
may drop past it, thus discouraging attempts of subjects to
raise the doors. Or, if the floor is cut full length, a strip nailed
across the box just inside of the exit door will serve the same
purpose while giving support to the floor.

Each box should have a wire netting cover on top.

*

METHODS OF EXHIBITING REACTIVE TENDENCIES

21

All doors should slide vertically, upward, in wooden ways.
These are conveniently made by nailing strips of wood to the
side walls of the box. The strips serve the additional purpose
of supporting the side walls. The outside strip may either be
nailed to the end of the side wall or along the side. If nailed
to the end, it serves as the outside strip for adjacent doors and
thus reduces the amount of labor. In figure 2, the outside strip
for the entrance door is shown as nailed to the end of the side
wall. The writer prefers this method of construction.

The reward receptacle, or mechanism, must be so constructed
as to be concealed when the exit doors are down and fully ex-
posed when they are raised. It may be simply and conveniently
constructed by nailing outside the rear end of each box a block
of wood, of the dimensions suggested in the table, in the center
of which there is a hole large enough to receive a metal food
cup. Aluminum is preferable as material for the food cup, and
desirable dimensions for the various sizes of apparatus are sug-
gested in table 1. In the proper position on the outside of the
exit door, there should be screwed a metal plate, bent at right
angles in such wise as to cover completely and tightly the food
cup when the exit door is down. This is shown in figure 4.

Figure 4. — Metal cover for food cup. w, position of food cup under cover; z, point
at which cover is bent nearly at right angles; x, portion of cover which is at-
tached to exit door by means of wood screws, holes for which are indicated;
y, portion of cover which hides food cup.

The dimensions for this cover or cap for the food cup, also,
are indicated in table 1. For the small apparatus, heavy tin
is a satisfactory material for this cover; for the medium appa-
ratus, light galvanized iron suffices; and for the- large appa-
ratus, it is necessary to use galvanized iron which is so thick
that the large apes cannot readily bend the cover out of shape.
The thickness should be about 1/16 inch.

22 ROBERT M. YERKES

For most animals there is no necessity of locking the doors of
the apparatus, but when it is to be used with monkeys or anth-
ropoid apes, it is absolutely necessary that the experimenter be
able to securely lock any one or all of the sliding doors. It is
therefore essential to equip the large sized apparatus with locks
to be operated in connection with the mechanisms which raise
and lower the doors. Each door should lock automatically when
lowered and unlock when the raising mechanism is operated.

Just behind and a trifle above the release box, an observer's
stand or record table should be constructed, separated by a
screen from the apparatus so that the animal shall not be able
to see the observer. On this table there should be placed a
keyboard, or lever device, by means of which any one of the
twenty-six working doors 4 of the apparatus may be raised or
lowered quickly and quietly.

For the small apparatus the various doors may be controlled
readily by means of a light cord, which runs from a screw eye
in the top of each door, through appropriately placed pulleys,
to a hinged lever key which the observer operates. This key
should be so arranged that when it stands in approximately
vertical position the entrance door is closed. When it is placed
in the horizontal position, the entrance door is open. A cord
from the exit door, carried similarly by pulleys, should be so
placed that it may be attached readily by means of hook and
ring, or ball and slot, to this key, so that if, when a given en-
trance door is lowered, the experimenter desires to raise, simul-
taneously, the exit door of the same box, the pushing of the
key to the vertical position will effect the appropriate move-
ment of each door, that is, will simultaneously lower the given
entrance door and raise the given exit door. The distance to
which the entrance door is raised may be altered by changing
the point of attachment of the cord to the key. This simple
hinged key and cord device renders necessary the use of only
fourteen keys for the operating of twenty-six doors, but the
scheme is feasible only so long as the doors in question are light
enough to be readily moved by means of a fairly small lever
key. The accompanying diagram, figure 5, indicates the rela-
tions of parts, as described above.

4 If both return alleys are used there are twenty-seven doors instead of twenty-
six to operate.

METHODS OF EXHIBITING REACTIVE TENDENCIES

23

Ex

En

Figure 5.— Diagram of lever-key mechanism for raising and lowering doors. En,
entrance door; Ex, exit door; T, observer's table: s, hinged lever key in vertical
position; p, same, in horizontal position; A, pulley for cord between entrance
door and lever key; B. pulley for cord between exit door and lever key; C,
second pulley for cord from exit door.

For the medium sized apparatus also, the lever key mechan-
ism is feasible, but it requires considerably more space and
much greater effort on the part of the experimenter. A sub-
stitute for it is the weighted cord mechanism. 5 A cord with
appropriate carrying pulleys is provided for each door, and to
the end of the cord, which drops in front of the experimenter's
table and within easy reach, is attached an iron or lead weight
which is just sufficient to hold the door in position after it has
been raised by the experimenter. If the weight is too heavy,
the door will tend to rise at inappropriate times; if too light, it
will not stay in position after being raised. This device has the
defect of varying in reliability with humidity and temperature,
since the door will slide more or less easily in accordance with
these varying conditions. The lever mechanism is preferable,

5 Described in previous papers on the multiple- choice method. A study of the
behavior of the pig Sus Scrofa by the multiple-choice method, Journal of Animal
Behavior, 1915, 5, p. 188. The mental life of monkeys and apes: a study of idea-
tional behavior, Behavior Monographs, 1916, 3, p. 14.

24 ROBERT M. YERKES

since it can be relied upon to place and hold the doors in a
constant position.

For the large apparatus, it is extremely desirable to devise
some type of lever mechanism which shall be easily manipu-
lated, reliable, and inexpensive. All of the mechanisms thus far
proposed are either too cumbersome or too expensive to be
feasible, but it is hoped that shortly a method may be discovered
by which the experimenter may conveniently and accurately
control the various doors by means of levers, the maximum
excursion of which shall not exceed eighteen inches. Since the
various doors must be raised a maximum of seventy-two inches,
it will probably be necessary to introduce one or more forms of
multiplying device. Already an automatic locking device, to be
operated in connection with the proposed system of levers, has
been designed.

In the absence of a satisfactory scheme for the use of levers,

weighted cords and locks, which are operated independently,

may be employed. But this system of control mechanism, as

has been stated above, is both unreliable and troublesome to

operate because of the numerousness of the parts. There must

be a separate weighted cord for each of the twenty-six doors

and a separate lock mechanism for each of the twelve boxes,

entrance and exit door in each case being controlled by the

same lock.

USE OF APPARATUS

The use of the convertible reactive tendency apparatus in con-
nection with each of the three methods in question will now be
described. For all of the methods alike, rewards and punishments
may be used as inducements to effort. As rewards, food pre-
sented in the food cups, or for children small presents similarly
presented, serve well. In certain exceptional instances, it may
prove desirable to present the reward for a successful choice, not
in the food cup of the correct box, but instead at the entrance to
the release box. As punishment, it has proved feasible to use
confinement in incorrect boxes. It seems probable that for cer-
tain organisms the electric shock may prove useful.

Hamilton Method

For use with the Hamilton method of quadruple choices, the
following procedure is suggested. This method involves the use
of only four reaction mechanisms. Boxes 5, 6, 7 and 8 may

METHODS OF EXHIBITING REACTIVE TENDENCIES 25

therefore be used, the fact that they are to be reacted to being
indicated by their openness, the entrance doors being raised in
case of each trial. Since the entrance doors of all other boxes
should remain closed and locked, there would be no persistent
tendency on the part of most organisms to attempt to enter
other than the four boxes referred to. For some purposes, it
may prove even more satisfactory to use boxes 2, 5, 8 and 11.

Incorrect choices would not be rewarded, and as seemed desir-
able the subject could be punished for such choices by being
confined in the boxes for a stated period. A correct choice, no
matter what the particular form of the problem, would naturally
be rewarded by the presentation of food in the food cup.

Various problems, in addition to that originally suggested by
Hamilton, may be presented by this method. The following
will suggest the range of possibilities: (1) An insoluble prob-
lem, such as Hamilton used, the several boxes serving as cor-
rect boxes in irregular order, but the same one never twice in
succession and each the same number of times in every hundred
trials (this problem is practically insoluble by even the most
intelligent organism) ; (2) the systematic use, as correct box,
of each in turn from the left end to the right end, that is, 5, 6,
7, 8, or in case of the other group of boxes, 2, 5, 8, 11, this suc-
cession being repeated indefinitely; (3) box at left end, box at
right end, box next to left end, box next to right end, the same
being repeated indefinitely. From these suggestions, it is evi-
dent that various degrees of complexity of order and relation-
ship might be utilized to elicit reactive tendencies and to dis-
play problem solving ability of different sorts.

The apparatus demands no special modification or adaptation
for use in connection with the Hamilton method. Further details
are unnecessary in view of the fact that Hamilton has already
published a fairly complete description of method and apparatus, 6
and has in press a still more elaborate account of procedure
and results. 7

Hunter Method

For the method of delayed reaction the apparatus demands
certain special appliances which, however, do not have to be
removed when either the Hamilton or the Yerkes method is

6 Hamilton, G. V. A studv of trial and error reactions in mammals. Journal
of Animal Behavior, 1911, 1, pp. 33-66.

7 Behavior Monographs, 1917, 3, no. 13.

26 ROBERT M. YERKES

in use. The special equipment consists of a concealed incan-
descent electric lamp for the illumination of each box and an
electric signal and timing mechanism for the operation of the lamps
and the door between the release box and the reaction chamber.

The method of delayed reaction may be used with various
groups of doors, according to the grade of difficultness of re-
sponse desired. Thus, as the simplest situation, boxes 6 and
8 may be used. In this case, the entrance doors of both boxes
should be raised in preparation for a trial. The doors of the
other boxes should remain closed. In accordance with a pre-
arranged plan, either the one or the other box would be indi-
cated, by momentary illumination, as the box to be chosen.

For the second grade of difficultness, boxes 5, 6, 7 and 8 might
be used, each of them having the necessary equipment and con-
nections for use as the correct box; for grade three, boxes 2, 5,
8 and 11 ; for grade four, boxes 1, 3, 5, 7, 9 and 11 ; and for grade
five, all of the twelve boxes might be subject to use, that is, the
entrance door of every box should be open and the subject should
be required to choose that one of the twelve which has previously
been illuminated.

The satisfactory use of this method necessitates not only the
presence of a lamp, but the installation of a mechanism which
shall control several important factors in the situation. The
experimenter, by pressing a simple key, should close a circuit
which at once illuminates a certain box (the particular box to
be determined by the setting of a switch), and at the same time
starts a timing mechanism. This mechanism should, after an
interval, with a range of 1 to 10 seconds, open the lighting cir-
cuit, thus cutting off the illumination of the correct box; and
after an interval of to 60 seconds it should cause the door
of the release box to open so that the animal may enter the
reaction chamber. For intervals longer than 60 seconds, it seems
best to have the experimenter determine the delay by means of
a stop watch and operate the door of the release box by hand.

There is no obvious reason why this twelve mechanism reac-
tive tendency apparatus should not be used in wholly satisfac-
tory fashion for the study of delayed reactions. The additional
electrical equipment should in no wise interfere with the other
uses of the apparatus and that portion of it which controls the
release box door might be made to serve the experimenter in
connection with all of the methods.

METHODS OF EXHIBITING REACTIVE TENDENCIES 27

Yerkes Method

For use by the method of multiple choices, the apparatus
demands neither modification nor special adaptation. The chief
features of the method have already been described several
times, and it is needless here to do more than formulate a set
of problems with wider range of difncultness than those hereto-
fore used in reported experiments on lower animals. Those
proposed problems, ten in number, are presented in brief form
below, with a series of ten settings for each. Thus, in case of
problem 1, for which the correct mechanism is always box
number 5, that is the fifth from the left end of the apparatus,
the first setting involves the use of boxes 1 to 6, the second
setting, of boxes 3 to 12, and so on. It is understood* that, if
possible, this series of ten settings (ten trials) shall be presented
to a subject once a day until the problem has been solved. If
for any reason the series of ten trials cannot be completed on
a given day, it should be resumed from the point of interrup-
tion on the following day. If more than one series per day can
be given, either the ten trials may be divided into two groups
of five each or the total series may be repeated.

In each of the series of ten settings, a total of sixty boxes is
presented. The average number of boxes open in each trial
is, therefore, six. Of these sixty boxes, ten are definable as
correct boxes. The probability of correct first choice prior to
experience is for any series of ten trials, one to five. In order
that this ratio of probable right to wrong first choices shall not
be disturbed, it is desirable that the experimenter make use of
the proposed settings.

Proposed Problems and Settings for Multiple-Choice

Method

Problem 1. Same box (box 5).

1-6 (5) ; 3-12 (5) ; 4-6 (5) ; 5-9 (5) ; 2-10 (5) ;

4-5 (5) ; 4-10 (5) ; 3-6 (5) ; 1-8 (5) ; 5-10 (5).
Problem 2. First at left end.

6-12 (6); 11-12 (11); 3-11 (3); 1-5 (1) ; 4-11 (4);

10-12 (10); 5-9 (5); 2-12 (2); 8-11 (8); 7-12 (7).
Problem 3. Middle.

1-7 (4); 10-12 (11); 6-10 (8); 1-11 (6) ; 1-3 (2);

4-10 (7); 1-9 (5); 9-11 (10); 1-5 (3); 6-12 (9).

28 ROBERT M. YERKES

Problem 4. Third from right end.

1-6 (4); 5-8 (6); 3-12 (10); 1-3 (1); 7-11 (9);
2-10 (8); 1-7 (5); 3-5 (2); 2-9 (7;) 1-5 (3).

Problem 5. Alternately left end and right end.
8-12 (8); 1-10 (10); 3-8 (3); 6-9 (9); 1-9 (1);
3-5 (5); 7-11 (7); 5-12 (12); 2-8 (2); 4-6 (6).

Problem 6. Progressively from right to left end of apparatus —

toward left bv ones.
10-12 (12); 6-12 (11); 3-10 "(10); 8-12 (9); 8-10 (8);
1-9 (7); 5-8 (6); 4-9 (5); 2-11 (4); 3-7 (3).

Problem 7. One place to left of middle key.
6-12 (8); 3-5 (3); 8-12 (9); 1-9 (4); 2-12 (6);
10-12 (10); 5-11 (7); 1-5 (2); 3-9 (5); 1-3 (1).

Problem 8. Alternately second from right and second from left.
6-12 (11); 2-5 (3); 1-8 (7) ; 5-9 (6); 1-5 (4);
4-12 (5); 5-10 (9); 9-11 (10); 2-9 (8); 1-5 (2).

Problem 9. To the right of mid-point in even group; or first

member of second-half of group.

3-10 (7); 1-4 (3); 2-7 (5) 1-2 (2); 3-12 (11);
8-11 (10); 5-12 (9); 1-10 (6); 5-10 (8) 11-12 (12).

Problem 10. Alternately to left of middle key and to right of it.
1-7 (3); 8-12 (11); 2-10 (5); 10-12 (12); 1-9 (4);
3-9 (7); 1-3 (1); 6-10 (9); 6-12 (8); 3-7 (6).

The various forms of problem serviceable in' connection with
the different methods and the detailed procedure for each remain
to be worked out. The methods have been thoroughly tried
out and have already yielded such valuable results that further
development and application is obviously desirable. There is
no reason why the same apparatus should not henceforth serve
for studies of reactive tendencies and ideational behavior by
the method of quadruple choices, that of delayed reaction, that
of multiple choices, and that of conditioned reflexes.

We experimenters shall . doubtless do well to use our devices
to the limit of their applicability, seeking no less assiduously
new ways of employing existing experimental equipment than
we seek to invent new mechanisms.

LIGHT REACTIONS OF THE CRIMSON-SPOTTED
NEWT, DIEMYCTYLUS VIRIDESCENS

A. M. REESE

West Virginia University

INTRODUCTION

The following experiments, which are extensive rather than
intensive in character, were started with a dozen salamanders
obtained in the month of November from the Marine Biological
Laboratory at Woods Hole. During the course of the experi-
ments, which extended over a period of more than a year, three
of the animals escaped, so that some of the later results were
obtained with only nine animals; they were obtained from
Woods Hole because of their comparative rarity in the neigh-
borhood of Morgantown when the work was begun. Later,
animals were caught in a local pond and these were also used
in the experiments.

No change in reaction, except in one possible case, was pro-
duced by prolonged residence (for a month or more) in a photo-
graphic dark room, though it was noted that all of the animals
were of a lighter shade of color when first brought from the
dark room. '

In all but one or two cases the animals were confined in a
rectangular glass aquarium, six inches wide by ten inches long,
with two or three inches of water. The water was used chiefly
for two reasons: because the newts were very much more active
in the water than they were in the empty aquarium, and because
the water, of course, acted as a heat-screen and practically elim-
inated heat as a stimulus.

A few tests were made without water, with no noticeable
difference in reaction except speed; the animals responded two
or three times as quickly when in water than they did when in
the merely moistened aquarium.

Observations made upon animals in an evenly illuminated
aquarium seemed to show that they have a certain tendency

30 A. M. REESE

to collect in groups, in one place or another, without regard to
the light stimuli to which they are subjected; this tendency,
then, has no apparent bearing upon the following experiments.

After hundreds of observations, extending over a period of
many months, upon several lots of animals, several sets of
observations were made upon one or two small groups of animals
immediately upon bringing them into the laboratory from their
native pond. Under these conditions the animals responded
either very indefinitely to the same light stimuli, or even in a
contrary manner to the animals that had been for some time
under observation. This irregularity in what had been con-
sidered the normal response was also noticed in a group of
animals that had been in the aquarium for a long time and had
not been used in the experiment for a considerable period.

It is possible that, after all, the responses of the animals
under these abnormal conditions may be quite different from
what would be seen under normal conditions in their native
habitat.

It is the intention of the author to carry on similar experi-
ments upon this species in the natural environment as soon as
a suitable spot can be found. (See Addendum.)

Experiment I. — This experiment was to determine whether
Diemyctylus is positively or negatively phototropic towards
white light.

Twelve animals were placed in the above-described aquarium
of water which was entirely surrounded by black except over
half of the top. Ten inches above the surface a 25-watt, 115-
volt tungsten lamp was so fixed as to illuminate exactly one-
half of the aquarium, the other half, of course, being thrown
in dense shadow.

At regular intervals of five minutes the numbers of animals
in both light and dark ends were noted. When an animal, at
the moment of observation, happened to be partly in light and
partly in shadow it was counted for that region in which the
greater part of its length lay, though occasionally an animal
was so near the exact center that it was not counted on either
side.

Table I shows that in 30 observations 95 animals were found
in the light and 250 in the dark. These observations were

LIGHT REACTIONS OF NEWT 31

taken on three different days; and after observations 5, 7 and 17
the light and dark ends were suddenly interchanged, thereby
throwing the larger proportion of the animals, that had collected
in the shadow, into the light. The last five observations were
made about two weeks after the first, during which time three
of the animals had escaped.

TABLE I

Observation. . . 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Light half 232220012042252

Dark half 10 9 10 10 10 12 12 11 10 12 8 10 10 7 10

Observation. . . 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Totals

Light half 126454354148955 95

Dark half 11 10 6 8 7 8 9 7 8 11 5 1 4 4 250

Another set of observations made under the same conditions,
except that only enough water to moisten the bottom of the
aquarium was used, gave 141 animals in the light to 243 in the
dark region. As noted above, without enough water to swim
in the newts are so sluggish that experimentation is not nearly
so satisfactory as when they are actively swimming.

It is evident, then, that, at least under the conditions of the
experiment, these newts are negatively phototropic.

Experiment II. — Variations of experiment I were tried to de-
termine the effect of temperature upon the phototropic reac-
tions of Diemyctylus.

The first variation was merely to start the observations, made
upon eleven animals, with the water at 10° C, the arrangement
of aquarium and lights being as in experiment I. Not only
was the aquarium surrounded by black, but the experiment was
performed in a photographic dark room. Beween the 15th
and 16th observations was an interval of two hours, during
which time the animals were in the dark. After the 30th obser-
vation, when all the animals were in the dark half of the aqua-
rium, the ends were reversed, throwing all the animals into the
light half. When the animals, as in this case, were sluggish it
would be some time before they would move into the dark
again, which would reduce the total preponderance of dark over
light. The total figures for 40 observations were 174 in the
light end to 278 in the dark, which was about the proportion
noted in experiment I when no water was used. At the end of

32

A. M. REESE

this part of this experiment the temperature of the water had
risen about 2° C.

In the second variation of this experiment the aquarium con-
taining the animals was placed out-of-doors for about five hours,
until the temperature of the water had fallen to 1° C ; it was
then brought into the dark room where the same arrangement
for vertical illumination of just half of the aquarium as in ex-
periment I was used. All of the animals at this temperature
were numb with cold, and lay motionless on the bottom of the
aquarium. One or two were apparently dead and when turned
over, ventral side up, made no effort to right themselves. At the
beginning of this series of observations six animals were placed
in the dark end of the aquarium and five in the light end.

TABLE Hi

Observations 123456789 10

Light half 5555544444

sssssSSsss
Dark half 6 6 6 6 6 7 7 7 7 7

sssssSSSSs

Observations . . 15 16 17 18 19 20 21 22 23 24

Light half 6667221000

S s s S S- S- A

Dark half 5 5 5 4 9 9 10 11 11 11

S S S S S- S- S A- A- A

Observations . . 29 30 31 32 33 34 35 35 37 38

Light half 3005 5 4 2252

A . . . . S- a A- A- S a A

Dark half 8 11 11 6 6 7 9 9 6 9

S S S S- S- S A- S a A

Observations . . 43 44 45 46 47 48 49 53 51 52

Light half 0644533157

.. A A- A A S S- A- a a

Dark half 11 577688 10 64

.. S A- s S A- S- A- A A

Observations . . 57 58 59 60 61 62 63 64 Totals

Light half 5 6 2 9 5 9-5 2 259

a A A A A a a a

Dark half 65926269 446

A A A A A a a A

11 12 13 14

4 4 4 4

s s s s

7 7 7 7

s S S S-

25 26 27 28

10 7 4 4

S- A A A-

1

4 7 7

s- s s

39 40 41 42

3 111

a A A A

8 10 10 10
A- A- A- A

53 54 55 56

6 6 8 8

a a a a

5 5 3 3

a A a A

Observations were begun at intervals of three minutes, but

1 In this and the following experiments the letters refer to the average activity
of the animals at the time of observation —

a = very active;. A = less active; A — = still less active,
s = very quiet; S = less quiet; S — = still less quiet.
A — and S — would probably be about the same state of ac tivity.

LIGHT REACTIONS OF NEWT 33

as no change of position had taken place at the end of fifteen
minutes the interval between observations was changed to five
minutes, from observations 7 to 21, after which it was again
made three minutes.

It will be seen from table II that, for the first 18 observa-
tions, lasting about one and one-fourth hours, there was very
little change in the position of the animals, which lay almost
motionless during that time. At the 18th observation the tem-
perature of the water had risen to only 7° C, and warm water
was carefully added until that in the aquarium was raised to
13.5° C. ; the animals soon began to become more active, and
after twenty minutes (22nd observation) all were collected in
the dark half of the aquarium. From the 22hd observation
until the end of the experiment observations were made at
intervals of three minutes. It will be seen by table II that
the light was changed after the 24th observation, throwing all
the animals into the light end; after fifteen minutes all the
animals had again collected in the dark region.

After the 31st observation, when the water of the aquarium
had risen to 15° C, warm water was again added until that in
the aquarium was raised to 24° C. ; this operation was repeated
after the 36th observation and the temperature raised to 33° C.
The animals were mostly very active but continued to collect
in the dark region, so that after the 43rd observation, when all
were in the dark, the ends were reversed, throwing all the
animals in the light end.

After the 50th observation, when ten of the eleven animals
were in the dark region, enough water was added to raise the
temperature to 36.5° C. ; this caused the animals to become
unusually active, to frequently give a squeaking sound, and to
come to the surface for air. After this, it will be noticed from
the table, there is no longer a tendency to collect in the dark,
possibly a slight tendency in the reverse direction. After the
59th observation water was again added until that in the aqua-
rium was raised to to 38° C. At this temperature the animals
acted as just described, but with more vigor. Some of them
were so seriously affected that they turned ventral side up and
could scarcely right themselves again, and it was evidently im-
possible to further increase the temperature without endanger-
ing the lives of the animals.

34 A. M. REESE

It is apparent, therefore, that low temperatures, not far above
the freezing point of water, cause these animals to become so
sluggish as to be more or less indifferent to differences of light
and darkness. As the temperature rises they become active and
seek the dark region of the aquarium. When the temperature
reaches about 36° C. they become abnormally active and again
become indifferent to light and shade differences. At somewhat
less than 40° C, about the temperature of human blood, (though
they could doubtless be acclimated to higher temperatures') they
are seriously affected or possibly killed.

Experiment III. — Another variation of experiment I was to
determine whether the animals would seek the dark half of the
aquarium when the illumination was from below.

The same aquarium and eleven animals were used as in the
preceding experiments, but the light was thrown from below
by the same tungsten lamp, placed six inches below the bottom
of the aquarium. In all, 60 observations, at three-minute in-
tervals, were made, with a rest of three and one-half hours be-
tween the 30th and 31st observations. The temperature of the
water was about 27.5° C. and the animals were active throughout
the experiment, those in the light being the more active, on the
average. The total number of animals counted in the light was
266; those in the dark, 360.

It is evident then, that Diemyctylus tends to come to rest
in the dark region of the aquarium when the light comes from
below, but that the tendency is not so strong as when the source
of light is above the water.

Experiment IV. — This experiment was to determine the re-
action of Diemyctylus in relation to the direction of white light.

In this and similar experiments both the region of the aqua-
rium where found and the position of the animal in relation
to the direction of the light were noted. It was noticed that
when the aquarium, described on page 29, was placed with one
end about eight feet from a window, but not in the direct sun-
light, on a fairly bright day, a large proportion of the animals
stayed in the end of the aquarium towards the light and swam
against the glass as though trying to get nearer the window.
No actual counts were made in this observation.

LIGHT REACTIONS OF NEWT 35

TABLE III

Observations 123456789 10

Facing light 8867767776

A- A- A- S- A A A- A A- A-
Facingdark 3354.. 54442

S S S S.. A-S-S-S S
In light end 7 6 4 5 7 5 6 8 4 4

A A- A- A- A A- A- A- A A
In dark end 4 5 7 6 4 6 5 3 7 7

S S- S S S A- S- S- S- S

Observations 11 12 13 14 15 16 17 18 19 20 Total

Facing light 5469868696 136

a A A A- A- a A A A- a
Facing dark 4412343425 68

S- S S S S S- S- S- S- S-
In light end 5467778676 119

aAAA-AAAA a a
In dark end 6 7 5 4 4 4 3 5 4 5 101

S-SSSSS-SSSS-

Table III shows the results of a series of observations upon
the same eleven animals used in the preceding experiments, The
aquarium, containing a few inches of water, was entirely sur-
rounded by black except at the end which was towards the win-
dow, in this case twenty feet away. The day, while not dark,
was overcast, and the light that entered the open end of the
aquarium was naturally quite dim. When an animal, at the
instant of observation, lay at right angles to the direction of
the light it was not counted. It will be seen that exactly twice
as many animals faced towards the light as faced away from it,
while the number of animals in the half of the aqaurium near
the window was not very much greater than the number in the
other half. It will be noticed also that, as a rule, the animals
facing the light were more active than those facing in the other
direction, and that those in the half nearer the light were more
active than the others.

This experiment shows that these salamanders are positively
phototactic even towards weak daylight.,

Experiment V. — This experiment or series of experiments was
to determine the reaction of the animals towards a much more
intense white light than the daylight of the preceding experi-
ment. The light here used was the same 25-watt, 115-volt
tungsten lamp that was used in experiment I; it was placed
six inches from the open end of the aquarium. The aquarium

36

A. M. REESE

was surrounded except at one end by a black cloth, and the
whole apparatus was operated in a photographic dark room.
Observations were made at intervals of five minutes. The tem-
perature varied from 16.5° C. to 19° C. Eleven animals were
used. At the beginning of the experiment the animals . were
quiet and equally distributed through the aquarium.

TABLE IV

Observations 1 2 3 4 5 6 7 8 9 10 11 12 13

Facing light 7 10 9 7 10 10 8 10 9 7 8 8 4

S-A-A-aAA a A A A A A A-

Facingdark 1124113124337

S S .. S- A- S S- S S- S- S- S A-

In light end 787788 10 876954

S-A-A aA aA a a a a a A-

In dark end 4344331345267

S S- A- S- A- S- S S S- S- S S A-

Observations 14 15 16 17 18 19 20 21 22 23 24 25 26

Facing light 6 9 10 7 10 8 8 9 10 8 11 9 9

A- A A A A A A a A A A- A- A-

Facingdark 32.. 2132213022

A- A- .. S- S S S S .. S- .. S S

In light end 5 7 6 7 6 8 8 9 6 4 7 9 8

A-AA aAAA a a aAA a

In dark end 6454533257423

A- A A- A- S- S A S- S- S- S- S- S

Observations 27 28 29 30 31 32 33 34 35 36 37 Totals

Facing light 88878987554 298

A A A A- S S- A- A- A A A-

Facingdark 13243.. 34662 90

S S S S S .. S S S-A-A-

In light end 97974444534 244

a a A A- A A A- A A- A- A

In dark end 24247777687 163

SSSSSSSSS-AA

For explanation of letters see page 32.

Observations 1 to 12 were made at night; the other observa-
tions during the morning and afternoon of the following day.
Between observations 22 and 23 was an interval of two hours
and five minutes during which the tungsten light was shining
into the end of the aquarium. After observations 9 and 25 all
the animals were gently pushed into the end of the aquarium
away from the light. After observation 12 and about one hour
before observation 13 the animals were fed as much raw meat
as they would eat. It will be seen from table IV that the aver-
age activity of the animals facing the light was greater than

LIGHT REACTIONS OF NEWT 37

that of the animals facing away from the light; and that the
animals in the near half of the aquarium were, as a rule, more
active than those in the half farther from the light. As before,
animals which lay, at the moment of observation, with the long
axis at right angles to the direction of the rays of light were
not counted. The total number of animals facing the light was
298, to 90 that faced away from the light; the number in the
near half of the aquarium was 244, to 163 in the half farther
from, the light.

Experiment VI. — Another series of 25 observations, taken
every five minutes, under conditions similar to those just de-
scribed, except that the aquarium was in an ordinary room and
covered with the same black cloth, gave 200 facing the light
to 78 facing away from the light, and 179 in the near end to
97 in the far end of the aquarium.

Experiment VII. — Still another series of 30 observations,
taken every five minutes, was made upon nine of the same animals
after having been in the dark for 32 days except for about two
and one-half hours three days before the present experiment.
This was to determine if prolonged residence in total darkness
had any effect upon their reaction to white light. The arrange-
ment of the apparatus was the same as in experiment V. One
hundred and ninety-seven animals were found facing the light,
to 72 facing away from the light; 202 were in the near half,
to 74 in the far half of the aquarium.

It will be seen by comparison with experiment V that, after
this long residence in darkness, the preponderance of animals
that faced the light over those that faced in the opposite direc-
tion was less than in animals that had been in the light; while
the preponderance of animals in the near half of the aquarium
over those found in the distant half was greater in animals
that had been in the dark than in those that had been in the
light. It is possible that these differences may have been due
to other causes than the prolonged residence in the dark.

Experiment VIII. — To see whether the same eleven animals
were positively phototactic to a light of even greater intensity
than the tungsten the aquarium, covered as before, with a black

38

A. M. REESE

cloth, was placed, in a dark room, with its open end fifteen inches
from the lens of an arc projection lantern. Observations were
taken at five-minute intervals. At this distance the light was,
of course, decidedly painful to the human eye.

The positive response was so marked that only 15 observa-
tions were made, which gave 116 facing towards the light to 41
facing away; and 105 animals in the near half of the aquarium
to 60 in the distant half. The animals facing the light and in
the near half were, as a rule, somewhat, though not a great deal,
more active than the others.

It appears, therefore, that the response to white light is about
the same whether the source of light be dim daylight or an
intense electric arc.

Experiment IX. — This experiment was to determine the effect
of low temperature upon the responses of Diemyctylus to white
light at the end of the aquarium.

TABLE V

Observations.. 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Facing light 1 7 2 2 3 5 2 3

S S- S- A- A- S- S- S-

Facingdark 10 11 11 4 9 9 8 6 9 8 8

S- S- S- A- S- S- S- S- A-

In near end 1 00 76 5 6 74 3

s S S S S- S- A- S- S- S- S-

In far end 10 11 11 4 5 6 5 4 7 8 11

s S S S- S- S S- S A- S- S- S- S- A-

Observations . . 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Facing light. .. 02 4 46243433254

S A- A- A- A- A- A- A- A- A-

Facing dark. ..443347774787 4..
A- S A- A- A- A- A- A- A- A-

In near end... 0000 10 774422222

S A- A- A- A- A- A- A- A- A-

In far end 11 11 11 11 14 4 7 7 9 9 9 9 9

A- S S A- A- A- A- A- A- A- A- A-

Observations 29 30 31 32 33 34 35 36 37 38 39 40 Totals

Facing light 4 4 3 5 7 2 4 4 6 3 3 6 122

A- A- A- A- A- A- A- A- A

Facing dark 556649765853 231

A- A- A- A- A- A- A- A- A

In near end 338567555543 140

A- A- A- A- A- A- A A- A

In far end 88265 4 666678 266

A- A- A A- A- A- A- A- A

LIGHT REACTIONS OF NEWT 39

The same animals and arrangement of apparatus were used
as in experiment V, the only difference in the experiments being
that in the present one, table V, the aquarium containing the
animals had been placed outdoors for three hours, until the
temperature of the water had fallen to 0° C. and a thin skim
of ice had formed.

As in experiment II the animals at the beginning of the obser-
vations were all stationary, as though dead, and were evenly
scattered over the bottom of the aquarium. Observations 1 to
6 were taken at ten-minute intervals ; 7 to 40 were at five-minute
intervals. After observations 5, 17 and 30 the ends were re-
versed, thus putting the animals from the far to the near end
of the aquarium; this, of course, raised the total number for
the far end and lowered the total number for the near end. For
the first 3 observations, or one-half hour, practically no change
in the position of the animals took place; at this point warm
water was carefully added until the temperature of that in the
aquarium was raised to 8.5° C, and by the end of the experi-
ment the temperature had slowly raised until it was 12.5° C.
After the addition of the warm water the animals began to
show signs of life, though they remained rather sluggish to the
end of the experiment.

In 40 observations 122 animals faced towards the light to
231 away from the light; 140 were in the near half of the aqua-
rium, 266 in the half farther from the light.

Experiment X. — This was a continuation of the preceding ex-
periment under exactly the same conditions except that the
temperature of the water at the first observation was 4.5. C.
instead of 0° C, and the animals were moving about very slowly
instead of lying perfectly still. After the 4th, 5th and 14th
observations all the animals were pushed into the near half
of the aquarium. Observations were somewhat irregular, being
every ten minutes for about the first half of the observation, every
five minutes for the latter half of the observations. After ob-
servation 16 warm water was added until the aquarium tem-
perature was 23.5° C. ; the dark room being cold, this tempera-
ture was lowered 2° by the end of the experiment. In 28
observations 134 animals were found facing the light to 147
that faced away from the light; while 107 animals were counted

40 A. M. REESE

in the near half of the aquarium to 212 in the far half. It is
noticeable, however, that in the 16 observations before the
warm water was added 67 animals faced the light to 83 that
faced in the opposite direction, while in the 13 observations
after the addition of the warm water 67 animals again faced the
light but only 64 faced away from it. Again, in the first 16
observations 50 animals were counted in the near half of the
aquarium to 126 in the far half, while in the last 13 observations
57 animals were in the near half to only 86 in the far half.

Experiment XI. — This was a continuation of experiment X
on the following day. The water at starting was 5.5° C. and
was raised, after observation 12 to 23° by the addition of
warm water. The first 7 observations were at somewhat irreg-
ular intervals of ten minutes; the remaining observations were
at five-minute intervals. In 30 observations 149 animals faced
the light to 154 that faced in the opposite direction; while 143
were noted in the near end to 198 in the distant end. In the
first 12 observations, however, when the maximum temperature
of the water was 11°, only 43 animals faced the light to 73
that faced away from it; while in the last 18 observations, when
the water had been raised to 23°, 106 animals faced the light
to 62 that faced the other way. Again, in the first 12 obser-
vations 48 animals were in the near half to 94 in the far half,
while in the last 18 observations the numbers were 95 to 104
respectively. In a total of 98 observations for the last three
experiments, 405 animals faced the light to 532 that faced in
the opposite direction; and 390 animals were counted in the
near half of the aquarium to 676 that were found in the far half.

From the last three experiments it seems that low tempera-
tures tend to inhibit or even reverse the positive phototaxis
of Diemyctylus as seen in movements towards the light and
orientation of the body so that the animal faces the light.

Experiment XII. — This experiment was to determine the
responses of Diemyctylus to white lights of different intensities
acting simultaneously at opposite ends of the aquarium.

Nine of the same animals used in the preceding experiments
were employed here; they had been in darkness for 15 days.
The same aquarium in the same dark room was used; it was

LIGHT REACTIONS OF NEWT 41

entirely covered with black cloth except at the ends where the
light entered. Two 25-watt, 155-volt tungsten lights were used;
they were not tested as to candle-power, but they were of the
same supposed power and were of the same age. One light
was six inches from one end of the aquarium, the other light
was twenty-four inches from the other end. The first 50 obser-
vations were taken at five-minute intervals, except that one
and one-half hours intervened between observations 31 and 32,
during which time the animals were in the darkness.

In 50 observations 265 animals were seen facing the more
distant (24 inches) light to 170 that faced the nearer and, there-
fore, more powerful light. Two hundred and sixty-nine animals
were found in the half of the aquarium nearer the more distant
light, 174 in the region towards the nearer light. The weaker
of these two lights, then, seems to have the greater attraction
for the animals.

Experiment XIII. — The arrangements were exactly as in ex-
periment XII except that the lights were six inches and twelve
inches from their respective ends of the aquarium. Two and
one-half hours in darkness intervened between observations 19
and 20. In 40 observations 141 animals were found facing the
nearer (6 inches) light to 185 that faced the more distant light;
while 163 were found in the half of the aquarium towards the
nearer light, to 190 in the other half. The weaker of the two
lights seems again to be the more attractive to the animals,
though in a less marked degree than in experiment XII.

Experiment XIV. — The same experimental conditions as in
the preceding except that the lights were twelve inches and
forty-eight inches from their respective ends. Between obser-
vations 15 and 16 was an interval of three days, and between
observations 45 and 46 was an interval of one day; during both
intervals the animals were in the dark. As in the preceding ex-
periment, the observations were taken every five minutes.

In 60 observations 289 animals were found facing the nearer
(12 inches) light, to 229 that faced the farther (48 inches) light.
Two hundred and eighty-three were seen in the half of the
aquarium towards the nearer light, to 255 in the other half.
It seems that, while the differences between these sets of figures

42

A. M. REESE

are not great, the nearer (12 inches) light has a somewhat greater
attraction than the more distant (48 inches) light.

Experiment XV. — The conditions of this experiment were ex-
actly the same as in the preceding except that the lights were
twenty-four inches and seventy-two inches from their respec-
tive ends. There was an interval of twenty-one hours (in the
dark) between observations 5 and 6. In 40 observations 203
animals faced the nearer (24 inches) light, to 133 that faced the
farther (72 inches) light ; and 204 were in the half of the aquarium
towards the nearer light, to 144 in the other half.

Experiments XII to XV may thus be placed in tabular form
for comparison:

f 6" distance.

Experiment XII..

24" distance.

{ 6" distance.

[facing 170
[near 174
ffacing 265
(near 269
ffacing 141

Experiment XIII •

12" distance.

Experiment XIV.

12" distance.

■\

48" distance.

'24" distance.

Experiment XV ■

72" distance .

near 163
(facing 185

near 190
ffacing 289

near 283
ffacing 229

near 255
ffacing 203

near 204
ffacing 133

near 144

Experiments XII and XIII seem to indicate that when -one of
two sources of light is very intense the animals tend towards the
less intense light; while experiments XIV and XV show that
when neither source is very intense, perhaps not reaching a
certain optimum, the animals tend towards the more intense
light.

LIGHT REACTIONS OF NEWT 43

REACTIONS TO RED LIGHT

Experiment XVI. — In this experiment the same nine animals
and the same arrangement of apparatus as in experiment V were
employed; but between the tungsten lamp, placed six inches
from the end of the aquarium, and the aquarium was a filter
composed of two glass jars each 20 mm. thick containing an
aqueous solution of crystal violet and of potassium monochro-
mate respectively, after the formula of Landholt.

In 30 observations, taken at five-minute intervals, 225 ani-
mals were noted facing the red light to 46 facing away from the
light; and 221 animals were found in the end of the aquarium
nearer the light to 49 in the farther end.

Experiment XVII. — This experiment was an exact repetition
of experiment XVI, made thirty days later, during which period
the animals had been in the darkness of the photographic dark
room. In 12 observations, at five-minute intervals, 86 animals
faced the red light to 19 that faced in the opposite direction;
and 82 were seen in the red end of the aquarium to 26 in the
other end. Comparison of experiments XVI and XVII with
experiment V shows that the proportion of animals attracted
by the light was greater with the red light than with the white.
This may have been due to the decrease in intensity rather than
to the red color.

REACTIONS TO BLUE LIGHT

Experiment XVIII. — This experiment was performed thirty-
six hours after experiment XVII ; during the interval the animals
were in complete darkness. The experiment differed from the
other only in the substitution of a blue filter for the red. This
filter consisted of two similar glass jars containing solutions of
crystal violet and copper sulphate after Landholt 's formulae.
The five-minute intervals between observations were somewhat
lengthened on four occasions by interruptions to the experiment.

In 30 observations 197 animals were found facing the blue
light to 73 that faced away from it; 195 animals were found in
the half of the aquarium near the light to 84 in the other half.
It is evident that the proportion of animals attracted by the
blue light is less than was attracted by the red light.

44 . A. M. REESE

REACTIONS TO GREEN LIGHT

Experiment XIX. — The arrangement of this experiment dif-
fered from the last only in the substitution of a green filter for
the blue. This filter consisted of solutions of copper chloride
and potassium monochromate, after the formulae of Landholt,
in jars like those described in the two preceding experiments.
The same nine animals were used; they had been in total dark-
ness for twenty-nine days, and had been fed upon earthworms
the day before the experiment. In 30 observations, at five-
minute intervals, 210 animals faced the green light to 51 that
faced away from it; and 199 animals were found in the near
half of the aquarium to 71 in the half farther from the light.

The attraction of the green light is apparently more marked
than the blue but less marked than the red.

REACTIONS TO WHITE LIGHT ON VARIOUS PARTS OF THE BODY

Experiment XX. — In order to be able to throw a small, sharply-
defined spot of white light on any part of an animal a small
electric bulb was mounted in the tube of a microscope, as de-
scribed by Bradley M. Patten in Science, January 22, 1915,
pp. 141-2. By using different low-power objectives a sharply
defined spot from 1 to 5 mm. in diameter was directed upon all
parts of the body of several animals. These animals were in a
black rubber developing tray in sufficient water to cover them.
In one case they had been in a dark room only an hour; in
another series of trials they had been in the dark for a week or
more. Some of the animals were of the lighter shade with very
bright crimson spots; other animals were of the darker type
when experimented upon. During experimentation just enough
light was admitted to the dark room to faintly see the animals,
so that any movement could be noted. The spot of light was
thrown, as has been said, on all parts of the body, from the head
to the tip of the tail ; on the crimson spots and between them ; it
was varied in diameter from 1 to 5 mm. No certain reactions
could be determined for any of the animals used. Doubtful
reactions were sometimes obtained when the spot was made
large enough to cover the entire anterior half of the head.

When the spot was thrown on the black bottom of the tray
near the animals they followed it actively and snapped at it,
evidently taking it for food; they seemed to be able to see the
spot at a maximum distance of about 3 cm.

LIGHT REACTIONS OF NEWT 45

Experiment XXI. — The same animals that failed to respond
to the white spot from the microprojection apparatus responded
promptly when a beam of sunlight was thrown, by a small
mirror, upon various parts of the body. When the light was
thrown upon the tail they either started forwards suddenly or
drew the tail sharply forward along the side of the body. When
the light was thrown upon the head the animal usually backed
away from it. Animals in a cloth covered aquarium in a brightly
lighted room responded about as promptly as those in the dark
room.

Animals that had been for some time in the dark responded
more promptly than those that had been exposed to the light;
some of the former fairly jumped when the beam fell upon them.

Little or no response was obtained when a small beam from
a 5 mm. mirror was used instead of a beam that was large enough
to illuminate a large area of the animal at one time.

The animals responded in the same way, and almost as prompt-
ly, to a beam of light from below.

These reactions to a beam of sunlight are quite similar to
those described by the author for Necturus (2).

SUMMARY

1. Under the conditions of these experiments Diemyctylus is
almost always markedly negative in its phototropic reactions to
white light, at ordinary temperatures.

2. At temperatures near 0° C. and 36° C. Diemyctylus is
indifferent to white light from above.

3. The above reactions are the same whether the light fall
from above or come from below, though they are usually less
marked in the latter case.

4. Diemyctylus is positively phototactic to lights of all in-
tensities, from very weak daylight to an intense arc light. -

5. At low temperatures this phototactic reaction is inhibited
or reversed.

6. With an intense white light at each end of the aquarium
the animals tend towards the less intense light; if neither light
be of great intensity, perhaps not reaching a certain optimum,
the animals tend towards the more intense light.

7. Phototactic reaction to pure red light was the same as
to white light, possible a little more marked.

46 A. M. REESE

8. The reaction to green light is the same as to the red, but
less marked.

9. The reaction to blue light is the same, but still less marked.

10. A small spot of white light from a micro-electric torch
produced no effect when thrown upon various parts of the
animal's body.

11. The animals responded promptly to a beam of sunlight
thrown on various parts of the body, either from above or below,
by a small mirror, though if the mirror threw a beam of 5 sq.
mm. or less there was little or no response.

REFERENCES

1. Pearse, A. S. The Reactions of Amphibians to Light. Proc. Amer. Acad.

1910. Arts and Sc, 45, pp. 162-206.

2. Reese, A. M. Observations on the Reactions of Cryptobranchus and Nec-

1906. turus to Light and Heat. Biol. Bull., 11, pp. 93-99.

3. Sayle, Mary Honora. The Reactions of Necturus to Stimuli Received Through

1916. the Skin. Jour. Animal Behav., 6, pp. 81-102.

ADDENDUM

As a check upon the preceding laboratory experiments, the
following experiments were tried upon a number of newts of
the same species, under as near natural conditions as could be
obtained. The work was done, during the latter part of August,
in a small, fresh-water pond, about two miles from the labora-
tory at Woods Hole, Mass.

Twenty-eight animals were obtained by sweeping a dip net
through the grass of this shallow pond. They were caught
during the morning, and were confined until night, and during
intervals between experiments, in a 12 in. x 12 in. floating live-
box, with wire top and bottom, which was partly filled with
grass and dirt from the pond.

During experimentation they were confined in a cage 1 ft. x
2 ft. in area, 6 inches deep, and open above, made of one-quarter
inch wire netting. This cage was sunk about 5 inches into the
water so that it was surrounded by the grass of the pond. A
few of the animals escaped, during the experiments, by climbing
out of the cage.

Only sunlight and artificial white light were used, the latter
being supplied by a miner's acetylene lamp with a reflector;
this lamp gave a fairly brilliant though rather variable light,
but its candle-power was not determined.

LIGHT REACTIONS OF NEWT 47

Experiment XXII. — This experiment was performed during a
fairly dark, moonless night. One-half of the wire cage was
covered with a board, while the other half was brilliantly illu-
minated by the acetylene lamp, fixed about 10 inches above
the surface of the water. Fifteen observations, at 5 -minute
intervals, were taken, during which 65 animals were noted in
the light half of the cage to 355 in the darkened half of the
cage, — a proportion of more than five to one. This proportion
would have been still greater but for the fact that after obser-
vations 2, 6 and 11 the light and dark ends were suddenly re-
versed, thus throwing the larger group of animals into the
light area.

Experiment XXIII. — In this experiment the same cage and
animals were used, but the light was bright sunlight. Of course,
on account of the diffused light, the shaded half of the cage
was not nearly so dark as in the preceding experiment. In 16
observations, at 5-minute intervals, 103 animals were counted
in the light half of the cage to 297 in the dark; this proportion
of nearly three to one would have been greater but for the sud-
den reversal of light and dark ends after observations 9 and 12.

It is evident, then, from experiments XXII and XXIII, that
under these conditions the negative phototropism to white light
is even more marked than in the laboratory experiments.

Experiment XXIV. — In this experiment the acetylene light
was placed in a large, glass aquarium jar, which was sunk into
the water of the pond so that the light was thrown into one
end of the wire cage, the observations being made, of course, at
night. This arrangement was not very satisfactory, as the dark
color of the pond-water made the illumination of the far end
of the cage very dim.

In 26 observations, at 5-niinute intervals, 213 animals were
noted in the half of the cage nearer the light to 263 in the farther
half. After observations 9, 20, and 22, since it was difficult to
reverse the ends of the cage, all the animals were pushed into the
light half; this tended to decrease the excess of those in the dark
end; but the experiment was hardly conclusive, perhaps on ac-
count of the unsatisfactory conditions.

Experiment XXV. — This experiment was performed with the
clear sun shining down upon the end of the submerged cage, at
an angle varying from 40 to 25 with the surface of the pond.

48 A. M. REESE

The cage being uncovered, the light was evenly distributed over
the bottom, but entered, as said, from one end. Under these
conditions, in 16 observations, at 5-minute intervals, 191 animals
were noted in the half of the cage towards the sun, to 120 animals
in the other half. After observations 8, 11, and 14 all the ani-
mals were gently pushed to the center of the cage, which dimin-
ished the preponderance of those in the half towards the sun.
This experiment seems to indicate that, where the light is suf-
ficiently bright, the animals tend to go towards it, as in the
laboratory experiments.

These outdoor experiments, then, seem to substantiate, so far
as they go, the results of the laboratory experiments.

THE INTERFERENCE OF AUDITORY HABITS IN

THE WHITE RAT

WALTER S. HUNTER

ASSISTED BY

JOS. U. YARBROUGH

The University of Texas

The present paper has grown out of the work which one of
the authors has already published on audition in the rat. 1 On
pages 324-5 of the last of these a report is made of some tests
on the retention of auditory habits. It was these tests that
gave us our cue. Negative results only had been secured by
attempting to train rats to turn in one direction through a box
when a tone or a chord was sounded and to turn in the opposite
direction when silence was given. This was a direct attack upon
the problem of tone sensitivity by the association method. It
occurred to us that working indirectly through habit interfer-
ence further data of value might be secured. By such a method
one could redetermine whether for the rat certain tones are
equivalent to silence. Should such a method succeed, its data
would be similar to that secured by the conditioned reflex method.
In the present paper we shall deal only with the tests bearing
upon habit interference. An immediately succeeding article will
stress the auditory sensitivity data secured by this method and
combine them with other observations from this laboratory.

The same T-shaped discrimination box was used here that
has been described in the previous papers. The buzzer was
held on a wire over the apparatus in the same location indi-
cated for the tuning forks. The initial plan (which was much
supplemented as will be seen) called for 20 rats as follows:

A. 20 rats train to turn rt. for handclaps, 1ft. for silence.

B. 4 ra ts of set A train for 30 days to turn rt. for buzzer.

1 Hunter, Walter S. The auditory sensitivity of the white rat. Journal Animal
Behavior, vol. 4, p. 215, 1914, and vol. 5, p. 312, 1915.

50 WALTER S. HUNTER AND JOS. U. YARBROUGH

C. 4 rats of set A train for 30 days to turn rt. for tuning

fork 256 d. v.

D. 4 rats of set A train for 30 days to turn 1ft. for tuning

fork 256 d. v.

E. 4 rats of set A train for 30 days on regular series of pre-

sentations on auditory stimulus.

F. 4 rats of set A tested for retention after 30 days rest.

G. Rats of sets B, C, D, E retested on handicaps.

This program calls for a measure of the relative retention of a
simple co-ordination in five groups of animals, each group having
been kept under different conditions for an interval of thirty
days.

Only 18 rats completed the work of set A. Of these 13 were
females (numbers 1, 4, 9, 10, 11, 14, 15, 16, 18, 19, 20, 21, 24),
and 5 were females (numbers 7, 8, 17, 22, 23). All were about
three months old at the beginning of the tests. With the excep-
tion of nos. 1 and 4, they were untrained. No. 1 had been
trained on the inclined plane problem box. No. 4 had worked
with light in a two-choice discrimination box. Nos. a, b, c, 25,
26, 27, 28, 29, whose records are given below, were also about
three months old at the beginning of the tests. All of these
animals were females. The tests here reported, like most studies
of animal learning, have been long and tedious. They have ex-
tended from January, 1915, to June, 1916.

Prior to the regular tests, each rat was fed on the experiment
table and was permitted to run through the box on each of two
days. Care was taken that no position habits were developed.
Those rats that manifested a preference for a certai n side of
the box were immediately forced through the opposite side.

Discrimination was regarded as established when the average
percentage of correct reactions for four days together was 87|%
with no one day's record below 80%.

II

Learning the first habit. — Table 1 gives the total number of
trials required by each rat to set up the habit of running to the
right for handclaps and to the left for silence. The period of
learning is the period up to the 40 trials made at the standard
per cent. The rats underscored are males. Figure 1 shows

INTERFERENCE OF HABITS IN THE WHITE RAT 51

the distribution curve. All but six of the rats had mastered
the problem within 500 trials. I am inclined to attribute the
irregularities largely to position habits which appeared during
the learning and which had to be overcome. Fear caused by-
punishment retarded the last part of the learning in rats 25-29.
No sex differences appear. The form of the learning curve will
be shown in section VIII.

TABLE 1
Number of Trials per Rat in Learning First Habit

Rat

Trials

Rat

Trials

Rat

Trials

1
4

260
210

16

17

260
300

a
b

350
420

7

430

18

450

c

460

8

500

19

450

25

420

9
10
11

300
390
370

20
21
22

710
610
620

26
27
28

550
320
610

14

370

23

470

29

570

15

360

24

260

Each rat of set A, when it had completed its four days with
a general average of 87 J%, was given three controls, each of
which in general alternated with a day on the normal stimulus
of handclaps. Control 1 — no stimulus was given. The reac-
tion was counted correct if it agreed with the series of presenta-
tions. This control was to test the animal's dependence upon
extra-auditory cues. Control 2 — an electric buzzer was sounded
in place of the handclaps. This control was given to each rat
on each of four successive days (40 trials). Control 3 — a tuning
fork, 256 d. v., placed over the apparatus as described in pre-
vious papers, was sounded by striking with a felt hammer. This
tone was substituted by the experimenter for the handclaps.
The necessity for the first control needs no comment. Con-
trols 2 and 3 were used in order to determine the relation of the
respective stimuli to the habit just established. It is very im-
portant, if interference effects are to be studied, that the mutual
relations of the stimuli (i.e., their transfer relations) be known, —
in the present case, whether the buzzer and the tuning fork
would be substituted readily for the handclapping in this par-
ticular co-ordination.

The results of control 1 indicate that the animals were depend-
ing upon auditory cues. Only one rat's (No. 17) record was

52

WALTER S. HUNTER AND JOS. U. YARBROUGH

n

100

ito

300 yoo

n

fl

JH

rs

ito

(,00

too

Figure 1. — Distribution curve of the original learning

the least ambiguous. The accompanying table (No. 2), how-
ever, will show that an extended use of the control produced
low percentages. Control 2, where the buzzer was substituted,
gave the following results: On the first day, 7 rats (Nos. 15,
18, 20, 22, 9, 10, 11) made below 80%. On the second day, 5
rats (Nos. 20, 22, 10, 11, 14) made below 80%. On the third
day, 3 rats (Nos. 21, 8, 11) made below 80%. On the fourth
day, Nos. 9 and 14 fell below 80%. Seven rats (Nos. 1, 4, 7,
16, 17, 19, 23) never fell below 80%. The adjustment or trans-
fer was thus usually made either at once or by the end of the
second day. On control 3, 6 rats at different times made a
day's record with 80% correct. This occurred, however, with
a majority of days at 50, 60 and 70% and is to be regarded
not as evidence of auditory sensitivity, but as an accident in
the grouping of kinaesthetic factors.

TABLE 2

Records on Controls Giving Correct Choices Out of 10
The text states that not all records are from successive days

Rat

15

16

17

18

19

20

21

22

23

1

4

7

8

9

10

11

14

Con. 1

6,5, 6

7, 4 5

8! 5, 7, 8, 6, 5

5, 6, 6
4,5,7

6, 6, 6
6, 6, 5

6, 5, 5
5, 5, 7

6.5, 6
6,7,5

7, 7, 6
5, 6, 6

6.6, 6
6,3,4

5, 3, 5

6, 5, 6

Con. 2

5, 10, 8, 8
8, 9, 10, 8

8, 9, 8, 9

6, 8, 9, 10
10, 10, 9, 9

7, 5, 10, 9

9, 9, 6, 9

7, 6, 8, 10

8, 8, 9, 9

10, 8

9, 10, 8, 9
9, 10, 8, 8
9, 8, 7, 8

6, 8, 9, 6

7, 5, 8, 10

7, 5, 6, 9

8, 7, 9, 6

Con. 3

7,6, 7
6, 6, 8, 6
8, 7, 7, 10, 7, 7
8, 9, 5, 8, 5
6, 7, 8, 7

6, 6, 5

7, 6, 7

6, 7, 4

7, 8, 7, 7
6,5,5
6,7,5
5,7,7
6,6,7
5,6, 5

7, 6, 6
7, 6, 8, 6
5,6,7

INTERFERENCE OF HABITS IN THE WHITE RAT

53

Table 2 presents the results for these controls. The four
days' records preceding control 1 were made at or above 87^%
correct. Each day's record with controls 1 and 3 alternated with
a day on the normal stimulus (handclaps) save when 80% was
made. In these cases the same control was used on the suc-
ceeding day.

Ill

Thirty-day Tests. — After control 3 had been given, each rat
of set A received the normal stimulus for four days or until the
standard 87 \°/ was reached. The animals were now started
upon the interference periods (B-E) as outlined above for 300
trials or 30 days. No rat learned his problem within this period.
Only one rat (No. 15) made 80% during any one-sixth (50 trials)
of the period. This rat made 82% during the third 50 and 84%
during the fourth 50 trials. (These are general averages for the
50 trials.) After this he broke down, so that on the final sixth
50% only was made. There is no available explanation for this.
Neither fear nor position habits intervened. It is one of those
anomalous cases that will occur. (It took this rat 180 trials
to relearn the normal habit. This, however, cannot be corre-
lated with the high percentage in the interference period because
long periods of relearning appeared in other rats where the high
percentages were absent.) Each of the succeeding 50 trials for
the rats in these sets averaged practically between 50 and 65%.
Table 3 contains a record of the number of correct reactions in
each 50 trials made by each rat during the 30 days.

TABLE 3

Trials Correct in Each 50 During the 30-day Test and the Relearning

30-day training

Sets

B

C

D

E

F

Rats 15

23

7

11

1

17

18

16

14 10

20

9

19

21

4 24 8 22

28

31

13

20

28

32

29

35

29 27

26

24

28

14

32

34

19

25

26

27

27

43

33 26

32

24

28

15

41

29

23

24

23

26

24

34

33 35

24

27

30

19

42

33

26

30

20

31

26

33

35 27

26

25

33

21

31

17

24

19

27

21

28

35

26 32

29

30

28

21

25

28

27

20

23

29

24

33

28 28

29

29

27

19

54

WALTER S. HUNTER AND JOS. U. YARBROUGH

Relearning

36 35 24
33 of 36
41 40 of
39 . . 40

46 37 38 35
.. 42 37 10
.. 41 35 of
.. 9 37 10
... of 40 . .

38 39 44 35
.. of 10 of
.. 40 of 40
. . . . 10

34 37 45
of of 9
40 40 of
. . . . 10

37
of
40

45 33 29

. . 37 39

.. 40 27

42 of

9 .. ..

41 30

of . .

.. 10 18 ..

16

10 . .

. . . . of

of

. . . . 20

20

Total trials on relearning

210 40 90

50 160 270 60

50 40 60 40

40 40 60

40 50 270 130

On the day following the close of the 30-day period each rat
was retested on handclaps to the right. The results are in-
cluded in table 3. Our criterion of the degree of retention has
been the length of the period of relearning rather than the per
cent of correct reactions on the first day. Table 4 shows that
in the present case there is no way of predicting the amount of
relearning from the percentages made on the first days. It will
be seen from table 3 that all sets of rats are essentially on a par
with respect to retention. In other words, so far as these rats
are concerned, 30 days of diverse training has not produced effects
in retention.

TABLE 4

F stands for number of trials correct in first 10 of relearning.
T is the total relearning time in trials.

B

D

E

Rat F T Rat F

Rat F T Rat F

Rat F

7
15
23

5 90
3 210
8 40

1 5 60

11 10 50

17 7 270

18 2 160

10 8 40

14 10 40

16 8 50

20 8 40

9
19
21

7 40
9 40
9 60

4 9 40

8 7 270

22 2 130

24 7 50

IV

Sixty-day Rats. — Four rats (Nos. 1, 11, 17, 18) had- been
tested in the work outlined above. During the 30-day period
an effort was made to train them to turn right for the tuning
fork 256 d. v. without success. These rats were then idle, al-

INTERFERENCE OF HABITS IN THE WHITE RAT 55

though kept in good physical condition, for the following inter-
vals of time: Nos. 1 and 11 went 10 days; No. 17 went 23 days;
and No. 18 went 40 days. At the close of these intervals of
time, all of the rats were brought back to the standard percent-
age of correct reactions on turning to the right for handclaps.
They were then put into training again on going right for 256
d. v. and left for silence. They remained in this series for 600
trials, 10 per day, punishment and reward. No. 11 was the
only rat of the four that improved during the 60 days. He
learned the reaction in 270 trials. The senior author was away
for the summer at this time and no control tests were made to
determine the basis of the response. Inasmuch, however, as
no other rat in the laboratory has learned to react to tone in
this fashion since the work was begun in 1913, and inasmuch as
this rat learned rapidly, it is most probable that the reaction
was due to secondary cues accompanying the tone. This ex-
periment is confirmatory of work previously published indicat-
ing the insensitivity of the rat to certain tones.

At the close of the 600 trials, retention tests for handclaps
to the right were given. No. 1 came back to standard in 10
trials; No. 17, in 60; and No. 18, in 30. This is practically
perfect retention and is as good a record as that made by the
30-day rats. The results are practically comparable, although
not absolutely so inasmuch as the 60-day animals were some-
what overtrained relatively on h. c. to the right.

The same results with the same limitations were secured with
rats 4, 8, 22 and 24. These were the rats listed under F in
the 30-day tests. The retention tests in that series brought
these animals back to the standard. They were then idle for
60 days at the close of which period they were again retested
on h. c. to the right. Rat No. 4 came back to standard in
20 trials; No. 8, immediately; No. 22, in 30 trials; and No. 24,
in 40. In order to compare the results given here and in the
above paragraph with those listed under ' ' Total trials on re-
learning " in table 3, it is necessary to subtract 40 from each
of the totals in that table. The results given in the present sec-
tion are the number of trials up to the 40 made at the standard
per cent.

Rats Nos. 7, 15 and 23 had been through the 30-day tests
in set B, — turn left for the buzzer. After intervals of rest as

56 WALTER S. HUNTER AND JOS. U. YARBROUGH

follows they were brought back to standard on handclaps: No.
7, 9 days; No. 15, 38 days; and No. 23, 47 days. They were
now retrained on going to the left for the buzzer and to the
right for silence. The intention was to train them upon this
for 60 days, unless the habit was established sooner, and then
test their retention of h. c. to the right. Rat No. 7 learned in
54 days, 540 trials; No. 23 learned in 35 days, 350 trials; and
No. 15 learned in 45 days, 450 trials. If we add to this only
the 300 trials which they had previously had on the same prob-
lem in the 30-day test, No. 7 learned in 840 trials; No. 15, in
750 trials; and No. 23, in 650 trials.

At the close of the 40 trials at the standard percentage for
rats 7, 15 and 23 as just noted, they were retested on h. c. to
the right. No one of the three fell below 80% for 30 trials.
In other words, there was perfect retention. When given con-
trol 1 — tests made without the auditory stimulus — the per-
centages ranged between 30 and 50. On one day and with only
one rat did it go as high as 70%. So there could be no doubt
that the rats were dependent upon the auditory stimulus. Here
we have a case where two opposite habits are present simul-
taneously in the organism although the respective stimuli were
not originally differentiated. The process of the differentiation
has been a successive formation of habits and not a simulta-
neous one as is usual in discrimination tests. And the inter-
esting thing is that the formation of the second (and opposite) habit
has not interfered with the retention of the first habit. A second
automatism has arisen gradually and independently of the first.

Further tests were made upon rat No. 7 to determine the
nature of the difference between the buzzer and the handclaps.
These results will be published in a separate paper.

V

Ninety-day Rats. — Three untrained rats, Nos. a, b and c,
were trained to go right for handclaps and left for silence. The
number of trials required in learning is shown in table 1.
At the close of this series, control 1 was alternated with
normal for three days in order to be sure that the animals were
not depending upon extra-auditory cues. The percentages were
all around 50. These three rats were then given a period of
idleness for 90 days. During this period, they remained in

INTERFERENCE OF HABITS IN THE WHITE RAT 57

splendid physical condition for experimentation. At the close
of the period, they were retested on h. c. to the right. It is
needless to give the data in detail. No one of these rats aver-
aged above 70% for any 50 trials although their retraining ex-
tended through from 34 to 45 days. Their behavior at the
beginning of the retesting indicated that the apparatus and
method were still familiar to them, but that was all. The re-
sults as a whole indicate that these rats had lost all measurable
traces of the original training. It may be well that in a habit
so difficult as the present one continued or retained familiarity
is too slight an aid to manifest itself in shortening the period
of relearning. The disintegration of this habit in the white
rat apparently takes place between 60 and 90 days. The 60-
day tests indicated practically perfect retention at the close
of that period, but the two sets of data are not strictly com-
parable. The rats in the 60-day tests had been retrained at
different intervals on h. c. to the right after the original learn-
ing. Hence the habit was considerably overlearned.

VI

Effect on retention of learned vs. unlearned habits. — It would be
interesting to know just what went on in the rats' nervous
systems during the 30 and 60-day periods of training. We seem
forced to assume that certain synaptic connections have per-
sisted in spite of the attempts of incoming stimuli to disintegrate
them. Inasmuch as either continued training (?) or the lapse
of time will result in the disintegration of these connections,
definite problems arise under each condition. We have indi-
cated that with the mere lapse of time, the dissolution of the
particular habit in our rats occurred between 60 and 90 days.
The present section contributes data throwing light upon the
comparative disintegrations brought about in the h. c. habit
by the 30 days' ineffective training on B and by a period of
training during which B was mastered.

Of the 18 rats used on the 30-day test described above, 9
made the standard 87 \°/ immediately upon being re-tested on
h. c. to the right. Four others did essentially as well. Two
hundred and seventy trials was the maximum period of re-
learning and was found in two rats. Table 3 gave the data in

58 WALTER S. HUNTER AND JOS. U. YARBROUGH

detail. It will also be recalled that no one of these rats im-
proved during his 30-days' training upon B.

Three untrained rats, Nos. 26, 27 and 29, formed the original
h. c. habit as indicated in table 1. They were then trained on
B until it was mastered. (I shall discuss certain details of this
training in a following section.) At the close of the 4 days
on B made at 87|%, these rats were retested on h. c. The
results for all save the original learning are given in table 5.

TABLE

5

Correct

in successive 50 trials in learning B

No. 25

No. 26

No. 27

No. 29

14

15

15

5

10

15

19

20

18

20

22

18

17

12

38

21

18

22

36

21

19

21

33

19

20

22

30

17

22

20

34

24

25

29

33

32

28

32

38

36

* 28

37

34

29

30

37

36

31

29

34

32

34

31

38

38

35

34

39

38

36

34

37

15 of 20

39

42

38

32

Unfinished

32

41

7 of 10

13 of 2

Correct in each 50 in

retest on h. c.

29

36

26

31

36

35

30

32

24

31

33

33

36

38

34

24 of 30

41
8 of 10

8 of 10

No 26 required 280 trials for the re-learning here in question.
No. 27 required 310; and No. 29, 260 trials. The intervals for 26
and 29 are a little too small inasmuch as these two rats grew
sick and died. Each, however, had reached 80% correct and
so was within 7|% of the standard. The indications, from this
test are that marked progress must be made in the formation
of a second contradictory habit before the retention of a first
habit is noticeably affected. This can be represented graph-

INTERFERENCE OF HABITS IN THE WHITE RAT 59

ically as indicated in figure 2. The three lines to the left are
based upon rats 26, 27 and 29. The three lines to the right
are based upon the 18 rats of the 30-day test. The first line in
each column represents the average number of trials in learn-
ing the original h. c. habit; the second line, the trials given
on habit B; and the third line, the re-learning time. The de-
tailed data have already been given in the tables. The rats
represented in the right hand column averaged about 5 months
old at the beginning of the relearning tests. This was approx-
imately 2 months younger than the other set of animals at the
corresponding point of their tests. Both sets were composed

w m

he -

8 gfe& 300

he m - 1™ tW 50

Figure 2. — Effects on retention of learned vs. unlearned habits.

of active animals, however, and in view of the marked difference
in results as compared with Hubbert's, 2 1 am inclined to discount
age as an important factor in determining the present data.

A comparison between these data and the results' of the 90-
day test points the way toward interesting interpretations. The
90-day rats had lost all measurable traces of the original h. c.
habit whereas rats 26, 27 and 29 relearned within an average
of 260 trials or 26 days. These three rats had spent 85 days
on habit B. Unless these are accidental variations, then, it
would seem that the training on B favored the retention of
h. c. The rats seemed equal in physical fitness for the tests.
If we now consider the relations of the data given in figure 1,
it would seem that the loss in retention of the first habit is prob-
ably caused as much or more by the lapse of time than by the forma-
tion of the contradictory habit. It was found in the 30-day test
that training had no greater effect on retention than lack of
training. It is thus suggested, although not clearly proved by
our tests, that the disintegration of certain habits in the rat is due
to a temporal factor and not to habit interference.

2 Hubbert, H. B. The effect of age on habit formation in the albino rat. Be-
havior Monographs, 2, no. 6, 1915.

60 WALTER S. HUNTER AND JOS. U. YARBROUGH

VII

The Strength of Habit.— Rats 25, 26, 27, 28 and 29, whose
learning periods were described in the first section of the paper,
were further tested as follows: At the close of the 40 trials at
87 1% made on the first h. c. habit, each rat was given control 1
on three days alternating with the normal. All of the rats
failed to respond correctly in this control. They were then
each given two consecutive days on control 2 (buzzer substi-
tuted for h. c). In case a rat fell below 80%, a day with the
normal stimulus was interpolated. The results are in table 6.

TABLE 6
No. 25 No. 26 No. 27 No. 28 No. 29

9

10

9

8

8

8

6
8

9

8

3

8

9

7

8
8

8

7

8

8

H.C..

Con. 2.
H. C.
Con. 2.
Con. 2.
H. C

It will be seen from this that rat 26 did not rate the buzzer as
identical with the handclaps and that No. 29 failed also, but
on the first day only. No 28 became sick on the third day
and was dropped from the tests. At the close of the tests in
the above table, Nos. 25, 26, 27 and 29 were immediately started
on learning "buzzer to left, right for silence " which was the
opposite habit to the extent shown in the table. The progress
of learning B in successive fifties was shown in table 5. The
very important point that I wish to emphasize is that no one
of these four rats learned in less than 770 trials while two were
as high as 910 and 920. It took these rats approximately twice
as long to break the h. c. habit as it had to form it. The figures
are: No. 25, h. c. habit-420, buzzer habit-850 (?); No. 26,
h. c. habit-550, buzzer habit-910; No. 27, h. c. habit-320, buzzer
habit-770; No. 29, h. c. habit-570, buzzer habit-920. These
figures exhibit in a striking manner the tenacity of habits in
the rat. The original habit need not be literally broken, how-
ever, because in each case a period of retraining reinstated it.
The situation is probably more accurately described by saying
that the first h. c. habit interfered with the formation of the
buzzer habit, although the latter but slightly (if at all) affected

INTERFERENCE OF HABITS IN THE WHITE RAT 61

the former. The amount of the interference will probably
depend much upon the ease of discrimination between the
stimuli for the two habits. We are not prepared to contribute
upon this point. (Because the rats ranked the buzzer as the
same as handclaps we have felt justified in assuming that un-
trained rats would learn " buzzer to the left ' as readily as
" hand claps to the right.")

Mrs. Binnie Pearce, in research from this laboratory as yet
unpublished, found even more striking interference in visual
habits. Using the same T-shaped box, she trained rats to
run one way for light and the other way for darkness. When
she then attempted to train them to reverse this behavior, the
task was found all but impossible.

We are not familiar with any other work where an animal
has had to learn the opposite of a previously acquired habit.
There are many cases where different habits have been set up
in succession and where interference has been more or less ex-
plicit. However, in order to secure comparable data, it is neces-
sary that the stimuli be known and the responses simple. The
study of interference in mazes, latch boxes, etc., suffers for this
reason. Not only must the stimulus be known in the case of
the first habit, but the second stimulus must be known physi-
cally and also physiologically in terms of the first one. Thus
one can know whether or not the stimulus for the second habit
is for the subject in that situation the same as the first stimulus
(positive transfer). Where the type of habit set up is kinaes-
thetic as opposed to auditory or visual, the control of the stim-
ulus is very difficult because the stimulus lies in the animal's
movements. The most feasible procedure is to reduce the
problem to such an extent that only one or two prominent kin-
aesthetic experiences are presented. The senior author is work-
ing upon this problem at the present time, although interference
is but one phase of the study.

VIII

Relative rates of error elimination in interfering habits. — With
particular reference to the 30-day rats and rats 25-29, it is of
interest to raise the following question: In what parts of the
learning curves does the interference, as measured by the rela-
tive rates of error elimination, occur?

62

WALTER S. HUNTER AND JOS. U. YARBROUGH

Table 7 gives data for rats 25, 26, 27 and 29. The numbers
represent the percentages correct in each succeeding one-tenth
of the learning process. 3 The first columns for each rat are

TABLE 7

25

26

27

29

Av.

50 23

54 29

50 31

50 23

51 26

57 34

63 29

50 45

70 39

60 36

57 37

38 35

43 75

64 42

50 47

57 38

60 45

59 64

68 35

50 45

78 50

54 49

43 64

61 58

59 55

66 53

72 67

59 70

64 64

65 63

64 60

61 71

75 68

70 64

67 65

59 59

76 73

65 71

77 69

69 68

59 68

76 75

75 71

82 82

73 74

78 76

74 71

81 75

66 75

74 74

87.5 90

87.5 87.5

90 87.5

90 90

88 88

the records for learning the original h. c. habit. The second

columns are the records for learning B. These figures are

secured as follows: No. 25, e.g., learned h. c. in 420 trials.

This is divided into 10 parts of 42 each. Of the first 42, 21 or

50% were correct. This method when applied to all members

of the group enables us to construct a curve which throughout

its length is representative of the group. The bottom numbers

in each column of table 7 represent the percentages of correct

reactions made in the last 40 trials. Sometimes this runs over

the standard 87.5%. The values above S are from the forties

made at or above the standard per cent.

If the curves of figure 3 are examined, the curve for B is seen

to start much lower than the curve for h. c. and to lag markedly

behind throughout eight-tenths of the learning. (These curves

are plotted from the average values in table 7.) This lag would

be even greater, but for the accidental fact that learning h. c.

was retarded toward the last by the fear that arose in the rats

from punishment. The marked interference of the two habits

is seen when the last of h. c. is compared with the first of B,

and also when the first parts of the curves are compared. B is

more than a new habit. It is interfered with from the start

by h. c.

3 This method of treating the learning process is taken from Dr. S. B. Vincent's
Function of the vibrissae in the behavior of the white rat. Behavior Monographs,
1, no. 5, p. 17, 1912.

INTERFERENCE OF HABITS IN THE WHITE RAT

63

Figure 3. — The relative rates of error elimination in the hand clapping habit and
the buzzer habit. Based on rats 25, 26, 27 and 29.

TABLE 8

15

23

7

Av.

52 57

57 48

39

46

49 50

52 57

61 65

48

50

53 55

61 64

61 60

51

48

57 57

36 55

70 65

55

53

53 57

63 68

59 54

62

48

61 56

63 75

59 54

69

57

63 62

80 55

76 57

83 .

51

79 52

88 71

76 68

67

55

77 64

83 75

72 ■ 65

86

59

80 66

66 71

78 51

67

61

70 61

95 90

90 90

87.5

90

90 90

Table 8 gives data for rats 15, 23 and 7, used in set B of the
30 and 60-day tests. In this table again the first columns are
the original learnings ; the last columns, the learnings of B in the

64

WALTER S. HUNTER AND JOS. U. YARBROUGH

60-day test. These rats had received 300 trials in the 30-day
test followed by some intermediate training on h. c. If this
data were included in the curves, there would be no variation
in their essential relations. If anything the interference would
be more apparent.

id

* 3 f r <> 7 Z ?

Figure 4. — Relative rates of error elimination in h. c. and in B.

on rats 15, 23 and 7

S

Based

The curves in figure 4 begin at essentially the same height
and go along together throughout the first six-tenths of the
learning. It is during the last four-tenths of the curves that
the B -curve remains markedly below that for h. c. (There is
no evidence that this was caused by fear.) The interference
of the two habits is seen here and in a comparison of the last
of h. c. and the first of B. In the average B is no more than
a new habit with these rats. Its curve begins no lower than
that for h. c. The details are further brought out in table 9,
which gives the correct responses in each 10 trials of the first

INTERFERENCE OF HABITS IN THE WHITE RAT

65

100 trials of the 30-day test with B. It will be seen from this
table that there is no essential difference between the initial
stages of the two habits.

TABLE 9

7

15

23

h. c.

B

h, c

B.

h. c. B

5

3

7

3

3 4

3

9

3

5

7 6

3

2

6

7

5 5

5

2

5

8

10 8

2

4

6

5

5 8

4

4

7

5

8 7

4

5

3

5

4 6

7

6

5

7

6 8

6

1

4

7

6 6

3

3

9

8

5 7

,

IX

Conclusions. — The present paper opens up problems in an all
but unexplored field of animal behavior. Keeping in mind the
limitations imposed by the number of animals and the type of
experiment, the following conclusions may be stated as the
more important ones to which our work points:

1. Habit interference occurs in the white rat between a first
habit and the formation of a second one.

2. This interference may or may not manifest itself at the
beginning of the second habit and may or may not manifest
itself later during the second learning.

3. ' Interference ' is most marked between the end of the
perfected habit and the beginning of the new habit. In many
cases this may show not genuine interference, but merely the
beginning of a new habit.

4. Habit interference may serve greatly to slow up the forma-
tion of a new habit. Clear evidence of this forward reference
has been found. We have brought to light no evidence that
learning the second habit as such interferes with the retention
of the first habit.

5. It seems clear that in some cases the lapse of time may be
more effective than intervening training in disintegrating a habit.

THE CRITERION OF LEARNING IN EXPERIMENTS

WITH THE MAZE

K. S. LASHLEY

The Department of Psychology of the Johns Hopkins University

In comparative studies of the rate of learning in which ani-
mals are trained in the maze the selection of a proper criterion
by which to judge the progress of habit-formation in different
groups of animals offers a rather difficult problem. There can
be little doubt that the ability to thread the maze without error
is the final test of learning, but whether a single trial without
error, three successive trials as used by Hubbert, or a still larger
number of errorless runs should be required before the habit
is considered as established has so far been determined largely
by the convenience of the experimenter. The question is chiefly
one of economy of the experimenter's time, but not wholly so,
for, although all animals may become automatic in running the
maze after long training, an occasional error still appears and
no method of evaluating these has been devised.

In some tests dealing with the effects of drugs upon the rate
of learning I have recently trained 94 rats in the Watson cir-
cular maze, obtaining data which makes possible a limited com-
parison of such criteria of learning.

The animals were all given five trials per day in the maze
with food at the end of each trial. At the beginning of the
experiments, as an arbitrary standard of ' perfect learning,"
a single record of three successive errorless trials on the same
day was selected. After this degree of proficiency is once at-
tained the animals make very few errors, so that this standard
actually represents very nearly the limit of training, but it was
chosen simply because it could be attained after about ten
days' training.

To test the reliability of this standard in estimations of the
difference beween groups of animals its results have been com-
pared with those of another standard, that of the number of

THE CRITERION OF LEARNING 67

trials preceding the first which was made without error. This
comparison is best made by correlating the number of trials
preceding the first errorless run with the number preceding
' ' perfect learning ' ' for all the animals. The former varied from 10
to 75 with a mean at 23.8±.977, the latter from 10 to 150 with
the mean at 47. 3± 2.99; the correlation in the variations of the
two is 0.632±0.061. The coefficient of regression of the varia-
tions in trials preceding the first errorless run over those pre-
ceding " perfect learning " is 1.304, that of variations in " per-
fect learning " over first trial is .306. This means that if we
are dealing with fairly large numbers of animals and have found
a given difference between two groups, as measured by the aver-
age number of trials required to make one perfect run, we may
expect that the difference in the number of trials required for
" perfect learning " will be in the same direction and 1.304 times
as great. Conversely, if we know the difference in trials re-
quired for " perfect learning " we may predict a difference .306
times as great in the number of trials required for one error-
less run.

It follows from this correlation that that group of animals
which has made the most rapid progress up to the time when
the first errorless run is made will continue in the lead until the
limits of training are reached; will, indeed, increase that lead.
As a test of the application of this principle, the groups of ani-
mals which were treated differentially in the experiments have
been graded in the order of the average number of trials re-
quired by them to attain to each of the two standards. The
results of this are shown in table 1. The different methods of
rating result in an interchange in the order of some of the groups
but in no case is the position of any one group changed by more
than one place.

The groups included in the table are not all strictly compar-
able. The methods of training were the same in every case
but some of the groups differed in the heredity and age of their
members, in the season during which they were trained, as well
as in certain drugs administration during training. In the sepa-
rate experiments, all these factors were controlled and the groups
a, J, g, and i, c, and d and b, c, h, and j are mutually comparable
and differ only in the drugs administered. The order of these
by the two criteria of learning is —

68 K. S. LASHLEY

" P. L." a f g i: c d: b e h j

1st P. R a g f i: c d: b e h j

The order is changed in this case only between the groups f
and g, and the difference between them is not great enough to
be significant in either case. There is essential agreement in
the results obtained by the two criteria.

As will be noted in the table and from the coefficients of
regression, the difference between the groups is greatest when
measured by the difficult standard of three perfect trials. 1 Are
these differences more significant on this account? At first sight
it might seem so. The number of animals considered remains
constant and hence, other things being equal, the ratio of the
difference to its probable error increases. But the probable
errors are dependent also upon the amount of variability and a
further analysis of the data shows that the coefficient of varia-
tion remains constant or is even increased when the more diffi-
cult standard is used. The figures in table 2, which are taken
from groups c and d, illustrate this. The probability that
the first difference in the table (3.54) is due merely to chance
is about 1/3; that the second (4.12) is due to chance is 1/1 or
greater. A glance at the probable errors for the averages of all
the rats (page 70) shows that these are quite consistent with
the results for the smaller groups. 2 The coefficient of variation
in the number of trials preceeding the first errorless run is .5900,
for those preceeding " perfect learning " is .6107 and the prob-
able error of the average of the latter is proportionately greater
than that of the former . If two such groups were compared by
the two criteria the differences obtained would obviously bear
the same relation to their probable errors as do those in the
smaller groups.

The general results of this analysis point to the following
conclusions : 1 . Where there is a difference in the average capacity
of two groups of animals for habit-formation, the more difficult
the problem that they are required to learn the greater will be
the apparent difference between the groups in the practice re-

1 Some exceptions occur, but this is to be expected from the small number of
animals included in the groups.

2 No great importance could be ascribed to this fact alone as it does not follow
that there is any correlation between the variability within the subordinate groups
and the variability of all the animals taken together, but the fact that the same
results are obtained for both the small and large groups does seem significant.

THE CRITERION OF LEARNING 69

quired for learning. 2. With the increasing difficulty of the
problem there is an increase in the extent of variation between
the members of the same group so that the greater difference
between the groups looses its significance through the increase in
the probability of chance variation of the averages. 3. Hence
there is no advantage, for reliability of results, in prolonged
training where the problem is that of a statistical comparison of
different groups of animals by a single standard of achievement.

These conclusions apply only to a specific technique, but one
which has been used extensively in studies of the effect of age,
sex, distribution of practice, etc., upon the rate of learning. It
may be argued that long training permits the comparative study
of the rate of learning at different stages of proficiency. This
is quite true, but the analysis of learning curves based upon
the averages of several animals has contributed remarkably
little to our knowledge of the mechanism of learning and in
statistical studies of the sort under discussion there is not time
for that detailed analysis of the individual behavior of the
subjects which is of value in the interpretation of the form of
the learning curve. On the other hand the results of studies
of the modifiability of the course of learning by environmental
factors are for the most part questionable because of the small
number of cases upon which they are based. In many cases
differences which are smaller than their probable errors have
been regarded as significant, seemingly only because they sup-
port the hypothese of the writers.

The use of an adequate number of animals is difficult for

the reason that the groups to be compared should be trained

at the same time to rule out possible seasonal differences, of

which we know nothing at present, while only a limited number

of animals can be trained by one man at one time. A possible

solution of the difficulty is the cooperation of several students

upon a single problem but there is not enough data upon the

influence of the experimenter's personal equation to permit of

this as ye t. 3 The alternative seems to be the simplification of

3 The use of two or more criteria as in the experiments reported, while reducing
the probable errors of the average difference found, removes hereditary and like
individual differences from the category of chance variations and places them
on an equal footing with the experimental differences (age, sex, or whatever differ-
ence is being studied) as the cause of the diverse rates of learning revealed by the
experiments.

70

K. S. LASHLEY

the problems presented to the animals so that a greater number

may be trained. If the evidence given above can be verified

by more extensive data this solution will doubtless prove to be

the most satisfactory.

TABLE 1

The average number of trials required by differentially treated groups before
reaching the standards described in the text. The number of animals from which
the averages were taken is given at the left and the relative rating of the groups by
the two standards on the right.

Trials

Trials

Number

preceeding

preceeding

Rating

Rating

Group

of

"perfect

1st perfect

by

by

animals

learning"

runs

"P. L."

1st P. R.

a

9

24.5

14.8

1

2

b

6

30.0

14.3

2

1

c

16

31.0

16.6

3

3

d

16

35.2

19.6

4

5

e

6

42.5

18.0

5

4

f

10

43.5

23.0

6

7

g

10

48.6

20.4

7

6

h

6

65.3

31.3

8

8

1

9

74.4

32.4

9

9

J

6

82.6

43.0

10

10

TABLE 2
Differences between groups c and d as measured by the two criteria of learning

Group

Trials preceding first
errorless run

Trials preceding three
successive errorless runs

Mean

Probable
error

Coef.
of Var.

Mean

Probable
error

Coef.
of Var.

d
c

19.60
16.06

1.480
1.684

.448
.622

35.12
31.00

5.922

2.428

1.000
.464

Difference

3

.54±2.26

3

i

1.12±6.4C

I

THE REACTIONS OF DROSOPHILA AMPELOPHILA

LOEW TO GRAVITY, CENTRIFUGATION,

AND AIR CURRENTS

WILLIAM H. COLE

Contributions from the Zoological Laboratory of the Museum of Comparative

Zoology at Harvard College

No. 288

INTRODUCTION

Geotropism is characteristic of many animals and is often
closely correlated with equilibration. The ear in vertebrates
and the statocysts in invertebrates are evidently concerned with
this reaction. In insects, however, there are no semi-circular
canals or statocysts and it has not been proved that the so-called
" static " organs (chordotonal, etc.) have to do with geotropism.
Some other explanation is therefore to be sought. The experi-
ments here described were carried out with the common fruit-
fly, Drosophila ampelophila Loew, for the purpose of deter-
mining (1) whether or not it is negatively geotropic; (2) how it
responds to centrifugation and air currents; and (3) what mechan-
ism can control these responses.

Carpenter ('05) concluded that gravity acted on Drosophila
as a ' directive ' stimulus only, some ' kinetic ' stimulation,
such as photic or mechanical, being necessary to induce loco-
motion. If this is true, how will Drosophila react to centrifugal
force and air currents under conditions where light and mechan-
ical stimuli are not effective ? This question was suggested by
the fact that the flies, without mechanical stimulation, were
found to respond negatively to gravity in the dark as well as
in the light. If it should be found that Drosophila reacts nega-
tively to centrifugation or to air currents, then it would seem
that gravity is a kinetic stimulus as well as a directive one.
Another question closely related with this one, which must be
considered is, by what means is the stimulus of gravity received?

72 WILLIAM H. COLE

The work was done under the direction of Professor G. H.
Parker, to whom I wish to express my sincere thanks for guid-
ance and suggestions throughout its progress.

EXPERIMENTS

1. Effect of gravity in the dark. — The first experiments were
carried out in a dark box modelled after the one described by
Carpenter, except that no heat screen was used. 1 The glass
cylinder employed was 18 cm. long and 4 cm. in diameter, and
was marked off by fine ink lines into six regions of equal length,
to facilitate locating the flies at the end of the experiments. A
small number of flies were put into the cylinder and attracted
to the top end by a strong light. Quickly but carefully the
cylinder was placed, this end down, inside the box. After a
period of one minute the door was opened, the lights turned
on and the position of the flies noted. Observations were also
made with a single animal, with smaller and larger cylinders of
celluloid as well as of glass, but since the results were always
the same it is not necessary to describe these modifications in
detail.

The results of 58 trials involving 26 different animals showed
that an average of 82 per cent went to the uppermost third of
the cylinder after it was inverted, that 4.8 per cent remained
in the lowest third and that the others stopped creeping in the
middle third. The individual readings for those at the top
varied from 67 to 92 per cent. In other words the animals re-
acted negatively to the stimulus of gravity in the dark. Whether
or not this response is due entirely to gravity without regard
to the mechanical stimulus of turning them over will be con-
sidered later.

One of the sets of records in this series of experiments is given
in Table I.

2. Effect of gravity on flies equally illuminated from above and
below. — The dark box was converted into a light box by the intro-
duction of two electric lights, one at each end. These were either
carbon-filament lamps of 16 candle power or 15 -watt Mazda
lamps. As before, the flies were attracted to the top of the
cylinder, which was then inverted and placed in the light box.

1 Carpenter's heat screen, because of the thinness of the water layer, was prob-
ably of no great value in preventing the action of the heat on the flies.

THE REACTIONS OF DROSOPHILA

73

TABLE I

Showing the position of 5 flies in 14 trials, after having been in the dark box
one minute. At the beginning all the flies were in section 6. 85.71 per cent crept
to the uppermost third of the cylinder (sections 1 and 2).

Number of Flies in the Different Sections of Cylinder

Trial number.

1
5

2

4

3

4

5

6

7

8

9

10

11

12

13

14

Total

Section 1

3

4

4

3

4

4

5

3

2

3

3

4

51

Section 2

1

1

1

1

2

1

2

9

Section 3

1

1

1

1

1

5

Section 4

1

1

1

1

1

5

Section 5

Section 6

After one minute the readings were taken. Eighty per cent of
the flies used in 50 trials went to the top section, 9 per cent
remained at the bottom and 11 per cent went to the middle.
Here also the flies responded negatively to gravity.
A set of records from this series is given in Table II.

TABLE II

Showing the position of 5 flies in 14 trials after having been one minute in the
light box with equal illumination at top and bottom. 78.57 per cent crept to the
uppermost third.

Number of Flies in the Different Sections of Cylinder

Trial number.

1

2

3

4

5

6

7

8

9

10

11

12

13

14

Total

Section 1

3

5

5

2

3

5

3

3

2

4

1

3

2

1

42

Section 2

2

1

2

1

1

1

1

1

1

2

13

Section 3

1

1

2

4

Section 4

1

1

1

1

4

Section 5

1

1

1

3

1

7

Section 6

3. Effect of gravity on flies illuminated either from above or
below. — In order to study the effect of unequal illumination, a

74

WILLIAM H. COLE

single lamp was used either at the top or the bottom. When
the top lamp was lighted 98.5 per cent of the flies went to the
top after one minute, the others reaching the middle section.
Twenty trials with 12 different animals were made.

With illumination from below 70 trials on 21 flies resulted
in 61 per cent going to the uppermost third and 22.5 per cent
remaining in the lowest third. Thus when light acts contrary
to gravity a smaller number of flies are found at the top. It
is interesting to note that the light stimulus, contrary to expec-
tation, did not predominate over gravity. An increase of the
light intensity from 16 candle power to 40 made no difference
in the results.

A set of records from an experiment in which the light was

below the cylinder and therefore acted contrary to gravity is

given in Table III.

TABLE III

Showing the position of 5 flies in 14 trials after having been in the light box with
a single lamp (15-watt Mazda) below the cylinder; 55.7 per cent crept to the upper-
most third.

Number of Flies in the Different Sections of Cylinder

Trial number.

1

2

3

4

5

6

7

8

9

10

11

12

13

14

Total

Section 1

2

2

3

2

1

2

1

3

2

4

2

2

2

1

29

Section 2

1

1

3

1

2

1

1

10

Section 3

1

1

2

1

1

6

Section 4

2

1

1

1

o

Section 5

1

1

1

1

1

2

1

8

Section 6

1

1

1

1

1

2

1

2

1

1

12

These results corroborate previous work on this subject in
so far as a negative response to gravity is concerned. But in
this, as well as in all previous work, mechanical or light stimuli
have been operating. The former cannot be eliminated in such
experiments since it is impossible to invert the cylinder without
moving it. Consequently I next tried a series of centrifuging
experiments in which these two kinds of stimuli could be neg-
lected. So far as I am aware the effect of centrifugal force on
Drosophila has never before been studied.

THE REACTIONS OF DROSOPHILA 75

4. Effect of centr if ligation. — The centrifuge consisted of a
table mounted on a base capable of revolving, on a fixed axis
in a horizontal plane. A small water motor attached to an
ordinary faucet furnished the motive power. On the table could
be fastened glass tubes of various lengths and bores. In these
tubes, the ends of which were tightly corked, one or more flies
were placed in the desired position; the tube was then revolved
about its middle point as a center at a known speed, the time
of revolution usually being one minute.

In the preliminary trials it was found that at a certain speed
the flies in the ends of the tube crept toward the center and
remained there. If the speed was greatly increased, they were
thrown out to the ends. It became necessary therefore to de-
termine the maximum and minimum limits within which a
definite response could be noted. The calculation was made

according to the formula, F = where F represents the

r

centrifugal force, m the mass, v the velocity of revolution and r

the radius. Experiments showed that when F was equivalent to

gravity the flies began creeping toward the center. When it was

considerably larger than gravity the flies were thrown out

to the ends. Furthermore when F was just equivalent to gravity

the flies crept toward the center until they reached a point

where the force was less than gravity, the speed remaining the

same. To induce further creeping toward the center the speed

had to be increased, since the shorter the radius of revolution

the greater the speed necessary to generate the same force.

The tube ordinarily used was 50 cm. long with a diameter

of 2 cm. Applying the formula, n = , which can be de-

47r2r

rived from the previous one, the speed necessary to generate a
force equivalent to gravity is easily calculated. When the flies
are in the ends of the tube, therefore, it must revolve approx-
imately once every second; as they move toward the center
the speed must be gradually increased. But since the flies can
creep against a force much greater than gravity without losing
their equilibrium, a constant speed can be found at which they
will creep all the way to the center. This is about 85 revolu-
tions per minute. Experiments carried out in darkness, in dif-

76

WILLIAM H. COLE

fuse daylight, and with a bright light at one side all gave similar
results. A check experiment, in which the speed of revolution
was very low (from 1 to 40), showed the flies creeping about
indifferently; therefore it was concluded that mechanical and
light stimuli did not affect the response in these experiments.
One hundred trials with 40 different animals, under the various
conditions described above, showed that a speed of 60 revolu-
tions per minute was necessary to start them moving toward
the center. As the flies approached the center the speed had
to be gradually increased in order to keep them moving toward
the center. At a distance of 2 cm. from the center a speed of
approximately 210 revolutions per minute was necessary to ac-
complish this. At a distance of 25 cm. from the center any
speed greater than 100 revolutions per minute mechanically
prevented the flies from creeping toward the center. Table IV
gives the data for several trials taken at random from the series

of 100.

TABLE IV

Showing the position of 14 flies in 8 trials after one minute of revolution at dif-
ferent speeds.

Number

of

flies

Position

at
beginning

Number

revolutions

per min.

Time

of

revolution

Position at end

of

experiment

2

End

50

1 min.

End

2

End

72

1 min.

\ way from end

1

End

90

1 min.

Center

1

f way from end

120

1 min.

End

2

I way from end

72

1 min.

f way from end

2

\ way from end

96

1 min.

Center

3

End

100

1 min.

End

1

2 cm. from center

210

1 min.

End

These experiments demonstrate a very definite response and
prove that Drosophila reacts negatively to a centrifugal force
equal to or slightly greater than gravity, as well as to a gravi-
tational one, without regard to other stimuli. We may there-
fore consider gravity a kinetic stimulus as well as a directive one.

THE REACTIONS OF DROSOPHILA 77

5. Effect of air currents. — The next question considered was,
How does Drosophila respond to air currents? Horizontal, up-
ward, and downward currents, produced by an electric fan, were
directed into a glass cylinder like the one used in the dark box.
Their strengths were so adjusted that the flies did not lose
their equilibrium.

(a) Horizontal currents. — These trials, carried out in diffuse
daylight, did not give as definite a response as could be desired.
The flies were liberated singly from the bottle containers at the
open end of the cylinder, and their course of locomotion noted.
In only 11 trials out of the 40 made, could the response be called
definite. In 7 of these the flies crept against the current, in 2
they crept against it for about 10 cm. and then flew with it, and
in the other 2 they flew with the current. Every case of creeping
was against the current and every case of flying was with the
current. In the control experiments with no air currents the
flies crept or flew in any direction.

(b) Upward vertical current. — In these trials 29.5 per cent of
the flies crept upward with the current, 59 per cent flew upward,
and 11.5 per cent crept downward. Gravity is here acting con-
trary to the force of the current and the 29.5 per cent creeping
up is probably a purely negative geotropic response. The creep-
ing downward was very slow and intermittent. The largest
number (59 per cent) flew with the current.

(c) Downward vertical current. — The results of 61 trials showed
that 27.8 per cent of the flies crept upward against the current,
23.2 per cent flew upward, while 49 per cent flew downward with
the current. An interesting observation was that practically
all the flies crept upward a short distance before carrying out
the main response. In the control experiments, with no air
current and the cylinder in a vertical position, the only reaction
that could be noted was a negative geotropic one, the other
movements being indifferent.

There is therefore a tendency for Drosophila to fly with the
air current, a positive response, and to creep against the current,
a negative response. Since there were extremely few flying
responses in the experiments with gravity and centrifugal force,
no comparisons can be made with them. But the creeping
against the currents corresponds with the negative response to
gravity and centrif ligation.

78 WILLIAM H. COLE

DISCUSSION

The responses, other than mechanical ones, of animals to
centrifugal force and air currents have not been thoroughly
investigated. Only a few references to centrifuging experiments
are found in the literature. Loeb ('91) stated that Cucumaria
cucumis responds to centrifugation by contracting its body and
remaining motionless. This condition persists for from one-
quarter to one-half an hour afterward, when crawling is begun
again. Although he studied the geotropic reactions of certain
caterpillars, ephemerid larvae, coccinellids and blattids, no refer-
ence is made to testing the effect of centrifugal force on them.

Jensen ('93), having found that Paramoecium was negatively
geotropic, discovered that it moved centripetally with weak cen-
trifugation. Davenport and Perkins ('97), after concluding that
' gravity acts as an irritant to which the organism makes a
response, belonging to the category of adaptive responses," say
that this irritating pressure " may be replaced by a centrifugal
pressure, when the same geotactic orientation will occur." Har-
per ('11) also reported that Paramoecium reacted negatively
with weak centrifugation. He believes, however, that " the
response of Paramoecium to gravity is a purely mechanical
tropism."

On the other hand, geotropism of animals has been exten-
sively studied, and many theories put forth for its explanation.
It is generally accepted that the ear or some ' static " organ
controls this tropism in certain forms. But for insects there is
much doubt as to how the stimulus is received. Kafka ('14)
reviews this question and summarizes the different theories, as
follows: Loeb believes that the chordotonal organs at the base
of the halteres of some Diptera are the organs of reception.
Pfliigstaedt and Weinland describe other structures which might
serve as sense organs. Similar organs have been described by
Hochreuther for Dytiscus, by Janet for Hymenoptera, and by
Baunacke for nepid larvae. But conclusive proof that any of
these organs, the functions of which are little understood, control
the response to gravity is entirely lacking.

The reactions to the three kinds of forces described above
suggest an explanation as to how the stimuli are received. When
the fly is creeping upward against gravity the weight of the
body is on the legs. There is, therefore, a tension on the leg

THE REACTIONS OF DROSOPHILA 79

muscles distinct from that caused by creeping. When a fly is
creeping against centrifugal force a similar tension of the leg
muscles is produced. Furthermore, creeping against an air
current causes the same kind of tension. Very probably, then,
tl;e stimuli in all three cases are due to this tension and are
received by the sensory nerves of the leg muscles, the response
being an attempt to preserve the equilibrium of the body. Nega-
tive geotropism in Drosophila, then, is concerned with the muscle
sense. Radl ('05) expressed the view that the insect muscles are
capable of acting as special sense organs when he wrote ' ' das
Gehor der Insekten ist ein verfeinertes Muskelgefuhl."

The flying response does not fit into this explanation and it
may be that it is not influenced at all by gravity. It is a matter
of common observation that the house flies on a brightly illu-
minated window usually creep upward but fly in all directions.
The flying is much more indefinite than the creeping. In my
observations on geotropism only a very few cases of flying
(about 3 per cent) were seen. Cylinders with a diameter as
large as 12 cm. were used so as to allow flying, but no greater
proportion of cases was seen than in the smaller cylinders. When
disturbed the animals flew about indifferently for a short time
and then, after alighting, continued their upward creeping. In
the centrifuging experiments no flying at all was seen. The
air currents often caused flying, and in the large percentage of
cases the animals flew with the current, although they were able
to withstand it. It seems therefore that the response to gravity
is much less marked in flying than in creeping, where it is very
definite.

CONCLUSIONS

1. Drosophila ampelophila Loew, when creeping, reacts nega-
tively to gravity, to a centrifugal force which is equal to or
slightly greater than gravity, and to air currents without regard
to other stimuli. Gravity is, then, a kinetic as well as a direc-
tive stimulus.

2. The stimuli causing these reactions are probably received
by the sensory nerves of the leg muscles.

3. It is probable that flying reactions of Drosophila are not
influenced by gravity.

80 WILLIAM H. COLE

REFERENCES

Carpenter, F. W. The Reactions of the Pomace Fly (Drosophila ampelophila

1905. ' Loew) to Light, Gravity and Mechanical Stimulation. Amer. Nat.
vol. 39, pp. 156-171.
Davenport, C. B., and Perkins, H. A Contribution to the Study of Geotaxis in

1897. the Higher Animals. Jour. Physiol., vol. 22, pp. 99-100.
Harper, E. H. The Geotropism of Paramoecium. Jour. Morph., vol. 22, pp.

1911. 993-1000.
Jensen, P. Ueber den Geotropismus niederer Organismen. Arch. ges. Physiol.,

1893. Bd. 53, pp. 428-480.
Kafka, G. Einfiihrung in die Tierpsychologie auf experimenteller und etholo-

1914. gischer Grundlage. Leipzig, 8vo., xii+593 pp.
Loeb, J. Ueber Geotropismus bei Thieren. Arch. ges. Physiol, Bd. 49, pp. 175-

1891. 189.
Radl, E. Ueber das Gehor der Insekten. Biol. Cenlralbl., Bd. 25, pp. 1-5.

1905.

GEOTROPISM IN PLANARIA MACULATA

J. M. D. OLMSTED

Contributions from the Zoological Laboratory of the Museum of Comparative

Zoology at Harvard College

No. 289

Flat-worms, such as planarians, are commonly collected from
the underside of stones in a stream or pond (Bardeen, '01 ; Pearl,
'03; Whitehouse, '14). The resting position of these animals,
with their ventral surfaces uppermost, would seem to indicate
a negative response to gravity, since when moving they may
be in any position, depending upon the particular surface over
which they happen to be gliding. This investigation has as
its object the analysis of the resting behavior of these worms.
The specimens used were Planaria maculata Leidy and were
taken from Fresh Pond, Cambridge, Mass. A stock was brought
into the laboratory and kept in a large jar on a table about
four feet from a north window.

To ascertain the relative importance of light and gravity in
the reactions to be studied, an experiment in the following form
was carried out. One-half of one surface of a glass plate, 10 x
8 cm., was coated with black wax. This plate was supported
in a horizontal position by wax feet 4 mm. high on a second
glass plate. The pair were placed in a flat dish and covered with
water to the depth of 3 cm. The flat dish had a collar of black
paper about its sides, so that only light from above could fall on
the plates. Then twenty planarians were placed at one time on
the upper plate, at another on the lower one, and their positions
recorded twice a day. As the animals moved about over the
whole dish for an hour or more after beginning the experiment,
the fact that they had started from the upper or lower plate
was not significant.

The results of 30 readings showed that 30 per cent were not
under the plates, but usually in the shadow near the angle
between the bottom and side of the dish, 70 per cent being

82 J. M. D. OLMSTED

found between the plates, and always under the black half. Of
the latter, one-fifth were on the under side of the upper plate
(ventral surface up), and four-fifths on the lower plate (dorsal
surface up).

Since, as the preceding experiment showed, the influence of
light was so marked, it was decided to eliminate this factor
by conducting all the subsequent experiments in a light-proof
box. To eliminate thigmotropism and to provide a contin-
uous surface which should have all possible relations to gravity,
spherical balloon flasks were used. These flasks were 13 cm. in
diameter. They had a short neck 4 cm. long. Three regions
of equal area were marked off on the surface of each flask, a ring
about its equator and a segment at either pole. These three
regions were designated top, middle, and bottom. The flasks
were so marked that in one the neck came in the top, in another
in the middle, etc. In the experiments very few worms lodged
in the neck and the per cent of such was practically the same
whether it occurred in flasks with the neck in the top area, in
the middle, or in the bottom. In the tabulation of results
worms in the neck are not included.

Twenty worms were used in each experiment. Readings of
their positions were made at 9 A. M., 1. P. M. and 4. 30 P. M.
In a few cases readings were taken at intervals of two hours,
but even then the animals were at rest. They were made to
start moving before being returned to the box, as a means of
redistributing them for the beginning of another trial. It was
found that the positions of the planarians in the flasks changed
greatly during the first few days after being put into the dark.
At first the majority were to be found in the bottom of the
flasks. A few days later they were equally distributed in the
three areas. When they were fed there was a sudden depar-
ture from this equal distribution and the majority would be
found in the top. They again distributed themselves equally
in the three areas three or four days after feeding. Table I
gives a summary of results. The numbers are the per cents
taken from 10 or more readings. By ' from light ' is meant
worms taken from the stock which had been kept in the light.
" From dark ' means worms which had been in the dark-box
for a week or longer. ' Fed ' worms are those which were
fed on liver on the day of the experiment or every other day

GEOTROPISM IN PLANARIA MACULATA 83

during experimentation. ' Unfed ' are those which had been
without food for five days or more.

From these results it is evident that two factors are con-
cerned in the distribution of the worms: First, previous history
as regards exposure to light, and second, the state of metabolism
of the worms in relation to feeding. Both fed and unfed worms
which had previously been in the light were found to be mostly
positively geotropic immediately after being put in the dark.
The fed ones then became negative for a short time. Finally
both became indifferent if feeding was stopped. Those which
had been in the dark for a long time were negative when fed
and indifferent when unfed. Walter ('08) makes the statement
that Planaria gonocephala " seems, after several hours of ex-
posure to the dark, to be positively geotropic," while Kafka
('14, p. 151) says that Planaria gonocephala is negatively geo-
tropic after long retention in the dark. Both of these appar-
ently contradictory statements are probably true, since the length
of exposure to the dark may very well be an important factor
in the geotropism of Planaria gonocephala, as my experiments
show for its close relative, P. maculata.

That this negative geotropism of fed worms in the dark is
not in reality a response to oxygen from the open neck is shown
by the following experiment. A flask containing 20 planarians
was completely filled with water, and the mouth covered by a
glass plate. It was then immersed neck downwards in a jar
of water in the dark-box. Previous to the experiment the
planarians had been fed every other day for two weeks, and were
dividing so that at the end of the experiment there were 27
worms instead of 20. The per cents found in the three areas
of the flask under these conditions were as follows: Top, 58;
middle, 33; and bottom, 9. These are of the same order as the
last two series of the per cents given in Table I. Table II shows
this relationship.

84

J. M. D. OLMSTED
TABLE I

Area of the Flask

Month

Top

Middle

Bottom

Unfed

1st 2 days in box. . .

19

10

71

Nov.

1st 2 days in box . . .

17

21

62

Jan.

From light

5+davs in box

38

24

38

Nov.-Jan.

Fed

before

expt.

1st 2 days in box . . .

16

21

63

Nov.

3rd, 4th days in box.

63

21

16

Nov.

5th day onward. . .

36

23

41

Nov.

Unfed

36

30

34

Nov.

38

35

27

Dec.

35

36

29

Jan.

From dark

Fed
contin-
uously

58

28

14

Dec.

68

26

6

Jan.

TABLE II

Area of Flask

Month

Top

Middle

Bottom

Flask open to air

58

28

14

Dec.

From dark
and fed

68

26

6

Jan.

Flask submerged in water . . .

58

33

9

Jan.

Table II shows that the per cents were the same whether
the flasks were open to the air or entirely submerged in water.
If the worms had been responding to oxygen and not to gravity,
we should expect in this experiment to have found them in the
bottom near the mouth, where oxygenated water could enter
from the jar outside. They were actually found in the region

GEOTROPISM IN PLANARIA MACULATA

85

farthest from the supply of oxygen, so that their position was
a true response to gravity.

To find whether the presence of the slime tracks influenced
this behavior, indifferent animals were kept for a week (1) with
no change of water, but the slime washed out from the flask
daily, (2) with change of water daily, but the slime not washed
out, and (3) with no change of water and no cleansing of the
flask. Table III gives a summary of results.

TABLE III

Area of Flask

Top

Middle

Bottom

Slime washed out dailv

44

30

26

Unfed and from dark

Water changed daily

29

31

40

No washing or change of water

35

26

29

Since the planarians remained practically as indifferent to
gravity throughout the experiment as they were before it was
begun, the presence or absence of slime tracks probably had
little effect on their geotropism.

The results of these experiments are in line with observations
on the stock animals. They usually remain in the shadow under
the stones and along the side of the dish. The majority rest on
the underside of the stones, but a great many are to be found on
the sides of the dish. Immediately after they finish feeding, they
glide to the top and move about over the dish. If the water is
changed at this time they soon come to rest near the bottom
again. If the water is allowed to get foul after feeding, they
remain at the top, probably in this case on account of lack of
oxygen below. I have been unable to see daily migration such
as Walter ('08) observed.

It would seem reasonable, therefore, to suppose that the
collector who finds planarians ventral surface up on the underside
of rocks, sees those which have been feeding, while if he looked
in other places he might find the unfed ones in any position.

86 J. M. D. OLMSTED

Since Planaria maculata has no otocyst, it may be that after
eating, the food in the digestive tract serves as an otolith, and
after digestion and assimilation the animal becomes indifferent
to gravity because the food is no longer able to press upon the
digestive epithelium. This does not account for the fact that
fed worms are positively geotropic when first put in the dark.

I wish to thank Dr. Parker for suggesting the problem and
for advice as to methods.

CONCLUSIONS

1. Unfed Planaria maculata which have been in the light are
positively geotropic when first placed in the dark. After several
days in the dark they become indifferent to gravity.

2. Fed Planaria maculata which have been kept in the light
are likewise positively geotropic at first. But they become nega-
tive after two days and indifferent after five days.

3. Fed planarians which have been in the dark for some time
are negatively geotropic.

4. The presence or absence of slime tracks has no influence
on the geotropism of these planarians.

REFERENCES

Bardeen, C. R. On the Physiology of the Planaria maculata with Especial Refer-

1901. ence to the Phenomena of Regeneration. Amer. Jour. Physiol.,
vol. 5, pp. 1-55.

Kafka, G. Einfiihrung in die Tierpsychologie auf experimenteller und etholo-

1914. gischer Grundlage. Leipzig, 8vo., xii+593 pp.
Pearl, R. The Movements and Reactions of Fresh-water Planarians: A Study

1902. of Animal Behavior. Quart. Jour. Micr. Sci., vol. 46, pp. 508-714.
Walter, H. E. The Reactions of Planarians to Light. Jour. Exp. Zool., vol. 5,

1908. pp. 35-162.
Whitehouse, R. H. The Natural History of Planarians. Irish Naturalist, vol.
1914. 23, pp. 41-47.

JOURNAL OF ANI MAL BEHAVIOR

Vol. 7 MARCH-APRIL No. 2

THE DELAYED REACTION WITH SOUND AND

LIGHT IN CATS

JOSEPH U. YARBROUGH

From the Psychological Laboratory of the University oj Texas

The experiments herein reported on the delayed reaction in
cats were carried out during the session 1915-16 in the Psycho-
logical Laboratory of the University of Texas under the direc-
tion of Prof. W. S. Hunter. The purpose of the work was:
first, to determine the limits of the period of delay; second, to
ascertain definitely the behavior during delay; and third, to
describe as nearly as possible the method of reaction which
leads to success. Careful records were kept of the behavior
during the period of delay, and particularly of the bodily atti-
tudes maintained and of the orientations. Associations were set
up between movements that led to food and a light or buzzer,
as the case might be, which could be in either of three boxes.
With this association well established, tests were instituted in
which the stimulus was cut off before the reaction was com-
pleted. And throughout the remaining experiments the subject
had to respond in the absence of the stimulus that until now
had been present at the moment of response.

It was my purpose to use in this problem a method of pro-
cedure sufficiently similar to those already used with other
animals,— raccoons, rats, dogs, and children — that by comparison
the relative ranking of the cat in the solution of the problem
could be ascertained.

88 JOSEPH U. YARBROUGH

i

II

1. Cats tested on light. — The four cats used in these tests
were Jim $ , Tom J* , Fay 9 . and Bobby $ . Jim and Bobby
were both about ten months old, vigorous, healthy animals,
and their records may be accepted as typical. The other two
were young cats that had not been properly cared for. They
were weak and died before they were well into the experiments.

2. Cats tested on sound. — Four cats were used in the tests on
sound. Bess 9 and Phil cf were each about two years old.
Judy 9 was about one year and Kitty 9 at least two years
old. Bess and Phil continued strong and did excellent work
throughout the experiments. Judy and Kitty, on the other
hand, died early in the work. From this it is seen that four
cats were at work practically all the time, — two on the light
tests and two on the sound tests.

One would think from the number of deaths reported that
the cats were in poor physical condition. Such, however, was
not the case. Their general health was very good. Those that
died did not experience a long period of sickness, but died within
thirty-six hours of the appearance of distress. There was only
one exception to this, and in this case the cat was replaced by
another rather than risk her recovery.

It was much more difficult than 1 had expected for them
to become physically adjusted to their new environment. They
were kept in a wire cage 12' by 3|' by 6' high, in a room adja-
cent to the experiment room. Their room was well ventilated,
and a large east window provided an inlet to the morning sun-
shine. The difficult thing was to find the most nourishing food
for them. Milk, with a small amount of raw steak, proved to
be the most satisfactory.

Ill
DESCRIPTION OF APPARATUS AND METHODS

In Fig. 1 is shown the ground plan of the box used. The
box was made of \" boards and was 26" high, with the doors
at a, b, c, 10" high by 7" wide. The distance between these
doors was respectively 20", and the distance from the release
door E to each of the doors was 44". The door E of the release
box was raised by a cord passed over a pulley directly above
it and 6|' above the floor of the apparatus. Besides this pulley

THE DELAYED REACTION IN CATS

89

were three other pulleys through which passed cords from the
three sliding doors marked D in the figure. With the aid of
these cords, the experimenter could stand behind the release
box and control the door at each of the boxes. The release box
was covered with wire of \" mesh, and the board B upon which
was fastened the switches for both light and sound. The light
stimulus came from 8 c.p. lamps, so wired that any one of them
could be cut in at a time. The current was obtained from a
110 volt switchboard B.

B

h 's i

Figure 1. — Ground plan of apparatus

In order that the cat might not come in contact with the
lamps, and, also, not be hindered in entering the boxes, a hole
was bored in the back wall of each of the boxes and a lamp
placed outside and behind each box. The holes were of the same
size and 5" from the floor. The lamps were mounted on bases
which rested on the floor, and were placed behind the holes so
that they had equal intensities and could be observed with equal
ease from the release box. One 8 c.p. lamp hung over the
center of the apparatus and 4' from the floor throughout the
experiment. This light was shaded with a paper bag which
made it necessary to keep fresh sawdust on the floor of the box
to make the movements of the animals clearly visible. At the

90

JOSEPH U. YARBROUGH

outset I was compelled to cover the entire apparatus, as the
cats were free to jump out at will. The wire used for this pur-
pose was of \" mesh and its tendency to blurr the field of vision
made it still more necessary that the white sawdust be used.

Fig. 2 should give a clear presentation of the essentials of the
box when taken in connection with Fig. 1.

Figure 2

The cats on sound used the same apparatus as those on light,
the only difference being the change in stimulus. On the switch-
board B, Fig. 1, were placed three buttons which corresponded
to each of the three light boxes, a, b, and c. In each of these
boxes a buzzer was suspended directly over the door and 12"
from the floor of the apparatus. These buzzers were suspended
by a coiled wire, and were not in contact with the apparatus.
The system of wiring was the same as that of lighting — i.e., any
buzzer could be sounded at the wish of the experimenter by
pushing the proper button on the switchboard B. Such an
arrangement made it possible for the experiments on both sound
and light to be carried on without any interference. So far as
the knowledge of the experimenter goes, the cats on light never
found the buzzers, nor did the cats on sound find the lamps.

The general method of experimentation was as follows: The
animal to be tested was put in the release box which is

THE DELAYED REACTION IN CATS 91

shown in Fig. 1. Now, suppose, for example, that the
lighted box were the one on the left, c; its exit door would be
opened and its light turned on. When the experimenter was
sure that the cat had seen the light or heard the sound, the
animal was released. A careful, detailed record was kept of
the direction in which the animal was oriented at the moment
of release and its path to the exit. Any unusually wide turn
in the path was always recorded. Hesitation and zig-zag move-
ments were especially noted whenever and wherever they ap-
peared in the cat's response. In these experiments the cats
should go straight to the lighted box, and through its exit door
and back to the entrance of the release box where they got food.
With the cats on the sound problem, the reactions were the same.
With them, however, the " lighted " box was a " sound ' box.

When the cats were sufficiently trained to choose the stimulus
box (lighted or sounded, as the case may be) almost perfectly,
delays were begun. The periods of delay were much the same
as those used by Hunter. 1 The first delay was to turn the
stimulus off just as the animal reached the box. In the second
delay, the stimulus was cut off when the animal was half way
to the box. In the third delay, the stimulus was stopped just
as the animal made its first move in response after the door of
the release box was raised. And, in the fourth delay, the stim-
ulus was cut off just before the door of the release box was
raised. In this fourth stage a genuine delay first enters in. The
first three stages of delay were of little or no value as delays.
Their primary purpose w r as to bridge over the period of stim-
ulus to non-stimulus, to bridge that period between acting in
the presence of a stimulus and acting in the absence of a stimu-
lus. All that was necessary to make a correct response was, in
each case, for the cat to continue in the direction he was going.
There was no further choice to be made. The fourth delay,
however, was genuine, although of small duration. The stimulus,
was cut off before the cats were released. Throughout the re-
mainder of the experiments the cats were compelled to react in
the absence of the stimulus that until now had been present at
the moment of response.

There was no definite standard adopted by which to promote

1 Hunter, Walter S. The delayed reaction in animals and children. Behav.
Mon., vol. 2, no. 1, 1913.

92 JOSEPH U. YARBROUGH

from one period of delay to another. The general method used,
however, was to promote the animal as fast as possible, and only
demote when the records showed him unable to bridge the delay.
There were no special arrangements made for punishment in case
of error. It was easy to observe, however, that there was a
certain degree of punishment following each error. These pun-
ishments were: having to back out of a box, and having food
and freedom deferred for a longer period of time. Although the
cats were expected to go straight to the stimulus box, no wide
turn in their pathway is recorded wrong unless they approached
the entrance to one of the other boxes. The apparatus was
so constructed that the animal could not see the position of
an exit door, i.e., whether it was open or closed, without actually
approaching the particular box.

IV

EXPERIMENTAL RESULTS
1. Three Compartment Experiments

A. Learning the association. — Although the primary purpose
of this investigation is a study of the delayed reaction proper,
it is well to make additional note of the learning process. Table
I gives the number of trials required by the cats of set A to learn
the association between the light and the getting of food. Each
cat was given 10 trials daily. Fay, the last reported in the table,
died at the end of 50 trials. Her results are reported, however,
because 75% of her last 20 trials were successful.

TABLE I

Cats Tested on Light

Number Per cent

Number Number Per cent correct of correct of

Cat of trials correct correct last 50 last 50

Bobby 130 96 73 47 94

Jim 110 84 76 49 99

Tom 170 112 65 45 90

Fay 50 25 50 25 50

The number of trials required by the cats on sound, set B,
are given in table II. The last cat reported in this table died
at the end of 40 trials. For this reason no record of her work
appears in the last two columns of the table.

THE DELAYED REACTION IN CATS 93

TABLE II

Cats Tested on Sound

Number Per cent

Number Number Per cent correct of correct of
Cat of trials correct correct last 50 last 50

Bess 180 123 68 43 86

Phil 70 44 63 37 74

Kitty 110 68 61 32 64

Judy 40 26 65

These results indicate, first, that the cats of each set learned
the association readily. The learning curve would appear short
and steep. And, second, they indicate that it is more difficult
to maintain a high efficiency in set B than in set A. This is
indicated by the fact that, while Bess and Phil, both of set B,
made 86% and 74% respectively correct in the last 50 trials,
Bobby and J m, of set A, made 94% and 99%.

Although the differences of results, as given in the tables
above, are not conclusive, the experimenter is of the opinion
that the sound tests present the more difficult problem. These
variations may well be explained on the basis of individual
differences, but it is to be noted that the animals tested on light
have the better records. This probable increase in difficulty in
the sound tests is due, no doubt, to the timidity on the part
of the cats when approaching the sound. The records show, as
is indicated in the next paragraph, that the cats were for some
time rather frightened by the sound of the buzzers. This caused
an increase in the number of errors and so a decrease in the
percentage of correct reactions. Before a definite conclusion can
be reached a sufficient number of cats, to eliminate errors from
individual variations, must be tested.

Observations of the behavior of the animals during the learn-
ing period on sound, which were recorded from day to day,
suggest several smaller divisions. (1) A period of disregard.
My notes read, " Bess appears to give no attention to the buz-
zer," and, again, the next day, " Bess walks about freely with-
out noticing the buzzer." (2) A period of disturbance. This
period may be characterized by a behavior whic i may be termed
" awareness " or " worry." The cat stops, turns head, looks,
and calls as if in danger. This note is recorded, " Bess dislikes
to go to the sound. She appears shy and afraid of the buzzer.
She will venture to the door, stop, and squat; look up at the buz-

94

JOSEPH U. YARBROUGH

zer and sometimes rise up and ' sniff ' at it before going into the
box." (3) A period of hesitation. The behavior of this period
is characterized by wavering and by starts and stops. And, in
period (4), the cat gives strict attention to the stimuli. Here
the behavior becomes more nearly perfect, the path of reaction
has been made straight, and the percentage of correct reaction
is high. With the animals tested on light, set A, the same
learning period divisions could be made. In this case, however,
the period of disturbance was not accompanied by so much
timidity and fear.

During this period of experimentation all possible care was
taken to prevent any preference for particular boxes. Should
such a tendency be observed, control tests were given to break
it up before the position habit was well developed. At the end
of the first 60 trials each box had been presented 20 times, and
the records show that no box was chosen more than 26, nor
less than 16 times by any one of the eight subjects.

For comparison we bring together in table III data on learn-
ing the association obtained by Hunter in his "study of animals
and place beside it that of our own subjects.

It is of interest to note that all the cats fall in the class with
Bob, Hunter's most rapid raccoon. Bob learned the association
in 120 trials while the eight cats used in these tests ranged from
50 to 180 trials with an average of 107 trials each. The curve
representing the learning period for the discrimination of the
three compartments was very short and steep, yet broken and
irregular. With continued practice, this irregularity would un-
doubtedly have been eliminated; and the cats of each set would
have attained perfect mastery of their problem.

TABLE

III

Number of

Number of

trials on

trials on

learning

learning

Raccoons —

Rat

Bob

120

No. 9

280

Betty

340

No. 12

440

Jack

540

No. 13

250

Jill

825

No. 15

No. 16

220

480

Dogs —

Blackie

560

Brownie

650

THE DELAYED REACTION IN CATS

95

TABLE III— Continued

Rats-
No. 2.
No. 4.
No. 5.
No. 6.
No. 7.

Number of
trials on
learning

176
175
505
800
361

Cats-

Bobby.
Jim . . .
Tom. .
Fay . . .
Bess . . .
Phil . . .
Kitty . .

Number of
trials on
learning

130
110
170

50
180

70
110

(c) Controls used. — In the construction of the apparatus, every
effort was made to eliminate all possible differences in the com-
partments which could be used as guides to correct reactions.
The backgrounds surrounding the entrances to the compart-
ments were all alike painted black. Since the backgrounds were
all of the same brightness, and, since everything remained con-
stant with the single exception of the exit doors to the com-
partments, controls were put in to determine their possible effect.
In order to test this, the three doors were all opened and the
tests were given by the usual method under conditions in all
other respects normal. The results were entirely negative. In
no case did an animal make use of the doors as cues to its reactions.

Again, control tests were introduced to determine whether or
not the animals were really depending upon the applied stimuli
(light or sound) for cues for guiding their reactions. To test
this, experiments were made under normal conditions except
that each time the stimulus (sound or light) was withheld .30%
correct reactions was the highest made by any subject under
these conditions. It is clear, therefore, that normally the reac-
tions were made either to sound or to light.

Not being able to secure the same pitch and intensity in each
of the three buzzers, control tests were made to determine whether
the animals had formed associations between them on the basis
of quality. The buzzers were all interchanged — buzzer a took
the place of b, b the place of c, and c the place of a. No case
was found where the differences in pitc'i and intensity were used
as cues for reaction. These qualitative differences could well
have been effective during the period of learning the association ;
but, on the delayed experiments, they could be of little or no
value. The essential cues in handling delays must be factors

96 JOSEPH U. YARBROUGH

that are variable from trial to trial otherwise they cannot be
selective in nature.

No temperature controls were used. They were thought to
be unnecessary because of the following: 1. The lights were
turned on but for a short time. 2. They were outside of the
main apparatus. 3. The cats oriented immediately when the
lights were turned on and reacted precipitately when released.
And, 4, the behavior of the cats on light was the same as that
of those on sound where temperature could not be involved.

B. "Delayed" experiments. — Since in the first four delays used
the entire reaction was not performed after the stimulus had been
removed, it is probable that they should not be termed delays
at all. The stimulus was always continued until the experi-
menter was convinced from all external evidence that the cat
had become aware of its presence.

The cats tested on sound and those on light were all pre-
sented their problems by the method described above, but for
convenience the data will be discussed separately.

(a) Set A (cats tested on light). —

Delay I. — In delay I the light was turned off just as the cat
reached the correct compartment. Bobby was given 30, Jim,
20 trials; and for both of them each trial was successful. With
the association well established, the turning off of the stimulus
at this point in the reaction effects no change in their percentage
of correct response.

Delay II. — In this delay the stimulus was cut off when the
cat was half way from the release box to the correct compart-
ment. 2 Jim was given 10 trials with all of them correct. Bobby
was given 60 trials with 56 correct. There appears to be no
difficulty in making the step from delay I to delay II, even
though the cats here made one-half of the distance of response
in the absence of the stimulus. After the cat is well set out, then,
on his reaction, the stimulus may be withdrawn without affect-
ing the response.

Delay III. — The only difference in this delay and number II

is that here the stimulus is withdrawn before the cat is well on

2 In case the cat started from the release box in a different direction from that
of the stimulus, e.g., if he started toward c when the stimulus was at a, the stimulus
was not turned off until he did turn in the direction of the stimulus compartment,
and so in this case, was well on his way.

THE DELAYED REACTION IN CATS 97

his way, while in II the reaction was half completed. Jim was
given 20 trials all of which were correct. Bobby was given
60 trials with 57 correct. The reader will notice that the cats
have still met no difficulty.

Delay IV. — Bobby was given 80 trials of which 66, or 82%,
were correct. Jim received 130 trials, 107 of which were cor-
rect, making also 82%. Here the first difficulty of bridging
over a period of delay appears. The door of the release box and
the cutting off of the stimulus were operated simultaneously and
without reference to where the cat was or what it was doing.
Thus the animal was forced to initiate the reaction and perhaps
make a choice of compartments, in the absence of the stimulus.
The data show that Jim took much longer to master this delay
than did Bobby, although he had held a higher percentage on
fewer trials in the three preceding delays. The fact that Jim
had received 100 trials less than Bobby in these preceding
delays can be offered as explanation of his need of 60 more trials
here. It seems natural that had he not been advanced so rapidly
from one delay to another he would have been better prepared
for this new delay, and, being better prepared, would have
bridged over it much more quickly.

Two seconds delay. — At this point in the experiments a metro-
nome was placed in an adjacent room to mark the period of delay
in seconds. At this distance its sounds could be easily heard
by the experimenter, yet they were not thought to be strong
enough to distract the attention of the animals.

Bobby was given 130 trials with 106, or 81% correct; while
Jim was given 200 trials, 143, or 71% of which were correct.
Of the last 40% of Bobbie's trials, 34, or 85% were correct; of
the last 40% of Jim's trials, 32, or 80% were correct. The
data do not show that the reactions were poor at the beginning
of the delay and grew better with successive trials, but rather
show an irregularity throughout. Bobby, e.g., was perfect on
the first 10 trials, while after having received 70 trials she made
only 50% on 10 trials. Again, Jim, after 160 trials, made only
30% on 10 trials, yet on the 10 just preceding, he made 90%
and 80% on the 10 immediately following.

Four seconds delay. — In the four seconds delay experiment,
180 trials were given Bobby with 141 correct, and 150 given
Jim with 118 correct. Each made 78% correct. Jim's last 30

98

JOSEPH U. YARBROUGH

trials showed much improvement, 29 of them being correct
reactions. Although Bobby had 30 more trials on this delay
than Jim, she made only 26 out of her last 30 trials. This is
readily explained in the light of the fact that the middle com-
partment was dropped out during this period with Jim, while
Bobby continued on three compartments. Jim had made 70%
on the last 40 trials preceding the 30 trials mentioned above, of
which he got 29 correct. At the end of these 40 trials, the middle
compartment was dropped out. Of the first 10 trials with only
two compartments, Jim made 100% correct. The records show
that Jim would have made a higher percentage than 70 much
sooner had it not been for a tendency to drop out the middle
box. This is not only seen in table IV, but by the fact that
when he received the stimulus only from boxes "a" and "c"
(the experimenter having dropped out the middle box), he made
100% on the first 10 trials.

TABLE IV
Daily Record on 4 Seconds Delay with Light

Number

Number

Distribution of

errors

Cat

of trials

correct

a

b

c

10

6

1

2

1

10

9

1

10

10

10

7

2

1

10

7

2

1

10

8

2

10

8

2

10

7

1

1

1

10

8

2

10

8

1

1

10

8

2

10

7

2

10

7

1

1

10

8

2

10

8

1

10

9

.

10

9

1

10

9

11

19

7

10

9

1

10

8

2

10

7

2

1

10

6

3

1

10

7

3

10

7

2

1

10

8

2

THE DELAYED REACTION IN CATS 99

* TABLE IN— Continued

Cat

Jim.

Number

Number

Distribution of errors

of trials

correct

a b c

10

9

1

10

7

3

10

7

3

10

7

3

10

*7

3

10

10

10

9

1

10

10

1 28 3

Six seconds delay. — Bobby was the only cat either on light
or sound that was tested on the six seconds delay before the
middle box was taken out. Although she had made 85% on
her last 40 trials on the four seconds delay, she fell to 50% on
the first 10 trials in the six seconds delay. After 90 trials with
only 50% of the last 40 correct, she was put back on the four
seconds delay where she was given 70 trials, making 80% on
the last 50. She was again given 20 trials on the six seconds
delay with 55% correct. After 60 more trials on the four seconds
delay with 85% correct, she was given 20 trials on the six seconds
delay with 70% correct. The records show that during the six
seconds delay she became restless and often turned around in
the release box. In such cases she usually went to boxes a or c,
depending upon the one she came in line with first in making a
circle by her turning in the release box. It is the writer's opinion
that with further training cats can bridge the six seconds delay
with three boxes.

(b) Set B (cats tested on sound). —

Delay I. — The two cats that continued the work on sound
after the death of their fellows were Bess and Phil. Bess was
given 60 trials, 48 of which were correct. Of the last 30 trials,
26 or 86% were correct. Phil received 30 trials with 25 or 83%
correct, and 9 of the last 10 correct. As in the case of the light
experiments, no difficulty was encountered by cutting off the
stimulus at this point in the reaction.

Delay II. — One hundred trials were given Bess, and she made
80% correct reactions. Phil received 50 trials which contained
90% corre ct. Of the last 20 trials 19 were correct. Although

* Middle box was dropped out here.

100 JOSEPH U. YARBROUGH

the cats were making the last half of the response in the absence
of the stimulus, no difficulty yet appeared.

Delay III. — On this type of delay, where the stimulus was
cut off the moment the cat left the release box, Bess was given
100 trials. Of this number 84 were correct, with 54 of the last
CO correct. Phil made 56 correct reactions out of 70 trials,
making a percentage of 80. These results mean that, having
started correctly, the cats are able to retain their cue for cor-
rect reaction even though the remainder of the reaction must
be made in the absence of the stimulus.

Delay IV. — In this delay, Bess was given 60 trials, 55 of which
were correct, and, of the last 40 trials, 39 were correct. Phil
was given 60 trials with only 60% correct. Of the last 30 pre-
sentations, 18 were reacted to correctly. Table V shows Phil's
tendency to drop out compartment b whenever the delays set in.

TABLE V
Daily Record on Delay 4 With Sound

Number Number Distribution of errors

Cat

of trials

correct

a

b

c

Bess

10

10
6

2

2

10

10

10

10

9

1

10

10

10

10

3

3

Phil

10

7

2

1

10

7

3

10

6

1

2

1

10

4

4

2

10

7

2

1

10

7

3

*10

8

2

10

9

1

10

8

1

1

10

7

3

2 22 5

Two seconds delay. — On this two seconds delay, Bess was

given 130 trials and of this number 116 or 89% were correct.

Of the last 80 trials, 74 were correct; with the percentage of 92,

she was promoted to the four seconds delay. Phil was given

* From January 23rd to February 3rd, Phil was being retrained on delays II and
III, receiving 10 trials each day.

THE DELAYED REACTION IN CATS 101

60 trials with 51 or 85% correct. Of his last 40 trials, 35 were
correct.

Four seconds delay. — One hundred and seventy trials were
given Bess with 121 correct responses. Of the last 30, only 18
were correct. With this low record, it was thought best to
return to shorter delays before trying to advance. From
January 24th, 1916, until February 14th, she was given 10 trials
daily on delay IV and on two seconds delays. Of the 200 trials
given during this period 120 were given on the two seconds
delay, the last 30 of which netted 28 correct reactions. This
high percentage of correct reaction on the last 30 is due to the
dropping out of the middle box ; so, also, may the low percentage
of correct reaction immediately preceding be explained by the
tendency to drop out the middle boxes the delays were length-
ened. The percentage of correct responses in the last 30 trials
immediately preceding the dropping out of the middle box was
66, while the percentage of the first 30 after its being dropped
out was 94.

(c) Maximal interval of delay with three boxes. —

In table VI the maximal delays attained by my cats are
given, and for comparative purposes similar data on Hunter's
animals and Walton's dogs are included. The reader should
remember that these tests were made with a choice of three
boxes and that training stopped here because of a well developed
tendency to drop out the middle box. In the case of Phil, the
last cat reported in the table, this tendency was not well devel-
oped. As is shown in the table, he was making a good record
on the two seconds delay, and there are no indications that
he could not have bridged a longer period of delay with three

boxes.

TABLE VI

Maximal

Number

Per cent

Animal

delay

of trials

correct

, ..Rats-

No. 13

4 sees.

. . .

88

No. 15

1 sec.

86

No. 16

1 sec.

50

No. 17

7 sees.

68

Dogs —

Blackie

5 mins.

80

Brownie

2 sees.

68

102 JOSEPH U. YARBROUGH

TABLE VI — Continued

Maximal Number Per cent

Animal delay of trials correct
Raccoons — ■

Jill 3 sees. . . . 93

Jack 20 sees. ... 85

Bob 25 sees. ... 90

Walton Dogs 10 sees. ... 64

Present work . . Cats —

Set A (light) Bobbv 4 sees. 160 85

Jim... 4 sees. 120 78

Set B (sound) Bess . 4 sees. 170 71

Phil . 2 sees. 40 83

It will be noted that the longest delay mastered by the cats
was a period of four seconds. I am sure that with continued
training they can bridge a much longer period than this. But,
since I was more interested in the behavior during delay than
in the maximum period of delay; and since at this point there
had developed a tendency to drop out the middle box; and, again,
since time was limited, I thought it best not to give further
training on three boxes.

(d) Longer delays. — ■

Scattered throughout the experiments are correct reactions
over periods of delay much longer than those mastered in the
regular series. These periods were willingly lengthened by the
subjects themselves. This in itself is good evidence that with
sufficient training a much longer interval of delay could be
mastered. At three different times Bess made 9 correct reac-
tions out of 10 trials with a delay period of six seconds. And,
at another time she made, with the same interval of delay, 17
correct responses out of 20 trials. Twice she responded cor-
rectly after a delay period of twenty-six seconds. It will be
recalled that Bess was tested on sound. Jim, also tested on
sound, bridged at one time a period of eight seconds, at another
a period of eighteen seconds, and a third of thirty-four seconds.
The cats on light seemed not to have hesitated so often as did
those on sound. In all the work on the three box experiments,
Bobby was the only cat tested on light who voluntarily length-
ened her period of delay. On this occasion she sat for sixty-six
seconds in the release box, after which she went directly to the
proper compartment. All the periods of hesitation were not
measured and tabulated. Animals, both of Set A and Set B,

THE DELAYED REACTION IN CATS 103

hesitated often from one to three seconds on a single reaction,
but their occurrence was so irregular and their duration so
brief that their measurement and tabulation were very difficult.
Therefore, no period was recorded in seconds unless it was of
considerable duration. However, all hesitations were entered
in the notes.

2. Two Compartment Experiments

A. "Delayed" experiments. — The delay work was continued
in the two compartment tests by the usual method. The series
of presentations of boxes was changed from

ab

cc

ba

ba

cb

bb

ca

ca

be

ca

ba

ca

bb

ca

ac

One of these three series of ten had been used each night. Each
one was used an equal number of times and at no time was
"one given twice in succession. In this way no one series was
given twice within three days. On the two compartment tests
the number of series was increased to four, as follows:

ac

ca

ac

ca

ca

ca

ac

ca

ca

ac

aa

ca

ca

ac

ac

cc

aa

ca

ac

ca

These were taken in their order beginning with the first, and
no one was, therefore, given twice within four days.

(a) Cats tested on light. — It will be remembered that in table
IV Jim is reported to have made 29 out of the last 30 trials
correct, after a four seconds delay. As his work progresses on
the two compartment experiments, the period of delay increases.
Since a very large proportion of his errors on the three com-
partment experiments were due to a tendency to drop out the
middle box, he would be expected to make a higher percentage
of correct reaction with this box omitted. Such is shown to
be the case in the data below.

Of the 40 trials given Jim on six seconds delay 34 were cor-
rect. He escaped from the laboratory on the third day, after
his work, and after 36 hours absence was recovered and made
80% on 10 trials. Feeling sure that the cat was experiencing
no difficulty, the experimenter increased the period of delay to

104 JOSEPH U. YARBROUGH

eight seconds. Jim was given 30 trials with this period of
delay 27 of which were correct. As he appeared to meet no
difficulty in bridging this period, he was set to work on ten seconds
delay where he reacted 28 times correctly in 30 trials. On
twelve seconds delay he made 90% on 30 trials. Since he had
so successfully bridged over these small advances in delays, the
next increase was double in length, i.e., four seconds. Forty
trials were given with a delay of sixteen seconds. The problem
did not seem to increase in difficulty for 36 of these 40 presenta-
tions were reacted to correctly.

The longest period of delay in which a regular series of experi-
ments were offered was eighteen seconds. One hundred tests
were given Jim on this period of delay, and of this number he
responded correctly to 91. During these experiments, Jim was
observed as closely as possible as to the orientation of head and
body when the door of the release box went up, and also at the
moment he initiated the movement of response. The matter
of orientation will be taken up again under the discussion of
" behavior during delay."

(b) Cats tested on sound. — Bess and Phil had been dropped
back to the two seconds delay before the middle box was dropped
out. Beginning with the two seconds delay they were promoted
simultaneously from one interval of delay to another.

Figured on the basis of 30 trials, Bess' percentage jumped
from 66 to 95, and Phil's from 80 to 96. On the four seconds
delay no difficulty was met. After 40 trials, — Bess with 93%
and Phil with 99, — they were promoted to the six seconds delay.
Here they received 40 trials, Bess making 95%, while Phil made
only 77%. This low percentage on the part of Phil was caused
by a pronounced position habit which appeared on the first day
and lasted through the second day of the series. They each
received 30 trials on both the eight and the ten seconds delays,
and each held a percentage of about 85. Since these periods
were bridged so easily, the period of delay was now lengthened
to fourteen seconds. On this interval of delay, 40 trials were
given, and Bess held 87%, while Phil made 98%. Ninety trials
were made by each cat on the sixteen seconds delay. Of these
90 trials, Bess was successful 84 times, and Phil 81 times. Dur-
ing this last period of delay of 90 trials, special observation was
made of orientation. These observations were recorded in detail,

THE DELAYED REACTION IN CATS 105

and will be carefully considered under " behavior during delay
and after release."

(c) Maximal interval of delay attained with two boxes. — Table

TABLE

VII

Maximal

Number

Per cent

Animal

delay

of trials

correct

Hunter

Rats-

No. 4

1 sec.

20

75

No. 11

5 sees.

70

81

No. 15

5 sees.

60

67

No. 16

5 sees.

50

90

Dog-

Blackie

3 mins.

30

86

Raccoons-

Jack . . .

20 sees.

40

85

Betty

10 sees.

30

86

Bob

25 sees.

20

90

Walton

1 min.

10

80

Present work

. .Cats — •

Set A-

-Jim

18 sees.

90

90

SetB-

-Bess

16 sees.

90

93

Phil

16 sees.

90

90

VII gives the maximal delay attained on two boxes by the
subjects studied by Hunter, Walton's dogs, and the cats of the
present experiments. The cats rank very well with Hunter's
raccoons in successfully bridging delays with two boxes. Just
what interval of delay could finally be bridged with the two box
tests is not known. It is evident from the above table that the
limit of the cats' ability was not reached. I see no reason why
the interval may not be increased even into minutes.

This opinion is based upon the fact that the records show
many reactions where the period of delay is of much longer
duration than eighteen seconds, the greatest recorded in the
above table. The following long periods of delay were each
followed by successful reaction. Phil lengthened his delay
period- twice during this period of 90 trials, once to twenty seconds
and once to twenty-two seconds. Jim reacted correctly after
three such periods of delay, twenty-four seconds, twenty-six
seconds, and thirty seconds. Bess was successful after the fol-
lowing delays : 1 twenty seconds duration, 3 twenty-two seconds,
1 twenty-four, 1 twenty-six, 1 thirty-two, 2 thirty-six, 1 forty-
two, and 1 fifty-two seconds duration.

It will be noted that all animals delayed much longer with

106 JOSEPH U. YARBROUGH

two than with three boxes. This is readily explained on the
basis of the relative complexity of the problems, and the effect
of continued training.

3. Behavior During Delay and After Release

Four different types of behavior appeared in our experiments:
(1) The animal maintained an orientation of all its body during
the interval of delay, i.e., it kept both its head and body point-
ing toward the proper box; (2) the animal kept either its head
or its body in perfect orientation; (3) no observable part of the
animal's body was retained in constant position, i.e., the experi-
menter could detect no orientation cues used by the animal,
(4) the animal held some certain position in the box, i.e., it
actually went to the point in the release box nearest to the
proper compartment and there awaited to be released. Types 1,
2, and 3 will be combined for convenience in the discussion, and
will be followed by a consideration of 4.

A. Orientation of whole or part of body. — Great pains were

taken to insure accuracy in the recording of orientations. Records

were kept not only of the body position, but of whether any

observable part of the animal remained in a constant position

during the delay period. Also note was made of any case where

the animal turned partly or entirely around, as well as of the

direction in which it turned. In order to obtain as accurate

data as possible on orientation, a series of 300 tests were specially

given where the orientations of both the head and body were

recorded at the moment the door of the release box went up, and

again when the animal made its first motion to leave the box. Tables

VIII and IX give a summary of these reactions showing just

what orientations preceded them. In the first table only those

tests are recorded where the orientation was different when the

animal started from what it was when the door went up. While

in the second table all tests are recorded where the orientation

was the same when the animal started as it was when the release

door went up.

TABLE VIII
When door went up: Correct Wrong

Good orientation of head only 108 1

Good orientation of body only 30 1

Good orientation of head and body 1 18

Poor orientation of head and body 18 24

THE DELAYED REACTION IN CATS 107

TABLE Will— Continued

When animal started : Correct Wrong

Good orientation of head only 2

Good orientation of body only 1

Good orientation of body and head 259 3

Poor orientation of body and head 9 26

TABLE IX

Good Bad

Good orientation lost between release and starting 107 1

Good orientation not lost between release and starting .40

Poor orientation at release and at starting 4 12

This table indicates plainly the similarity of the behavior of
my cats and Hunter's rats and dogs. The cats almost never
reacted in opposition to their orientation. (Here I mean, of
course, the orientation of both head and body, for many times
they reacted correctly in accordance with only the head or the
body.) Of 141 errors made by one of Hunter's dogs, 116 were
preceeded by faulty orientation. So, also, the cats' errors, as
the table shows, were in almost every case preceded by faulty
orientation. The non-orientation reactions are few enough to
be accounted for by chance.

B. Position in the box. — Owing to the fact that during the
period of long delays only two boxes were used and they were
located far apart, the cats could have shifted their behavior
from the use of orientation cues to the use of position cues. In-
formation on this point was hard to get: (1) Because of the
continuous movements of the animals, and (2) because the size
of the release box in comparison with the distance to the exit
box is so small that but little is gained by being at one side
or the other. However, from the few observations made, the
writer is of the opinion that the position of the animal in the
release box does aid its reaction.

4. Reaction Tendencies
In order to get representative data on errors and position
habits and the frequency with which these stereotyped forms of
response interfered with the work, I shall present 610 of Jim's
and 630 of Bess' reactions. It will be remembered that Jim
was tested on light and Bess on sound. The first 320 of Jim's
610 reactions were made on the three box experiments, while
the remaining 290 were made with only two. Position habits
in which one particular box was always chosen first, occurred

108 JOSEPH U. YARBROUGH

with each animal on each of the three boxes. Now one box
was chosen first and now another. For convenience these data
will be recorded in two separate tables (X and XI), the first
containing data recorded on the three box experiments, and the
second, those obtained when only two boxes were used.

TABLE X

Three Box Experiments Total

reactions

Order of response abc acb ab ac made

Jim 7 3 22 1 33

Bess 7 1 22 4 34

Order of response bac bca ba be

Jim 10 1 2

Bess 2 7 3 12

Order of response cab cba ca cb

Jim 8 6 3 23 40

Bess 3 4 15 22

Table X analyzes all incorrect responses made on the three
box experiments, here included, and gives the relative number
of times each subject followed the different possible orders. The
number of errors made beginning with boxes a and c is about
equal with both animals, while the number beginning with b
is very low. In fact, as the table shows, Jim only made 2 errors
when b was selected first. Bess, however, made 12 such errors,
or about one-half the number she made after selecting c first.
Of the 40 times Jim selected c first, he selected b next 29, or
72% of the time. And of the 33 times he selected a first, 29,
or 87% of the time b was the next box selected. When a was
selected first by Bess, she chose b next 29 times out of 34, or
85% of the time. And when she selected c first, she chose b
next 18 times out of 22, or 81% of the time. It may be con-
cluded then, that whenever the reaction began at the end of
the row of boxes, i.e., a or c, the tendency was to take the boxes
in order until the solution was reached. Only three times in
320 trials did Jim go to the same box twice in the same trial.
(These are listed in table XII as ' persistent errors.") The
form of this position habit was c b c b a, and was repeated three
times within 20 trials. Bess returned to the same box in the
same trial only one time, and the order of the boxes chosen
was babe. One further thing to be noted is that Jim made
25 " three place errors," responses where the animal tests each

THE DELAYED REACTION IN CATS

109

of the three boxes before the solution is reached. This type of
error was made 13 times by Bess. This form of behavior is
apparently less frequent than in Hunter's child 1 and far less
frequent than in Hamilton's dog. 4

TABLE XI

Two Box Experiments* Total

reactions

Order of response abc ac bac ba made

Jim 14 14

Bess 6 14 20

Order of response bca be cba ca

Jim 15 15

Bess 6 6

* Since exit b is no longer open, all orders of choice ending in b are omitted in
this table.

Table XI contains a record of the errors made in the two
box experiments. Jim had so completely lost the cue to b that
not one time after that box was dropped out did he return to
it. Although Bess never made b her first choice again, she
at six different times on her way from a over to c stopped by and
examined b. It will be noticed that the number of errors greatly
decreased with the elimination of the middle box. This may be
accounted for, first, by the increase in the simplicity of the
problem and, second, by practice.

It would be well worth while to put beside the reaction ten-
dencies of these two cats similar data gathered by Hunter on
rats, raccoons, dogs, and children. Table XII gives a sum-
mary of the errors made by his subjects, and includes, also,,
those for my two cats. Some explanation of this table is

necessary.

TABLE XII

Total Three Persis- Per Per

Number number place tent cent cent

of of errors errors errors of of

Animal trials A B C AtoBBtoC

Child 264 120 54 6 44 11

Raccoon— Bob 720 209 78 29 32 37

Dog— Blackie 570 127 75 25 59 33

Rat— No. 9 575 144 42 13 29 30

No. 2 345 152 69 47 45 68

Cat— Jim 320 75 25 3 33 12

Bess 330 68 13 1 19 7

3 Hunter, W. S. The delayed reaction in a child. Psych. Rev., 1917, vol. 24, 74-87.

4 Hamilton, G. V. T. An experimental study of an unusual type of reaction in
a dog. Jour. Comp. Neitr. Psy., 1907, 17, 329-341.

110 JOSEPH IT. YARBROUGH

The raccoon's records include delays from one second through
twenty seconds; those for the dog, from one second through
seven seconds; those for rat No. 9, from the third stage of delay
(turning light off just as the animal was released) through seven
seconds; those for rat No. 2, from the third stage of the delay
through one second ; and those for Jim and Bess, from one second
through four seconds. The data are compiled here for compara-
tive purposes and will be easily interpreted without further
comment.

Available data at the time of the preparation of Hunter's
paper on the delayed reaction in a child made it clear that there
were no marked differences between animals in the reaction
tendencies displayed under the experimental conditions in ques-
tion. It did look, however, as though there were marked dif-
ferences between the animals and the child. Our data here
presented place the cats in a class with the child. So far then
as this type of test is concerned, no essential differences between
man and other animals have been brought to light.

CONCLUSIONS

1. All the cats herein tested learned the initial association
within from 100 to 180 trials and therefore fall into a class with
Hunter's raccoons, so far as rapidity of learning is concerned.

2. No differences of method in solution of delays was observed
between cats on light and those on sound.

3. The minimum and maximum delays were two seconds and
four seconds on the three compartment experiments; while with
only two compartments, they increased to sixteen seconds and
eighteen seconds respectively.

4. The cats solved the problem by maintaining gross motor
attitudes of the whole or part of the body.

TEMPERAMENTAL DIFFERENCES BETWEEN

OUTBRED AND INBRED STRAINS OF

THE ALBINO RAT

NENOZO UTSURIKAWA

From Hie Harvard Psychological Laboratory

INTRODUCTION

About two years ago the writer sought, in the Harvard Psy-
chological Laboratory, training in the methods of comparative
psychology, since such training promised to be helpful to him
as an ethnologist. A problem was suggested to him by Pro-
fessor R. M. Yerkes, — evidently difficult and yet extremely
fascinating. Its thorough study would certainly require years
of diligent work. But the writer, because of his ethnological
interests, was able to give only one year to this psychological
investigation.

Obviously enough, from what follows, the ' materials to be
presented are fragmentary and inadequate for the description
of the differences in the strains of rat. Still, to throw them
away would seem too extravagant. With a humble sense of
obligation, the writer offers his limited data to the scientific
world. He wishes to take this opportunity to thank Professor
Yerkes, Dr. R. M. Elliott, and Dr. W. R. Miles, for valuable
assistance in the work.

PROBLEMS

The chief problem was to discover, if any, the temperamental
differences between outbred and inbred strains of the albino
rat. Such features of behavior as degree of nervousness or
timidity, of savageness and wildness, of sensitiveness to stimuli,
of persistence in response, quickness of response, and so on, may
be recorded as constituting the temperament of an animal. In
the terms of psychology, and in the last analysis, perhaps, tem-
perament is identical with the threshhold, quickness, amount,
and steadiness of response to a given stimulus or object. The

112 NENOZO UTSURIKAWA

problem, therefore, requires the measurement of the essential
components of temperament in order that comparisons of the
two strains may be made.

Although the inquiry was directed mainly to temperamental
characteristics, differences in behavior of other sorts were noted
and may here be reported. From the anthropologist's point
of view, the study of close inbreeding and its consequences,
even in case of lower animals, is of extreme interest and of some
practical importance. Anthropological data concerning this mat-
ter are meager, and as Topinard remarks, " the question is still
sub judice." Possibly it is not far from the truth to say that
such information concerning man will long be lacking, whereas,
through the study of infra-human organisms, we can readily
approach reliable information. Infra-human psychology gains
in importance as it allies itself with human psychology, and
this paper, if not projected upon the background of larger human
interests, will lose much of its significance.

There is such meager literature on temperamental character-
istics of lower animals that a historical summary is unnecessary.
The contribution of Basset 1 to the study of albino rats alone
has fairly direct bearing upon the materials of this paper.

SUBJECTS

Only albino rats were observed. All were obtained either
from the Wistar Institute of Anatomy and Biology in Phila-
delphia or from Miss A. E. C. Lathrop, Granby, Mass. The
several inbred rats were from the inbred strain of the Wistar
Institute. We are greatly indebted for them to Dr. H. H.
Donaldson and Dr. H. D. King. The accompanying list sup-
plies the reader with all available data concerning individuals
on whom the observations of this report were made.

Outbred Stock
Number Experiment

of rat Source and parentage Date of birth begun

251 &... .Granby* August 5, 1914 October 14, 1914

252 9 " " ' "

253 & " "

254 9 " "

1 Basset, G. C. Habit formation in a strain of albino rats of less than normal
brain weight. Behavior Monographs, 1914, 2, no. 4.

Number
of rat

1 C?.

2 9 .

3 d.

4 9.

261 d.

262 9 .

263 d\

264 9 .

265 d.

266 9 .

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS 113

Outbred Stock— Continued

Experiment
Source and parentage Date of birth begun

Wistar September — , 1914 . . . November 5, 1914

Granby, 251 c A x 252 9 . .October 28, 1914 March 2, 1915

<i «

a

u

. .November 28, 1914. . .
Wistar, 3d x 4 9 November 25, 1914. . .

Granby, 251 d x 252 9 . .November 28, 1914.
* Purchased of Abbie E. C. Lathrop, Granby, Mass.

Inbred Stock
Source and
Number generation of Experiment

of rat inbreeding Date of birth begun

201 d Wistar, 14th August 8, 1914 October 14, 1914

202 9 " " "

203 d " "

204 9 " "

5(? " September — , 1914 November 5, 1914

g 9 « a «

7 qP a a «

g Cj « « a

211 d 1 * Wistar, 15th December 18, 1914 April 11, 1915

212 9 * " " "

213 d 1 * " " "

214 9 * " " "

215 d* " " "

216 9 * " " "

* Offspring of 201 cT x 202 9 .

METHOD OF INQUIRY

Observations were made for above strains of rat as nearly
as possible at the same age, and for the sake of comparability,
on the same day. Certain of the observations were made under
the natural cage conditions; others, under definite experimental
conditions and for very specific purposes. These two kinds of
data, contrasted as the naturalistic and the experimental, tend
to supplement one another.

The naturalistic type of observation includes (1) observation

114 NENOZO UTSURIKAWA

of the position of an individual in the cage or nest box; (2) of
the relative positions of the two individuals, male and female,
in the cage; (3) of the degree of activity in the cage; (4) of sav-
ageness, viciousness, or tendency to bite. The experimental ob-
servation includes (1) measurements of quickness of response
to auditory stimuli; (2) of amount of movement in response to
the same stimuli; (3) of general behavior (restlessness) during
stimulation.

These several varieties of observation will now be reported in
tabular form, with scant discussion.

POSITION OF RAT IN CAGE

The animals were kept in rectangular, all-wire cages, the

floor dimensions of which were 16 inches by 14 inches. A single

pair of individuals, either outbred or inbred, was kept in a cage.

When the writer came into the animal room to feed the rats,

many of them would, as a rule, come forward in expectation of

food, but some would remain at the back of the cage or retreat

to the distant portion of the cage as the experimenter approached.

Some came forward singly; others, together. Sometimes the

individuals were found lying together in the cage ; at other times

they were observed to be distant from one another. The data

of table 1 concern, first, position of the two animals, male and

female, in the cage when the experimenter entered the room;

and second, the positions of male and female with relation to

one another.

TABLE 1

Position

OF

Rats in Cage

Outbred Rats

Date

Number
of rat

Position
forward

Male and female
together

Oct. 15 on

....251 d\.

77

. ... = .70...

110

9

=.08

110

«

253' " . .

76
. ... = .76. ..

7
= .07

100

100

Mar. 2 on

261 " . .

40
. ... = .67. ..

12
= .20

60

60

a

263 " ..

41
. ... = .68...

18
= .30

60

60

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS

115

TABLE 1 — Continued

Date

Number
of rat

Position
forward

Male and female
together

a

....265 (J 1 ....

57
60

.95

27

= .45

60

Nov. 5 on ... .

.... 1 "

65

.72

9
= .10

90

90

«

.... 3 "... .

43
80

.54

23

= .29

80

Average

437
700

.62

98

= .14

700

Oct. 15 on

....252 9 ....

82
110

.75

9

= .08

110

«

....254 " ....

71
100

.71

7

= .07

100

Mar. 2 on

....262 " ....

47
60

.78

12

= .20

60

«

....264 " ....

10
60

.17

18

= .30

60

a

....266 " ....

57
60

.95

27

= .45

60

Nov. 5 on ... .

.... 2 "... .

76

" " 90

.84 . . .

9

= .10

90

u

4 " .

65
90

.72

23

= .29

80

Average

490

.70...

98
= .14

700

700

Position of Rats in Cage
Inbred Rats

Date

Number
of rat

Position
forward

Male and female
together

Oct. 15 on

....201 & ....

40
....-- = .36...
110

43
— = .39

110

«

....203 " ....

40
. ... = .33...

120

24

= .48

50

Mar. 2 on

. ...211 " ....

26

. ... = .87. ..

30

11

= .37

30

116 NENOZO UTSURIKAWA

TABLE 1—

Continued

Inbred Rats

— Continued

Date

Number
of rat

Position
forward

Male and female
together

Mar. 2 on

....213 cT

30

. . =1.00...

30

19

=.63

30

«

....215 "

27
..— = .90...
30

12

= .40

30

Nov. 5 on

.... 5 "

28

. . = .35...

80

41

= .51

80

a

7 "

63

. . = .70...

90

22

= .24

90

Average

306

. . = .44...

700

293

= .42

700

Oct. 15 on

....202 9

39

. . =.35...

110

43
— = .39
110

it

....204 "

26

. . = .52...

50

24

= .48

50

Mar. 2 on

....212 "

26

. . =.87...

30

11

= .37

30

«

....214 "

29

. . = .97...

30

19

= .63

30

u

....216 "

27

. . = .90...

30

12
- - = .40

30

Nov. 5 on

.... 6 "

31

. . = .34...

90

49

= .54

90

a

.... 8 "

58

. . = .64...

90

22
= .24

90

Average

329

. . = .47...

700

293

= .42

700

The fractions of column 3 in this table indicate the relative
frequency of the forward position, that is, the presence of a
rat forward or its movement to that position on the approach
of the experimenter. The numerator of the fraction indicates
frequency of this position ; the denominator indicates the total
number of observations on the individual.

The chief results of these observations on the position of

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS

117

the animals in the cage and their relation to one another may
be stated thus:

(1) Outbred individuals more frequently come forward in the
cage than inbred.

(2) Females more frequently come forward than males.

(3) Individual differences are greater than are the differences
between the two strains or the two sexes.

(4) Inbred males and females are found together 3 times as
frequently as are outbred males and females.

ACTIVITY

The degree of activity of the several rats, as indicated by
their walking, running, climbing, washing, sniffing, seeking for
food, lifting the lid of the cage, and so on, was observed and
roughly graded by means of the numerals to 5, indicating
minimum activity and 5 maximum activity.

As in the previous case, tabular presentation is possible, and
in table 2 appear the comparable data for the two strains. In
this table, the numerator of the fraction is the sum of the various
grades given an animal in the total number of observations,
which appears as the denominator of the fraction. The average
grade in decimals in each case follows the fraction.

TABLE 2
Amount of Activity of Rats

Outbred
Date No. of rat Activity

286
Oct. 15 on.... 251 O 1 .... — = 2.6

110
221

....253 " .... = 2.2

100

145

Mar. 2 on.... 261 ".... = 2.4

60

103

....263 " .... = 1.7

60
94

....265 " .... = 1.6

60

Inbred
Date No. of rat Activity

163
Oct. 15 on. . . .201 (? — = 1.5

110

103

....203 " .... = .9

120

66
Mar. 2 on. . . .211 " ....-- = 2.2

30

59

" .. .213 "... . = 2.0

30
70

215 " ... . = 2.3

30

118

NENOZO UTSURIKAWA

TABLE 2-

—Continued

Outbred

Inbred

Date

No. of rat

Activity
191

Date

No. of rat

Activity
116

Nov. 5 on.

... ltf..

. = 2.1

90

Nov. 5 on.

... 5tf..

. = 1.5

80

u

. . . 3 " . .

101

.— = 1.3
80

u

. . . 7 " . .

174

. = 1.9

90

Average

139

= 2.0

70

Average

123

=1.8

70

Maximum
Minimum

2.6
1.3

Maximum
Minimum

2.3
9

Oct. 15 on.

...252 9 ..

266

. = 2.4

110

Oct. 15 on.

...202 9 ..

174

. = 1.6

110

a

...254

248

.— = 2.5
100

a

...204

51

. = 1.0

50

Mar. 2 on.

...262

147

. = 2.5

60

Mar. 2 on.

.212 " . . .

68
. = 2.3

30

a

...264

67

. = 1.1

60

«

...214

54

. = 1.8

30

u

...266 " ...

130

. = 2.2

60

«

...216 " ...

62

. = 2.1

30

Nov. 5 on.

. . . 2 " ...

213

. = 2.4

90

Nov. 5 on.

. . . 6 " . . .

120

.— = 1.3
90

a

... 4 " ...

202

. = 2.2

90

«

. . . 8 " . . .

143

.— = 1.6
90

Average

152

= 2.2

70

Average

117

= 1.7

70

Maximum
Minimum

2.5

1.1

Maximum
Minimum

2.3

1.0

Three comparative statements are justified by the data of
table 2 :

(1) Outbred rats are more active than inbred.

(2) The activity of the sexes differs much more for the out-
bred strain than for the inbred.

(3) Individual differences in activity are greater than either
sex or strain differences.

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS

119

SAVAGENESS OR VICIOUSNESS AS INDICATED BY THE
TENDENCY TO BITE

As a means of testing the savageness of the rats, a copper
wire was thrust into the cage from the front and from above,
and the floor was scraped or scratched with it. Some individuals
would, at this, dash forward and bite viciously and persistently
at the wire, whereas others merely noticed the disturbance and
were otherwise indifferent to it. The method of grading this
behavior was similar to that used in the case of activity.

The essential statistical data from these observations appear
as table 3.

TABLE 3
Savageness or Viciousness (Biting) of Rats

Outbred

Savageness

Inbred

Savageness

Date

No. of rat

(biting)

Date

No. of rat

(biting)

Oct. 15 on.

..251 d\.

92

110

.84

Oct. 15 on.

. . .201

o\.

31

. . = .28

110

«

...253 " ..

4
100

.04

(i

...203

a

109

. = .91

120

Mar. 2 on.

...261 " ..

2
' 60

.03

Mar. 2 on.

...211

a

79

. = 2.63

30

a

263

1
' 60

.02

u

...213

a

74

. = 2.47

30

a

265 " ...

48
' 60

.80

a

...215

«

26
. — = .87
30

Nov. 5 on.

. . 1 " . . .

10
' 90

.11

Nov. 5 on.

... 5

a

6
. — = .08
80

a

. . 3 " . . .

34
' 80

.43

a

.. 7

a

35

. = .39

90

Average

226
700

.32

Average ....

763

— 1.09

700

Maximum
Minimum

.84
.02

Maximum
Minimum

2.63
.08

Oct. 15 on.

..252 9 ...

6
110

.05

Oct. 15 on. .

..202

9...

86

. = .78

110

«

..254

52

100

.52

a

..204

a

161

. = 3.22

50

120

NENOZO UTSURIKAWA

TABLE 3-

—Continued

Outbred

Inbred

Savageness

Savageness

Date

No. of rat

(biting)

Date

No. o

f rat

(biting)

224

83

Mar. 2 on.

. . .262

$•■

. . = 3.73

60
99

Mar. 2 on.

. . .212

?■■

. . = 2.77

30

22

a

...264

ii

. .--= 1.65
60

60

u

...214

u

. . = .73

30
87

a

...266

a

. .--= 1.00
60
209

a

...216

a

. . = 2.90

30

163

Nov. 5 on.

... 2

it

. . = 2.32

90
301

Nov. 5 on.

... 6

11

. . = 1.81

90

440

u

... 4

u

.— = 3.34

90
1261

a

... 8

It

. . = 4.89

90

1710

Average

= 1.80

700

Average

= 2.44

700

Maximum

3.73

Maximum

4.89

Minimum

.05

Minimum

.73

The females far exceeded the males in degree of savageness.
They were, moreover, surprisingly quick and aggressive, whereas
the males were either on the defensive or indifferent. Whether
this remarkable sex difference is correlated with the maternal
instinct is not clear. It is, however, noteworthy that in the
feeding behavior the male very obviously asserts himself and
becomes the aggressor. It further appears from these obser-
vations that the tendency to bite is not directly proportional
to activity. Instead, there seems to be a tendency toward an
inverse relation.

The chief facts concerning savageness in the two strains are
these :

(1) Females are much more savage (exhibit the tendency to
bite more often and persistently) than are males.

(2) The inbred rats are much more savage than the outbred.

(3) Individual differences exceed either sex or strain differ-
ences, and they are especially great in case of the females.

EXPERIMENTAL OBSERVATIONS
The writer had intended to record the responses of his animals
to various stimuli and to more complex situations by means of
the galvanometer. But as the resources of the laboratory at

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS 121

the time did not permit of the development of a suitable form
of apparatus, this plan was abandoned in favor of a less expen-
sive and cruder preliminary mode of observation.

The apparatus finally devised and used consisted of a small
blackened rectangular box which rested at one end on two
pointed metallic posts and was suspended at the other end by
a delicate spring. This box was connected with a kymograph
by means of a marking lever so that any vertical movement of
the box was recorded on the kymograph surface. The rat was
placed in the box and so confined by means of movable parti-
tions that it was forced to hold its orientation with head pointed
forward toward the writing lever. The front end of the box
consisted of a wire screen. On the kymograph three records
were written: (1) A time line, indicating fifths of a second;
(2) a stimulus line; (3) a response line.

The only mode of stimulation here reported is the auditory,
and for this purpose an electric bell was used.

The reaction box and stimulus apparatus were enclosed in a
large pasteboard box in order that the animal should be some-
what protected from disturbing conditions.

QUICKNESS OF RESPONSE TO AUDITORY STIMULI

A rat having been placed in the apparatus and allowed to
become accustomed to its position, the various parts of the
mechanism were carefully adjusted, and when everything was
in readiness an auditory stimulus was given for .5 to .6 of a
second. After eight seconds, the stimulus was repeated. Then
the experimenter waited for an interval of another eight seconds
before again presenting a pair of auditory stimuli.

The quickness of response was indicated by the distance
between the point of stimulation on the stimulus line and the
point of initial response on the reaction line. The measurement
is extremely crude and inaccurate, but so far as may be judged,
not more so for the one strain or the one sex than for the other.
Where, because of long delayed or indefinite response, it was
difficult to decide on the initial point, the reaction was ignored.

The data of table 4 include the mean or average reaction
time for each individual in the first trial, that is, after initial
stimulus, and in the second trial, that is, with repetition of the
stimulus, the maximal and minimal reaction times, the total

122

NENOZO UTSURIKAWA

number of trials, and the number of trials in which no response
appeared. The chief results of these measurements of reaction
time may be stated thus:

TABLE 4

Quickness of Response to Auditory Stimulus
Outbred

First trial

Second trial

Number
of rat

Mean

Max.

Min.

No
resp.

No.

of

trials

Mean

Max.

Min.

No
resp.

No.

of

trials

251 c?

.35"*

.8"*

0*

3

10

.98"*

1.8"*

5

10

261 "

.39"

.9"

8

17

.21"

.35"

12

17

263 "

.51"

1"

3

16

.30"

1.2"

5

16

265 "

.51"

1"

7

15

.46"

1.2"

9

15

Aver.

Total

Total

Aver.

Total

Total

.47"

1"

21

58

.32"

1.2"

31

58

Average reaction time for the first and second trials = .40"

Total number of failures to respond in the first and second trials = 52

First trial

Second trial

Number
of rat

Mean

Max.

Min.

No
resp.

No.

of

trials

Mean

Max.

Min.

No
resp.

No.

of

trials

252 9

.43"*

1.0"*

3

10

.63"*

1.8"*

1

10

262 "

.38"

.7"

4

17

.43"

.95"

7

17

264 "

.15"

.3"

5

17

.30"

.7"

4

17

266 "

.28"

.6"

6

14

.80"

.8"

9

14

Aver.

Total

Total

Aver.

Total

Total

.27"

.7"

18

58

.51"

.95"

21

58

Average reaction time for the first and second trials = .39"

Total number of failures to respond in the first and second trials = 39

* Excluded from the averages, as the accuracy of measurements was uncertain.

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS

123

Inbred

First trial

Second trial

Number

No.

No.

of rat

Mean

Max.

Min.

No
resp.

of

trials

Mean

Max.

Min.

No
resp.

of
trials

201c?

.42"*

y*

.08"*

10

.33"*

7"*

0*

1

10

211 "

.18"

.3"

3

16

.26"

1.2"

3

16

213 "

.19"

.3"

1

16

.14"

.3"

9

16

215 "

.17"

.4"

2

16

.13"

.2"

7

16

Aver.

Total

Total

Aver.

Total

Total

.18"

.4"

6

58

.18"

1.2"

20

58

Average reaction time for the first and second trials = .18"

Total number of failures to respond in the first and second trials = 26

First trial

Second trial

Number

No.

No.

of rat

Mean

Max.

Min.

No
resp.

of
trials

Mean

Max.

Min.

No
resp.

of
trials

202 9

.26"*

2.0"*

.01"*

1

10

.30"*

.6"*

3

10

212 "

.23"

.85"

3

16

.22"

.65"

4

16

214 "

.29"

.5"

8

16

.08"

.08"

13

16

216 "

.28"

.8"

6

16

.23"

.8"

5

16

Aver.

Total

Total

Aver.

Total

Total

.27"

.85"

18

58

.18"

.8"

25

58

Average reaction time for the first and second trials = .23''

Total number of failures to respond in the first and second trials = 43
* Excluded from the averages as the accuracy of measurements was uncertain.

(1) Inbred rats respond more quickly to the auditory stimu-
lation than do outbred. (a) Inbred males respond most quickly
of all (.18 seconds), (b) Inbred females rank next in quickness
of response (.23 seconds), (c) Outbred females rank third (.39
seconds), (d) Outbred males are slowest of all (.40 seconds).

(2) The number of failures to respond obviously to the audi-
tory stimulation is both smallest and greatest for the males.

(a) The inbred males failed to respond 26 times in 116 trials.

(b) The outbred females failed to respond 39 times in 116 trials.

124

XEXOZO UTSURIKAWA

(c) The inbred females failed to respond 43 times in 116 trials.

(d) The outbred males failed to respond 52 times in 116 trials.

(3) The reaction time of the males is extremely variable; that
of the females is comparatively uniform.

(4) The sex difference in reaction time for inbred rats is slightly
more than for outbred.

(5) Reaction time, with certain exceptions, is shorter in case
of the second trial (repetition of auditory stimulus) than in the
first. Failures to respond are more frequent in the second trial
than in the first.

(6) Individual differences exceed sex and strain differences.

AMOUNT OF RESPONSE TO AUDITORY STIMULI

The amount of motor response to auditory stimuli was de-
termined by multiplying the maximum length (straight line) of
the response curve by the maximum height (straight line). The
result is expressed in square centimeters. These measurements
are even less reliable than those of reaction time, but again
there seems no reason to suppose that the errors are unequal
for either the sexes or the strains of rat.

As previously, the statistical data are arranged in tabular
form. They appear in table 5.

TABLE 5

Amount of Response to Auditory Stimuli

Outbred

First trial

Second trial

Number

No.

No.

of rat

Mean

Max.

Min.

No
resp.

of
trials

Mean

Max.

Min.

No
resp.

of
trials

251 c?

16.3

64

3

10

15.2

32

5

10

261 "

3.1

20

10

17

1.1

10

14

17

263 "

8.9

40

6

16

2.6

40

10

16

265 "

6.0

20

8

14

6.4

40

9

14

Aver.

Total

Total

Aver.

Total

Total

8.6

64

27

57

6.3

40

38.

57

Average amount of response for the first and second trials = 7.5 c.c.

Total number of failures to respond in the first and second trials = 65

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS

125

First trial

Second trial

Number
of rat

Mean

Max.

Min.

No

No.
of

Mean

Max.

Min.

No

No.

of

resp.

trials

resp.

trials

252 9

65.6

280

3

10

39.7

168

o

1

10

262 "

24.0

120

5

17

11.5

88

9

17

264 "

7.9

48

5

16

6.0

20

4

16

266 "

15.4

48

5

13

1.4

14

9

13

Aver.

Total

Total

Aver.

Total

Total

28.2

280

18

56

14.7

168

23

56

Average amount of response for the first and second trials = 21.5 c.c.

Total number of failures to respond in the first and second trials = 41

INBRED

First trial

Second trial

Number
of rat

Mean

Max.

Min.

No

No.
of

Mean

Max.

Min.

No

No.
of

resp.

trials

resp.

trials

201 cT

18.1

36

2

16

11.3

72

1

16

211 "

13.9

50

3

16

2.9

12

5

16

213 "

24.1

120

4

16

7.0

40

10

16

215 "

27.9

180

3

16

8.2

80

8

16

Aver.

Total

Total

Aver.

Total

Total

21.0

180

10

64

7.4

80

24

64

Average amount of response for the first and second trials = 14.2 c.c.

Total number of failures to respond in the first and second trials = 34

126

NENOZO UTSURIKAWA

First trial

Second trial

Number
of rat

Mean

Max.

Min.

No

No.
of

Mean

Max.

Min.

No

No.
of

resp.

trials

resp.

trials

202 9

133.1

640

1

16

35.3

192

3

16

212 "

29.0

96

4

16

17.1

168

7

16

214 "

14.4

48

8

16

11.3

144

12

16

216 "

9.5

44

6

16

8.0

48

10

16

Aver.

Total

Total

Aver.

Total

Total

46.5

640

19

64

17.9

192

32

64

Average amount of response for the first and second trials = 32 . 2

Total number of failures to respond in the first and second trials = 51

The following conclusions are indicated by the measurements
of amount of response:

(1)
bred.

(2)
(3)
(4)

The response of outbred rats is far less than that of in-

The response of females is greater than that of males.
The response is less in the second trial than in the first.
Individual and sex differences in amount of response far

exceed strain differences.

GENERAL BEHAVIOR DURING AUDITORY STIMULATION

As an indication of the restlessness or general tendency to
activity as a result of auditory stimulation, the total length of
the response curve was measured by means of a chartometer.
This could be done only very inaccurately because of the many
slight changes in direction of the curve, but as in the case of
previous experiments, there is no reason to suppose that the
degree of accuracy varies with the strains or with the sexes.

Measurements of restlessness were made for two conditions
of stimulation: (a) For separate auditory stimuli, the one
following the other after a stated interval and thus giving in
the records trial 1 and trial 2; and (b) for continuous auditory
stimulation instead of momentary.

Table 6 presents the data of restlessness or general behavior.

The principal results may be summarily stated thus:

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS

127

TABLE 6

Behavior During Experiment
Outbred

Number

With intermittent stimulation

With continuous stimulation

No.

No.

of rat

Mean

Max.

Min.

of
trials

Mean

Max.

Min.

of
trials

251 c?

17.4*

19.0

17.0

10

18.1

20.0

17.1

3

261 "

17.2

17.5

17.0

18

17.2

18.5

17.0

18

263 "

17.3

19.0

17.0

17

17.2

17.8

17.0

17

265 "

17.6

21.8

17.0

15

17.3

18.0

17.0

15

Average

Total

Average

Total

17.4

21.8

17.0

60

17.5

20.0

17.0

53

Average of both responses to intermittent and continuous stimulations = 17.5

Number
of rat

With intermittent stimulation

With continuous stimulation

Mean

Max.

Min.

No.

of

trials

Mean

Max.

Min.

No.

of

trials

252 9
262 "
264 "
266 "

22.5
17.4
18.4
17.3

38.2
18.0
22.7
18.5

17.0
17.0
17.0
17.0

10
18
17
14

23.3
19.5
18.4
18.1

26.2
38.5

21.0
21.0

20.5
17.0
17.0
17.0

2
18
17
14

Average
18.9

38.2

17.0

Total
59

Average
19.8

38.5

17.0

Total
51

Average of both responses to intermittent and continuous stimulations = 19.4
* Expressed in centimeters.

128

NENOZO UTSURIKAWA
Inbred

Number
of rat

With intermittent stimulation

With continuous stimulation

Mean

Max.

Min.

No.

of

trials

Mean

Max.

Min.

No.

of

trials

201 cT
211 "
213 "
215 "

17.7
17.8
17.2
19.2

20.0
24.0
17.5
18.2

17.0
17.0
17.0
17.0

10
16
16
16

17.5
18.9
17.6
17.4

18.0
21.6
21.0
20.0

17.0
17.1
17.0
17.0

5

16
16
16

Average
18.0

24.0

17.0

Total
58

Average
17.9

21.6

17.0

Total
53

Average of both responses to intermittent and continuous stimulations = 18.0

Number
of rat

With intermittent stimulation

With continuous stimulation

Mean

Max.

Min.

No.

of

trials

Mean

Max.

Min.

No.

of
trials

202 9
212 "
214 "
216 "

20.1
17.7
18.2
17.6

30.5
23.0
21.5
24.5

17.0
17.0
17.0
17.0

11
16
16
16

17.3
17.8
18.7
17.6

17.5
22.0
26.0
24.0

17.1
17.0
17.0
17.0

5

16
16
16

Average
18.4

30.5

17.0

Total
59

Average
17.9

26.0

17.0

Total
53

Average of both responses to intermittent and continuous stimulations = 18.2

(1) For inbred rats, the restlessness or continuity of response
is greater in case of momentary and repeated auditory stimula-
tion, whereas for outbred rats, the reverse is true.

(2) The amount of restlessness varies more widely for out-
bred than for inbred rats.

(3) The females of both strains show higher records for rest-
lessness than do the males.

(4) The records for the females are also more variable than
for the males, with averages as follows: Outbred males, 17.5;

TEMPERAMENTAL DIFFERENCES IN ALBINO RATS 129

inbred males, average 18.0; outbred females, average 19.4;
inbred females, average 18.2.

(5) Individual and sex differences greatly exceed strain dif-
ferences.

CONCLUSIONS

The outbred and inbred strains of albino rats contrast in
temperament as follows:

(1) Inbred males and females are found together in the cage
about three times as frequently as are outbred.

(2) The inbred rats come forward in the cage on the approach
of the experimenter much less frequently than do the outbred.

(3) Inbred rats are less active than outbred.

(4) The inbred stock exhibits savageness by biting to approx-
imately twice the extent of the outbred.

(5) The inbred animals respond more quickly and in greater
amount to momentary auditory stimulation than the outbred.

(6) The two strains differ also in restlessness or continuity
of response. For the inbred restlessness is greatest in case of
momentary and repeated auditory stimulation and less in case
of continuous stimulation, whereas for the outbred animals, the
reverse is true.

(7) The data indicate less difference between the sexes in the
inbred than the outbred rats.

RETROACTIVE ASSOCIATION AND THE ELIMINA-
TION OF ERRORS IN THE MAZE

HELEN B. HUBBERT AND K. S. LASHLEY

The temporal relation of the different activities which are
associated in the formation of a habit seems to have a direct
bearing upon the form in which the habit is fixed. Thus in an
experiment described by Bohn (Dontchef-Dezeuze, '14), carried
out in Pawlow's laboratory, the selection of stimulus and re-
sponse by their temporal relation seems established. In this
experiment an electrical stimulus was applied to the skin of a
dog, eliciting struggling and howling. Following this, food was
placed in the animal's mouth and salivary secretion was ob-
tained as a result. After this sequence had been repeated many
times the only reaction to the electrical stimulation of the skin
was the secretion of saliva. Why, in this experiment, was the
secretion of the saliva associated with the electrical stimulus
and not the struggling and howling with the taste of food? An
animal may be trained readily to reject food with the latter
reactions in a given situation by the use of punishment after
food is taken, 1 so it seems that the temporal order determined
which of the stimuli and responses were to become associated.

In the formation of complex habits it is probable that similar
and even more complex temporal factors modify the course of
learning. Hachet-Souplet ('13) has stated dogmatically that
when a series of actions, leading up to a pleasurable situation,
becomes connected into a habit the association occurs first
between the activities just preceding the pleasant result and
progresses to those more remotely antecedent. This may be
called the principle of retroactive association. If it is a fact
it has an important bearing upon theories of the mechanism of
selection in habit-formation.

In the work with the conditioned reflex it has been shown that
a well established reflex may serve as a basis for the formation

1 Hachet-Souplet ('14) describes an interesting case of this sort.

THE ELIMINATION OF ERRORS IN THE MAZE 131

of other conditioned reflexes as well as can an unconditioned
reflex (Bechterew, '13). If this condition obtains in the forma-
tion of a maze-habit, the situation involved in the last turn
before the food is reached may acquire from the getting of food
something (we can not be more definite at present) which makes
it capable of serving instead of the food in the fixation of the
next preceding activity. In this case the formation of a maze-
habit would appear as a series of secondary, tertiary, etc., con-
ditioned reflexes starting from the taking of food as the primary
reflex and progressing from the food compartment outward to
the entrance of the maze. Such an explanation avoids one of
the chief difficulties of theories of the fixation of habit; that of
accounting for the effects of getting food, or what not, upon an
activity which occurred half an hour or more earlier, such as
turning in a given direction at the entrance to the maze. 2

But such an hypothesis must be looked upon with suspicion
until the fact of retroactive association is established, and there
is little evidence in favor of this at present. With a slightly dif-
ferent problem in view Hubbert ('15) has made an analysis of
the order of elimination of errors in the maze. The results of
this analysis were not very certain owing to confliction in the
data given by different groups of animals. One point was quite
clear, however; there is no invariable sequence in the elimina-
tion of errors, if the records of single animals are considered.
In contrast to this, when the averages of very large groups of
animals are taken there does seem to be a progressive elimina-
tion of errors from the food compartment to the entrance of the
maze. This is shown by the following averages based upon all
the data given by Hubbert. (See figure 1 for the designation of
the alleys.)

a b c d e f

30.6 26.4 19.7 19.7 18.7 8.3 trials.

2 The theories of nervous drainage such as those formulated by Pawlow, Max
Meyer, Watson, and others seem to require the immediate succession of the acts
associated. In the experiments of Bonn, for example, they must assume that
the salivary reflex is excited while the nervous elements of the struggling reflexes
are still active and that the efferent fibres of the salivary glands drain off a part of
the energy from the cutaneous stimulation, thus opening the synapses from the
receptors of the skin to the salivary glands. None of the other theories of the
physiology of learning has been formulated in sufficient detail to take into con-
sideration the possibility of retroactive association.

132

HELEN B. HUBBERT AND K. S. LASHLEY

In the individual records we can not be certain either that a
single correct turn made by an animal in the maze proves that
the animal has learned the turn, or that a single error proves
the absence of the habit for the particular turn involved. The
former may be due to chance, the latter to a distracting stim-
ulus whose threshold of reaction is lower than that of the maze-
habit. In a question of this sort the most dependable results
are therefore to be obtained from averages, which rule out, to
a certain extent at least, the chance successes and errors. It
may be that while individual records do not show an exact pro-
gressive elimination of errors, the larger averages do reveal this

Figure 1. — Ground plan of circular maze, showing the position of the errors studied.

as one of a number of factors influencing the course of learning.
There is some doubt as to the comparability of the different
alleys in the circular maze and no great significance can be
attached to the progression except in the case of alleys b, c, d
and e, which are strictly comparable.

Vincent ('15, IV) has given data which indicates a more pro-
nounced progressive elimination than appears in the work of
Hubbert. In the form in which this is presented, however,
Vincent is not justified in applying the data to the problem. She
makes no statement as to what constituted a trial in her experi-
ments and in the two ' ' typical records ' of trials which she
reports in detail ('15, I) we find that the rat, after reaching the
food, was allowed to return and re-explore the maze. In the
first trial of her rat " No. 1 " the animal was allowed to leave

THE ELIMINATION OF ERRORS IN THE MAZE 133

the food compartment and explore the inner cul de sacs twice
before the trial was ended, and in the second trial the animal
left the food and explored the inner alleys five times. If the
final data is based upon trials conducted with this technique
it can have no bearing upon the problem of the elimination of
errors, for we can not determine whether the smaller number of
trials required for the elimination of the errors near the food box
resulted from retroactive association or from the fact that addi-
tional trials in the inner part of the maze were ignored.

Additional evidence dealing with the question of retroactive
association is now at hand and makes possible a more positive
conclusion concerning its role in habit-formation than could be
drawn before. The evidence consists of data upon the elimina-
tion of errors during the training of 56 rats in the Watson circular
maze. The method of analysis differs from that employed
earlier by Hubbert in that different types of errors are treated
separately.

The ground-plan of the maze is given in figure 1. As will
be noted, two chief types of error are distinguishable. The
first of these (type I) is the passing of a doorway through which
the animal should go. Errors of this type are marked with
even numbers in the figure. The other (type II) is that of an
incorrect turn after passing through the doorway. Errors of
this type are given odd numbers in the figure. In addition to
these types are errors of turning back upon the correct pathway.
While common enough in the early part of training these errors
are soon eliminated. In the 616 possible cases in the data
examined only one was found in which an error of turning back
followed the last error of types I and II in any alley (with the
exception of /, which, being without a partition, permits only
errors of turning back), so that they may be disregarded in any
study which deals only with the last error made at each given
point in the maze. In compiling the data any turn which car-
ried the animal for its own length or more into a blind alley
was counted as an error. Thus figure 2, a tracing of the path
followed in a single trial, shows errors at 2, 5, 7, and 11. The
records of all the animals were examined and note made of the
number of trials preceding the last in which each of the 11
possible errors was made. The averages of these for the 56 rats
are given in table 1.

134 HELEN B. HUBBERT AND K. S. LASHLEY

TABLE 1
The Average Number of Trials Required by the 56 Rats for the Elimina-
tion of Each of the 11 Errors Possible in the Watson Circular Maze.
The Numerals Indicate the Position of the Errors in Figure 1.

Errors of

passing a door

Errors of turning wrongly

Average trials

Average trials

Number

before

Number

before

elimination

elimination

2

30.07

1

36.62

4

23.19

3

36.48

6

19.18

5

37.97

8

12.48

7

39.21

10

21.05

9

32.46

Averages

21.19

Averages

36.55

Number 11

30.27

TABLE 2

The Number of Cases in Which Each of the Blind Alleys Was Not Ex-
plored Once Before the Maze-habit Was Thoroughly Established

Errors of passing a door Errors of turning wrongly
Number Number of cases Number Number of cases

2

1

4

2

3

6

3

5

8

22

7

10

4

9

2

"otal

31

Total

2

The most striking fact brought out by this table is that errors
of type I are eliminated in less than two-thirds as many trials
as are those of type II. Many rats never made certain errors
and the distinction between the types is shown here also. Table
2 gives the number of cases in which each of the errors was not
made once during the course of training. Blind alleys, entrance
to which constitutes an error of type II remained unexplored
in only two cases. In contrast to this, blind alleys beyond the
doorways remained unexplored in 31 cases.

What is the explanation of this fact? The first suggestion is
that there is a transfer of the reaction learned for one doorway
to other situations, but it may be also that a break in the smooth
walls of the alleys excites some instinctive mechanism which car-
ries the animals through the doorways. To test this latter pos-
sibility the records were examined to discover whether the
animals more frequently went through the first door that they
came to in their first trial in the maze or passed beyond it into
the blind alley, that is, whether or not error 2 was made at the
first possible opportunity. It was found that, of the 56 records

THE ELIMINATION OF ERRORS IN THE MAZE

135

of first trials 32 showed that the animal turned through the
first doorway and 12 that he passed the doorway. The remain-
ing 12 were uncertain owing to the confusion of lines made in
following the movements of the rats. Thus in three-fourths of
the available cases the animals, without previous training,
turned through the doorway instead of passing it; a fact which
furnishes evidence for an instinctive basis for the elimination
of errors of type I. It is probable that this accounts for the
difference in time required for the elimination of the two types
of errors.

Figure 2. — Tracing of path followed in one trial, showing method of counting
errors. Errors were made in this trial at 2, 5, 7 and 11.

An additional difference between the types of errors, which
is not explained by instinctive behavior, is the order of elimina-
tion within the series. Among the errors of type I there is a
very marked reduction in the number of trials required for the
elimination of successive errors from the circumference to the
center of the maze. Nothing of the sort is apparent in errors
of type II. 3 Is there a retroactive association in the case of

3 There is good reason for disregarding the two errors which are not in conformity
with the others, errors 10 and 11. After reaching the food at the center of th-
maze the rats frequently turn back and explore the inner alleys of the maze before
eating. These exploratory activities constitute one of the most characteristic in-
stinctive activities of the rat. In the home cages they are' usually not evident in
the scramble for food which takes place when several rats are together in a familiar
place, but in all relatively new situations the animals rarely begin to eat until they
have made a thorough exploration of their surroundings. In the maze the field of
exploration usually includes the food compartment, the alley /, and its doorway
which leads to error 10. When the rats have learned the maze they may eat as soon
as they reach the food, but any strange odor or startling noise may lead to a re-

136 HELEN B. HUBBERT AND K. S. LASHLEY

errors of type I and not of type II? If such association, in the
sense of a serial formation of conditioned reflexes, is an important
principle in habit-formation it must be obscured by other agents
in the elimination of errors of the second type. If not, it is for
some reason simulated in the elimination of errors of type I.
An inquiry into the relative complexity of the factors influencing
the elimination of the two types of errors gives data which seems
to favor the latter possibility. Movements toward the center
of the maze are much more influenced by gross orientation than
are those of turning to the right or left. After the third to sixth
trial the rats run with their heads near the convex (inner) walls
of the alleys of the maze and almost invariably confine their
efforts to climb out of the alleys to these partitions. Such be-
havior becomes more pronounced as they approach the center
of the maze and persists until the limits of training are reached.
It is improbable that this is a reaction to the odor of food in
the center of the maze, first, because it does not occur during
the first trials and second, because smearing the maze with food
does not alter the reaction essentially. It may be a reaction to
the curvature of the sides of the alleys but such an association
seems improbable in view of the slowness with which similar
sensory habits are formed in the discrimination box. The final
alternative is that of an orientation to the maze as a whole.
The work of Carr shows that the animals are usually well oriented
with respect to the direction of the food compartment from the
starting box and may even depend upon the direction of the
maze from their home cages for orientation. The behavior of
the rats strongly suggests that they very soon acquire this sense
of direction in the maze and that as they approach the center
of the maze the orientation becomes more certain, perhaps by
the summation of familiar stimuli, perhaps by the closer approx-
imation of the visible objects above the maze to their appear-
ence from the food-box.

The conditioning stimuli to an orientation determined either
by the curvature of the alleys of the maze or by the visible

tracing of this part of the path. This is evidence for an additional interfering
agent in the elimination of errors 10 and 11 which does not act in the case of the
other errors and justifies the view that the elimination of these errors is not directly
comparable with that of the others. In the experiments reported here every effort
was made to prevent this retracing of the path. The moment the animal entered
the food-compartment the experimenter hastened to close the doorway between
alleys e and /, and in very few cases did the rats escape into the outer alleys.

THE ELIMINATION OF ERRORS IN THE MAZE 137

objects above them, would obviously remain nearly constant,
no matter in what alley the animal happened to be. A similar
orientation with respect to right and left turns could not be
maintained equally well because the direction of the doorways
alternates with successive alleys. We have not had oppor-
tunity to test the mechanism of orientation but there can be
little doubt that the elimination of errors of type I is affected
by it to a greater extent than is that of errors of type II. We
can find, on the other hand, no evidence for any influence which
might hide a retroactive association in the elimination of errors
of type II. Such an influence would have to be equally strong
and equally uniform in its action with retroactive association and
the behavior of the rats gives no evidence of anything which
might produce this result.

Whatever the explanation of the serial elimination of errors
of type I, the process is evidently a complicated one and can
not be advanced as proof of retroactive association. No such
complication can be demonstrated in the elimination of errors
of type II. They show a pronounced uniformity in the amount
of practice necessary for their elimination; a greater uniformity
than should be expected if chance agents obscured the effects
of retroactive association, and it seems certain that their elimi-
nation presents a purer case of learning by the method of trial
and error than does that of type I. The number of animals
is large enough to make the averages dependable while still
allowing as great a chance variation as is found between the
averages in type II. Clearly the principle of retroactive asso-
ciation does not apply to errors of this type.

Since, as we have pointed out, the elimination of this type
of error is probably not complicated by transfer of training
from one part of the maze to another, by orientation toward the
center of the maze, or by reactions to the curvature of the pas-
sages, etc., we may extend our conclusion and say that retro-
gressive association is not an effective mechanism in the selec-
tion and association of simple series of motor activities in habit-
formation. 4

Evidence for the absence of retroactive association does not,
of course, throw much light upon the physiological mechanism

4 Here and elsewhere in this paper the word " association " is used to express
merely the production of a new temporal relation of stimulus and reaction.

138 HELEN B. HUBBERT AND K. S. LASHLEY

by which the getting of food fixes the habit. It indicates that
this is not the formation of a series of conditioned reflexes such
as was outlined in the first part of the paper. It suggests fur-
ther that the same mechanism is involved in learning similar
reactions in different parts of the maze and that the rate of
learning is the same irrespective of the temporal relation of these
to the getting of food.

LITERATURE CITED

Bechterew, W. von. Objective Psychologie oder Psychoreflexologie. Leipzig,

1913. Teubner.

Dontchef-Dezeuze, M. L'image et les reflexes conditionnels dans les travaux

1914. de Pavlov. Paris, Alcan.
Hachet-Souplet, P. De l'animal a l'enfant. Paris, Alcan.

1913.

1914. Notes psychologiaue sur les chiens de guerre. Jour, de Psych, norm.
et path., 11, 433-441.

Hubbert, H. B. Elimination of errors in the maze. Jour. Animal Behav., 5,

1915. 66-74.

Vincent, Stella B. The white rat and the maze problem. I. The introduction
1915. of a visual control. Jour. Animal Behav., 5, 1-24.
1915. The white rat and the maze problem. IV. The number and distribu-
tion of errors — a comparative study. Jour. Animal Behav., 5, 367-374.

A CAUSAL FACTOR IN THE RELATION OF THE

DISTRIBUTION OF PRACTICE TO THE

RATE OF LEARNING

K. S. LASHLEY

The Department of Psychology of the Johns Hopkins University

The fact is well established that, within limits as yet unde-
termined, the rate of learning varies inversely as the concen-
tration of the periods of practice. A number of hypotheses
have been advanced to account for this but none of them has
any experimental evidence in its support. One of them, offered
by Book 1 to account for improvement in typewriting during
periods of non-practice, assumes that during a long period of
practice the learner may acquire certain habits which, persist-
ing through the practice-period, limit his chance activities and
hence his possibility of improvement by the method of trial
and error. During rest-periods these habits, which are not very
well established, may be lost, in which case a distributed prac-
tice would permit of a greater diversity of activity than a con-
centrated one. In other words, during a long period of prac-
tice the learner is apt to get into a rut and intervals of rest
permit him to return to the problem with a new " set " and to
attack it in a different way.

The hypothesis applies not only to the periods of non-practice

dealt with by Book, but to any of the effects of the distribution

of practice. If it is correct, a detailed analysis of behavior in

the formation of any habit should show a greater diversity of

activity between successive practice-periods than within single

periods. In my experiments upon the acquisition of skill in

archery I observed the persistence of bad methods of aiming

through single periods of practice but, lacking time for detailed

descriptive or photographic records, was unable to determine the

role of these in modifying the effects of practice. 2

' x The Psychology of Skill: with special reference to its acquisition in typewrit-
ing. Missoula, University of Montana, 1908.
2 Acquisition of Skill in Archery. Carnegie Pub. 211. 1915.

140 K. S. LASHLEY

A simple method of recording significant motor activities for
a study of this problem is offered by the graphic maze. 3 With
it records are obtainable of all the errors made by the animals
being trained and it is easy to determine whether or not any
of the errors persist through single periods of practice or from
day to day. If such persisting errors are the real cause of
the results obtained with different distributions of practice, one
should expect to find, on the average, fewer errors common to
the last trial of one day and the first trial of the succeeding day's
practice than to any two successive trials made on the same day.

Comparison of the numbers of duplicate errors can not be
made directly owing to variations in the number of errors made
in different trials and the consequent difference in the prob-
ability of chance duplication. 4 Thus if nine out of a possible
twelve errors are made in each of two successive trials a larger
percentage of identical errors is to be expected from pure chance
than if only three errors are made in each trial. In making the
comparison I have used the records of all the errors made by
56 rats in learning the circular maze, when given five trials daily.
To avoid the influence of different numbers of errors in the
comparison of the number of duplicate errors appearing in single
and successive practice-periods the following method of com-
puting the results was adopted.

The records of the last trial in each day's practice and of
the first in the succeeding one, where a total of three or more
errors appeared in the two trials, were compared and the num-
ber of errors in each, together with the number of duplicate and
diverse errors, was tabulated. The records of the animal which
furnished this data were then examined for a case of two suc-
cessive trials on the same day each of which contained a number
of errors equal to that in the corresponding trial of the pair
made on successive days. The first pair of trials meeting these
requirements was taken for comparison and its numbers of dupli-
cate and diverse errors were arranged in a separate table. Where
no pair could be found which met the requirements the succes-
sive trials in different practice-periods were discarded.

3 Yerkes and Kellogg. A graphic method of recording maze reactions. Jour.
Animal Behav., 1914, 4, 50-55.

Watson. A circular maze with camera lucida attachment. Ibid., 56-59.

4 The methods of training animals and recording errors in the maze have been
described so frequently that they need not be discussed here. The reader who is
unfamiliar with the methods is referred to J. B. Watson, " Behavior," 1915.

RELATION OF PRACTICE TO RATE OF LEARNING 141

One hundred seventy-five pairs of successive trials made in
different periods of practice with an equal number of pairs made
during a single day's practice were obtained. The method gives
essentially a random sample of the two kinds of pairs of succes-
sive trials, with an equal number of errors and an equal distribu-
tion of errors between the members of the pairs of each series.
The role of chance in the production of duplicate errors may
therefore be disregarded, since it should be the same for both
kinds of pairs, and the relative proportion of duplicate to dis-
tinct errors in the two kinds of pairs may be considered the
result of the time intervening between the successive trials. In
computing the errors wrong turns only in the maze were con-
sidered and only one error was counted at each turn where
errors were made. The results of the analysis are given in the
following table. The totals of all cases are given and the figures
for the numbers of diverse and duplicate errors are directly
comparable, being based upon the same number of errors with
like distribution between the pairs of trials.

Successive Practice-periods

Total number of errors in last trials of first practice-periods 463

Total number of errors in first trials of succeeding practice-
periods 524

Total number of pairs of identical errors 247

Total number of diverse errors 493

Single Practice-periods

Total number of errors in first trials of pairs corresponding to

those above 463

Total number of errors in succeeding trials of pairs corresponding

to those above 524

Total number of pairs of identical errors 274

Total number of diverse errors 439

In both kinds of pairs the number of duplicate errors is slightly
in excess of the number of diverse ones. The maze, as arranged,
offers the possibility of 12 different errors and the average num-
ber of errors per trial was 2.82. The chances, then, that the
number of duplicate errors would equal the number of diverse
errors was only about one to eight. The fact that the numbers
were equal when the data included so many cases seems to prove
that there is some predetermining factor which causes errors
once made to be repeated both in single practice periods and
from one practice-period to the next.

142 K. S. LASHLEY

A comparison of the pairs of trials made during the same
period of practice with those made in successive periods shows
a greater diversity of errors in the latter. In them there were
247 pairs of like errors and 493 diverse errors; that is, 49.05%
of the errors made in successive trials were diverse. In the
successive trials made on the same day there were 274 pairs of
like errors and 439 diverse errors; 44.48% of the errors made
in these successive trials were diverse. The trials separated by
a 24 hour interval are thus seen to include 10% fewer duplicate
errors than those without intervening time: a greater diversity
of activity occurs where successive trials are separated by a
considerable interval of time than where they follow each other
immediately.

This is the result to be expected on the hypothesis considered
above and it seems to place the latter on a firm basis. Two
questions, however, are unanswered: first, is this the only factor
involved in producing the results noted when different distribu-
tions of practice are used; second, is the hypothesis applicable
to other types of activity than motor forms such as archery or
typewriting. It is not possible at present to say how great would
be the effect of the conflicting habits in retarding the rate of
learning in any particular case. If the diversity of activity is
reduced 10% by the concentration of practice from one to two
trials per day, a 10% increase in the amount of practice neces-
sary for learning is the least which could be expected, but the
fact that the disadvantageous activities are repeated more fre-
quently with increasing concentration of practice would help
to fix them as habits and thus produce an even greater retarda-
tion of learning. At a rough estimate, the persistence of the
same errors through prolonged periods of practice seems ade-
quate to account for such retardation of learning as has been
found to result from concentrated practice during the formation
of motor habits.

If the same explanation is not applicable to language habits, its
validity for motor habits is questionable. In view of our ignorance
of the mechanism of association in such activities generalizations
here must be made with caution. There can be little doubt,
however, that false associations occur in the formation of langu-
age habits and it seems probable that these may interfere with
learning in the same way that habits of making wrong turns in
the maze interfere with the learning of the correct path.

THE COURTSHIP OF PIERIS PROTODICE

PHIL RAU

The mating habits of Pieris protodice have been so uniform
in all of the cases which have come under my observation in
various times and places that I feel that this phase of their
behavior is fairly constant. The following cases are typical of
the usual performance.

On June 12, at 5:10 p. m., many of these white butterflies
were fluttering about one of their favorite haunts, a patch of
white-flowered milkweed in a sunny, treeless pasture. A female
was at rest on the upper surface of a leaf, with its wings spread
flat and its abdomen turned directly upward into the air and
held thus rigidly. A male was hovering above her with more
than usual activity; frequently he would flutter very near to
the upturned abdomen. Presently he grew more bold; he
repeatedly approached and beat his wings against hers with a
flapping motion, and darted toward her and touched her abdomen
with his own, apparently in an attempt to mate, until the ex-
citement grew intense after several such advances, and both
arose on the wing and fluttered and danced and fussed in an
almost quarrelsome manner in mid-air for a few seconds, when
the female settled to rest on another leaf nearby in the very
same position, and the whole performance was repeated. When
the male again became too familiar the female flew away as
before, the male following and the two fluttering 'round and
'round each other, high in the air. Then the coquettish female
suddenly darted away and hid under a leaf near the ground,
this time eluding her suitor.

Although there were hundreds of these white butterflies fly-
ing about and feeding upon the white flowers, only three pairs
were in copulo. Perhaps the late hour in the day might account
for the small number in mating.

On three occasions I chanced to see the beginning of this
strange performance, but I was unable to ascertain at any time
which sex made the first advance in this courtship. In each
case the female was feeding or resting on the flowers, with closed
wings in a normal, placid manner in so far as I could see, when

144 PHIL RAU

the male came dancing by on the wing. As he approached and
fluttered above her she promptly assumed the strange position
described above, the wings spread flat and directed slightly
backward, and the abdomen upraised. In each case they flirted
until the excitement became riotous and then they fluttered
away and must have separated, for they were lost among the
many others in the field. I have never seen the accomplishing
of actual mating, but in the several pairs observed already in
copulo they rested tail-to-tail. When disturbed they would fly
thus, in pairs. One pair continued flight for two and one-half
minutes without alighting. The leader (sex not ascertained) did
all the flying and seemed merely to carry the rear one.

On one occasion the pair continued their coquetry for fifteen
minutes, with several attempts at mating. During this period
the female moved from flower to flower, not with an air of trying
to escape, but rather leading him on; the male persistently fol-
lowed her, constantly on the wing and in a state of high agitation.
Finally he knocked her to the ground, where he followed her
still, and fell beside her, but he seemed exhausted and did not
try to mate, but lay as if spent. The female, when she found she
had exhausted him, arose on the wing and flew lightly and
indifferently away.

A pair in copulo were quietly at rest on a low shrub. I picked
them up in the fingers for examination and soon replaced them,
but after such disturbance they separated and went flying and
fluttering about each other, over the clay bank, over the stream,
over the low plants for about seven minutes, when they settled
for an instant and I thought they would reunite. But the female
dropped several inches to a lower stratum of leaves, remained
there for a few seconds and then darted away. I expected the
male to go in hot pursuit; instead, for the next ten minutes,
while the female was dancing over some shrubs a hundred yards
away this male was frantically going in and out among the
leaves in the spot on the lower stratum where the female had
paused for a few seconds. It was pathetic to see his eager search ;
the only spot that seemed to have an attraction for him was
the place where the female had been, even though he had not
seen her drop to it, but had only located the spot after she had
left it. As the male hunted frantically about her former loca-
tion, while the female was dancing in full view, I could not
help but feel convinced that in this case at least the sense of
sight was inferior to the sense of smell.

JOURNAL OF ANIMAL BEHAVIOR

Vol. 7 MAY-JUNE No. 3

THE DISTRIBUTION AND ELIMINATION OF
ERRORS IN THE MAZE

HARVEY CARR

University of Chicago

In the solution of the maze problem, animals do not distribute
their errors equally between the various cul de sacs; apparently
these blind alleys do not offer equal incentives to entrance.
Neither do the cul de sacs present the same difficulty in mas-
tery; the tendency to enter certain alleys is eliminated much
sooner than in the case of others. The principles determining
the relative frequency of entrance into the various alleys, and
the factors governing the order of their elimination are inade-
quately known.

Vincent 1 has published data concerning the first problem.
Hubbert 2 and Vincent 1 have investigated the second question,
and Watson, 3 on the basis of Hubbert's results, has contributed
to the discussion.

Watson contends that if food satisfaction is a causal agency
in selecting the true path and thus eliminating the cul de sacs
the latter part of the maze should be mastered first; hence the
cul de sacs should be eliminated in the order of their nearness
to the food box. He therefore concludes that food satisfaction
can not be regarded as a selective agency because only seven
of Miss Hubbert's eighty-four rats eliminated the six errors in
the exact order of their spatial contiguity to the food box.

Vincent. The White Rat and the Maze Problem. Jour. Animal Behav., 5,
367-374.

2 Hubbert. Elimination of Errors in the Maze. Jour. Animal Behav., 5.

3 Watson. Behavior, p. 268.

145

146 HARVEY CARR

Watson's argument, to my mind, contains two fallacies : 1. A
causal factor can mediate an invariable result (without any
exceptions) only under the ideal condition that it is the only
causal agency involved. Needless to say such a condition does
not obtain in any science, to say nothing of animal behavior.
The ideal effect of any hypothetical cause is always somewhat
altered, distorted, or even obscured by the influence of other
factors which can not be controlled. Hence general tendencies
and results must be utilized as diagnostic symptoms in the
search for causal conditions. Viewed in this light, Hubbert's
results support rather than disprove the efficacy of the alleged
principle of selection, for the general trend of the order of elim-
ination is that of the spatial contiguity of the alleys to the food
box. According to my computations, 80% of her rats eliminated
the 6th error before the 5th, 50% the 5th before the 4th, 47%
the 4th before the 3rd, 73% the 3rd before the 2nd, and 70% the
2nd before the 1st. Determining the average number of trials
necessary to eliminate each cul de sac, the order of elimination
for the entire group was 6-5-3-4-2-1, where the successive errors
are numbered in order from the entrance. This order gives by
the rank method a positive correlation of .943 between quickness
of elimination and propinquity to the food box. The average
number of trials necessary to eliminate the last three errors
was less than that for the first three for 90% of the rats. Surely
there is a very pronounced tendency for the errors to be mastered
in proportion to their nearness to the food box, and the devia-
tions from an exact correlation for each rat may well be due to
the operation of other causal agencies. The existence of other
efficacious agencies, viz., recency and frequency, is admitted by
Watson. 2. Granted that food satisfaction is the only effective
agency involved, yet a perfect correlation between speed of
elimination and nearness to the food box, as demanded by
Watson, would not obtain. According to the hypothesis the
order of elimination will be determined, not by the spatial
arrangement of the cul de sacs in relation to the food box, but
by the temporal relation of the errors to the food experience.
The temporal order in which the cul de sacs are entered is not
the same as their spatial order in the maze. Not all cul de sacs
are entered on each run. An animal may enter alleys 1, 3,

DISTRIBUTION AND ELIMINATION OF ERRORS IN MAZE 147

and 5, and skip 2,4, and 6. The matter is further complicated
by the phenomenon of returns. The tendency to return toward
the entrance box is very persistent during the early stages of
learning, and in any trial we may have a temporal order of
1-3-6-1-2-3-4-5-4. According to the hypothesis this trial will
tend to eliminate alleys 3, 4, and 5 prior to 6.

Hubbert's results, however, neither prove nor disprove the
efficacy of food satisfaction as an eliminative agency. Although
a high degree of correlation obtains between rate of elimination
and nearness to the food box, one should still refrain from gen-
eralizing on the basis of one maze. Different maze patterns
may give other results. A correlation between two factors does
not always indicate a causal relation between them; both may
be the result of some more fundamental condition.

This paper presents data on two mazes, and in addition Miss
Vincent has kindly furnished me records for six mazes. A cul
de sac was considered eliminated when but one entrance was
made in ten successive runs. The number of trials necessary
to eliminate each alley was determined for each animal and
the average number of trials for the group was thus computed
for the various cul de sacs. These values constitute the order
of elimination for the group. This temporal order of mastery
was correlated by the rank method with the spatial arrangement
of the alleys in relation to the food box. Table I designates
the various mazes, gives the number of cul de sacs in each, the
number of rats employed, and the correlation data for each maze
pattern. Mazes I-a to I-e have the same pattern, a slight
modification of the Hampton Court arrangement; they differed
only in the arrangement of sensory factors. I-a presented a
uniform sensory interior; I-b had the true path painted white
and the blinds black; I-c had the true path painted black and
the blinds white; in I-d an olfactory trail was laid in the true
path, while the trail was inserted in the cul de sacs for I-e. Il-a
and Il-b were alike except that the alleys of Il-b were without
sides. The records for the above mazes were obtained by Miss
Vincent. Mazes III and IV were somewhat similar in pattern
but radically different from patterns I and II. Maze V refers
to the circular maze used by Miss Hubbert whose data are given
for comparison.

148 HARVEY CARR

TABLE I

Correlation Between Quickness of Elimination and Nearness

to the Food Box

Number of

Number of

Percentage of

Maze

cul de sacs

rats

correlation

I-a

7

10

.607

I-b

7

10

.714

I-c

7

9

.178

I-d

7

6

— .643

I-e

7

6

— .358

Il-a

7

6

.750

Il-b

7

6

— .822

III

11

16

.563

IV

9

15

.666

V

6

84

.943

Varying degrees of positive correlation between the two factors
were obtained for six of the nine mazes. None of our values
are as high as that obtained for Miss Hubbert's maze. There
are, however, three exceptions to a uniform positive correlation;
in these three mazes the errors were mastered in proportion to
their nearness to the maze entrance. This lack of a uniform
positive correlation can be interpreted in two ways. 1. If food
satisfaction is an effective agency of elimination, its influence is
overcome by some other selective factors which are peculiar to
three mazes. 2. On the other hand we may suppose that food
satisfaction is non-effective and that all of the above correla-
tions (both positive and negative) are to be explained in terms
of a single principle. The latter hypothesis is the preferable one.

Miss Vincent suggested in her paper the possibility that the
ease of elimination is a function of the strength of the tendency
to enter an alley, and that in a general way the relative attrac-
tiveness of the various alleys can be measured by the frequency
with which they are entered. Cul de sacs with the greatest
error score offer the most inducement to entrance, and the
stronger the attraction, the larger will be the number of trials
necessary for elimination. On this basis a negative correlation
will obtain between the number of errors for each alley and the
order of mastery. To test the hypothesis, the various cul de
sacs were now ranked in the order of number of entrances made
by the group for the successive stages in the mastery of the
maze. Correlation data were computed by the ranking method
with much detail but since these results were uniformly consis-
tent for all stages of learning, we give in table II the values

DISTRIBUTION AND ELIMINATION OF ERRORS IN MAZE 149

for representative stages. The first horizontal column states
the correlation values for the various mazes between the order
of elimination of the alleys and the relative number of entrances
made by the group as a whole for the first trial. The second
column gives similar data in relation to the total number of
errors made during the second and third trials. The last three
columns state the results in reference to the total number of
errors made in the first five runs, the second five runs, and

4

from the eleventh trial until the maze was mastered.

TABLE II

Correlations Between Quickness of Elimination and Number of
Errors at Different Stages of Learning

Trials I-a I-b I-c I-d I-e Il-a Il-b III IV

1 -.571 -.821 -.535 -.714 -.892 -.321 -.642 -.654 -.133

2-3 -.535 -.643 -.892 -.857 -.643 -.107 -.321 -.300 -.166

4-5 -.642 -.571 -.821 -.535 -.500 -.642 -.964 -.764 -.350

1-5 -.714 -.892 -.750 -.857 -.857 -.143 -.642 -.758 -.116

6-10 -.857 -.857 -.678 -.750 -.857 -.857 -.571 -.518 -.583

11 -until learned.. -.500 -.785 -.857 -.535 -.321 -.928 -.000 -.973 -.866

With one exception, a uniform negative correlation was ob-
tained, the values ranging from —.107 to — .964. The excep-
tion refers to the final stage in the mastery of maze I I-b. This
record hardly invalidates the uniformity of the results as this
maze was mastered very easily and but few errors were made
after the tenth run. In considering the validity of this correla-
tion, the uniformity for all mazes and for the various stages of
learning must be emphasized. A single correlation value may
well be due to chance in view of the paucity of data for which
the correlation was calculated. But chance is pretty well elim-
inated when the computation is repeated 54 times and consistent
results are secured. We may then safely conclude that some
degree of negative correlation obtains for all stages of learning
between quickness of elimination and the tendency to enter the
cul de sacs, that in a general way those blind alleys which for
some reason offer little inducement to entrance are easily elim-
inated, while cul de sacs which present the most enticement to
exploration are the hardest to master. This correlation is a
priori logical, for it is reasonable to expect that the strongest
tendencies will be the hardest to overcome. Our results thus
establish one of the factors underlying the order of error elimi-

150 HARVEY CARR

nation, viz., that this order is a function of the strength of the
entrance attraction characteristic of the various cul de sacs.
It is admitted that other selective factors may also be effica-
cious; otherwise higher correlation values should have been
obtained.

The validity of the above principle of explanation is sup-
ported by the data of table III, giving the correlation values
between the number of errors for the various cul de sacs and
their spatial order in the maze. A positive correlation obtains
for the six mazes I-a, I-b, I-c, Il-a, III, and IV. In these cases
the animals entered the various cul de sacs with a frequency
roughly proportionate to their proximity to the point of entrance,
and it was for these six mazes that a positive correlation was
found between order of elimination and nearness to the food
box (table I). In other words the first cul de sacs were entered
the most frequently and were the last to be eliminated. In
mazes I-d, I-e, and Il-b, on the other hand, the animals for some
reason entered the last cul de sacs more frequently than the
initial ones (negative correlation between number of errors and
nearness to entrance), and in these cases the final errors were
the last to be eliminated (table I).

TABLE III

Correlations Between Nearness to Entrance and Number of
Errors at Different Stages of Learning

Trials I-a I-b I-c I-d I-e Il-a Il-b III IV

1 322 .929 .822 -.964 -.107 .000 -.607 .846 .750

2-3 250 .679 .036 -.928 -.107 -.392 -.214 .709 .666

4-5 232 .143 .215 -.321 -.750 .143 -.892 .373 .800

1-5 322 .715 .679 -.928 -.214 -.107 -.607 .846 .700

6-10 286 .786 .643 -.250 -.500 .465 -.107 .573 .733

11-until learned.. .215 .250 .179 -.285 -.071 .536 .393 .500 .417

It is thus the phenomenon of error distribution in the maze
that demands explanation in order to comprehend the order of
elimination. Our results enable us to offer but few explanatory
suggestions in regard to error distribution.

1. The persistent tendency for rats to keep returning to the
point of entrance after exploratory excursions operates to in-
crease the number of entrances into the initial cul de sacs. This
returning tendency is well known, and it is at once obvious that
these returns must result in repeated explorations of the initial

DISTRIBUTION AND ELIMINATION OF ERRORS IN MAZE 151

cul de sacs. All other factors being equal, we should expect
that the relative number of entrances into the various cul de
sacs will be roughly proportionate to their nearness to the point
of entrance. This influence of returns is further indicated by
two features of the data: a. In mazes I-d and I-e there is a
negative correlation between number of errors and nearness to
the entrance, the greater number of errors being made in the
final cul de sacs (table I). The opposite relation, however,
obtains for mazes I-a, I-b, and I-c, in which the greater number
of entrances were made in the initial cul de sacs. This differ-
ence in the distribution of the errors among the cul de sacs is
due entirely to differences of the sensory character of the mazes,
for exactly the same maze pattern was used throughout and the
objective environment of the mazes was identical in all cases.
Mazes I-d and I-e differed from the others in that an olfactory
trail was laid, either in the true pathway or in the cul de sacs.
This trail produced several characteristic peculiarities of beha-
vior, one of which was a noticeable diminution of the number
of returns. Miss Vincent kept no separate record of the number
of returns, but she noted this feature of the behavior and fre-
quently discussed its significance with the writer at the time.
The degree of returning also accounts for the different distribu-
tion of errors in mazes Il-a and Il-b (table III). Again the
same pattern was used and the maze was located in the same
objective environment in both experiments. The only difference
consisted in the presence and absence of sides to the runways.
Maze Il-b was without sides and their absence caused the ani-
mals to follow one edge of the platform with their paws or
vibrissae much in the same manner as did the olfactory trails.
Following an edge caused a noticeable diminution in the number
of returns and thus accounts in part for the difference of error
distribution in the two mazes, b. The returning tendency is
quickly eliminated in the course of maze mastery, and we should
thus expect that the relative frequency of entering the initial
cul de sacs will be gradually diminished during learning for
those mazes in which returns are an effective factor in error
distribution. Such a diminution obtains for mazes I-a, I-b, I-c,
III and IV (table III). In all five cases there is a high positive
correlation between number of errors and nearness to the en-
trance, but this correlation keeps decreasing with the number

152 HARVEY CARR

of trials; in other words the number of errors in the initial blind
alleys is decreased more rapidly than for the final cul de sacs.
2. The sensory character of the maze influences the distribu-
tion of errors in other ways than by minimizing the number of
returns. The tendency to follow the edge or an olfactory trail
evident in mazes I-d, I-e, and Il-b operated to reduce the total
number of errors. This fact is well demonstrated in Miss Vin-
cent's paper. The operation of the tendency was relatively
more effective in the initial than in the final stages of learning,
and Miss Vincent interpreted this fact as due to the shift in
control from olfactory and cutaneous cues over to kinaesthetic
factors; when the maze is run in terms of kinaesthetic stimuli,
the trail factor is no longer present and the rat is subject to the
enticements of curiosity, distractive odors, etc. I wish to sug-
gest a similar explanation of the greater frequency of entrance
into the final cul de sacs in these three mazes. Starting from the
entrance box, the rats are at once dominated by the stimulus
characteristic of the trail. This sensory trail is followed and
possible exploratory excursions due to curiosity, fear, and other
motives are eliminated. As a consequence relatively few errors
are made in the first sections of the maze. As the final sections
are reached, however, the dominance of the trail motive is weak-
ened and other enticements begin to operate. The weakening
of the trail motive may be conceived under such terms as habit-
uation or adaptation. The strength of other motives such as
fear, curiosity, and the returning tendency may progressively
increase with the distance traversed; the concept of summation
of stimuli may be applicable here. The strength of the olfactory
stimulus from the food box must necessarily increase as the
final sections of the maze are reached. One can hardly suppose
that the actions of an organism so complex, alert, variable, and
thoroughly alive as is the rat can long be continuously domi-
nated by a single motive. Some shift of motives during a run
must be presupposed irrespective of the explanatory difficulties
involved. Our hypothesis then assumes that the animals in
these three mazes start out on each run under the dominance
of the trail motive which tends to prevent errors. As the final
section of the maze is reached, there occurs a shift of motives,
and these new motives, such as food odor, fear, curiosity, etc.,
tend to produce errors. As a consequence of this shift of motives

DISTRIBUTION AND ELIMINATION OF ERRORS IN MAZE 153

during each run the greater number of errors will be made in
the final cul de sacs in these three mazes. It is also possible
that the same principle, a shift of motives, will account for the
greater frequency of entrance into the initial cul de sacs in
the normal mazes; in this case the shift will be from fear and
curiosity over to the food odor. The odor of food as the rat
reaches the final sections should be sufficiently directive as to
minimize the number of errors relative to those made when the
animal's acts are controlled primarily by curiosity and caution.

3. The distribution of errors is influenced by certain peculiar-
ities of the cul de sacs other than their positional relation to the
point of entrance. Since the influence of this factor may vary
according to the stage of mastery, a separate treatment is neces-
sary for its effects upon the initial and final distribution of
errors.

a. Initial distribution. — From the data of table III it is obvious
that the initial errors (first five trials) are never distributed in
the exact order of the spatial arrangement of the cul de sacs. A
few blind alleys are generally responsible for the deviations, and
these are listed in table IV. The first columns give the various
mazes with the number of cul de sacs belonging to each. Each
cul de sac is numbered in order from the point of entrance. In
the column headed ' plus ' are listed those alleys in which
the number of entrances exceeds that to be expected on the
basis of a perfect correlation. In the " minus " column are those
alleys in which the number of entrances is less than the normal.
The first group of mazes are those in which a positive correla-
tion obtains between the distribution of errors in the first five
trials and the proximity of the cul de sacs to the point of entrance.
In the last group of mazes the initial errors are distributed
proportionate to the spatial contiguity of the cul de sacs to
the food box. The five mazes, I-a, I-b, I-c, I-d, and I-e, are
identical in pattern; in the first three the sensory conditions
were such that the greater number of errors were made in the
initial cul de sacs; in the last two, the first cul de sacs were the
least attractive. In spite of this radical difference of error dis-
tribution in the two groups, the same cul de sacs are responsible
for the deviations from a perfect correlation in every case. The
first cul de sac is much less alluring in the initial trials than its
position would justify. On the other hand No. 6 was invari-

154 HARVEY CARR

ably attractive for it was entered much more frequently than
one would expect if the spatial order of the cul de sacs were the
only factor determining the distribution of errors in the initial
trials. Likewise No. 3 was relatively difficult in two cases, and
No. 4 three times. Comparing mazes Il-a and Il-b of like
pattern but with a different distribution of errors, we find that
No. 7 is easy in each case while No. 4 tends to be difficult.
Chance will account for the deviations in part, but chance can
not invariably favor certain cul de sacs. The tendency toward
uniformity in the character of the deviations for the same maze
pattern must be due to peculiarities of the cul de sacs other
than their spatial position in the maze. Certain ones are rela-
tively easy to avoid, while others are prone to be entered.

The susceptibility of the animals to these favored cul de sacs
is, however, an individual matter. In maze III, No. 8 received
more than its due share of entrances, yet this alley was not
entered at all in the first five trials by four of the sixteen rats;
in fact one animal did not enter this alley once in fifty trials,
while another rat entered but three times in fifty runs. This
avoidance of the difficult errors by certain rats can in part be
explained in terms of the habits acquired by those animals in
traversing the first section of the maze. Let us suppose that
the first section of a maze contains no blind alleys but consists
of a series of runways so arranged as to necessitate alternating
right and left turns of 90°. At the end of this section there is
presented a choice between two paths, — a ' straight ahead '
error, and a turn of 90° leading to the food box. It is conceivable

TABLE IV

Deviating Cul de Sacs

Number of
Maze cul de sacs Plus Minus

A. Mazes exhibiting a positive correlation

I-a 7 3, 6 1

I-b 7 4, 6 1

I-c 7 4, 6 1, 2

Il-a 7 4 1, 2, 3

III 11 8 7, 10

IV 9 3, 4, 8 6

B. Mazes exhibiting a negative correlation

I-d 7 3, 6 1

I-e 7 3, 6 1, 5, 7

Il-b 7 4, 6 3, 7

DISTRIBUTION AND ELIMINATION OF ERRORS IN MAZE 155

that a choice between two such possibilities might be determined
to a large extent by tendencies aroused in traversing the previous
zig zag course. On this hypothesis the attractiveness of any
cul de sac in a maze is in part a function of the motor tendencies
developed in the prior sections of the maze. The habits aroused
in a group of individuals in traversing the initial section of a
maze must necessarily have much in common, and yet any part
of a maze presents possibilities of wide divergence in the char-
acter of the pathway actually traversed. A group will thus
approach a cul de sac with some degree of uniformity of dispo-
sition toward it, but radical exceptions are possible.

b. Final distribution. — The final distribution of errors repre-
sents the order of elimination, for eliminated cul de sacs are
those which are not entered. As previously developed there is
a correlation between the initial distribution of errors and the
order of mastery of the blind alleys, but this correlation is far
from perfect. It is thus evident that the order of elimination,
or the final distribution of errors, is also dependent upon other
factors than initial attractiveness. Certain peculiarities of the
cul de sacs constitute a determining condition. Some blind
alleys are relatively difficult and others are relatively easy to
master, and this ease or difficulty of a cul de sac is to some
extent independent of its initial attractiveness.

The above principle is well illustrated by mazes III and IV.
The eleven cul de sacs of maze III fall into four rather well
defined groups as to rate of elimination. The progress in mastery
of four cul de sacs typical of these groups is represented by the
curves of fig. 1. The alleys chosen as types are 1, 2, 5, and 11.
The values represented are the total number of errors made by
the group for each successive five trials. Curves 1, 2, and 11
illustrate the first principle that the order of mastery is inversely
proportionate to the number of initial errors. No. 11 elicited
but few initial errors and was mastered first; No. 1 was the most
attractive but the most difficult, while No. 2 occupies a median
position between the two. As to progress of mastery the three
alleys are to be ranked in order, 1, 2, and 11. No. 5 is the
exception which illustrates the influence of the second factor.
This cul de sac was the hardest of the eleven to master, and
yet its initial attractiveness was no greater than its position
would justify. The number of entrances into this alley rapidly

156

HARVEY CARR

-

Ml !

t^rt II ' ■ ; i ■ ! i i

J 00 ; | ! | i i , | | | i 1

1]

1

V ! i 1

1 .

1

1 i

1 i

"111 ' ' ' ■ i i ' ■ : I '.

1 ill i ' i l 1 ' 1 l It'll II J 1 | II

! II 1 ' 1 1 1 i 11 M '"III III

1 ] ! | ' | M ' | | II i

i"

i i-i mi in ; '

ill mi '1 1 MM

..1-1 ' L_ 1 1

Tnn 1

1ZU I '

1 1 ' 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ! , 1 1 1 i| II 1 1 1 '

1 1 1 1 1 1 1 1 1 ii 1 1 r 1 1 1 i i i I i ' ! 1 1 1 1 II r II 1 1 1 1

Mil' 1 1 .1 1 1 1 i 1 i ' 1 1 1 ' I 1 1 1 1 1 • 1 ' 1 1 ' 1 1 1

1 ' ' 1

i 1 1 1 1 1 1 1 1 i . m ' m i ' i i ' 1 i i ' ' ' i 1 1 i m

1 I'M

• , ; 1 1 M M I i ■ ■ ' i M ' ' M

i i 1 | i 1 1

i mm M i I ! I ' 1 ! M

V I i i - 1 II- l

• 1 i i I 1 Mil 1 1 ' i ' ' ;

1 1 i i i 1 i 1 i i I III i Mill 1 II 1 III

1 , ''I ilM 1 1 ; ill

1 1 1 1 ' 1 1 | ' 1

1 1 | | | • | 1 III

1 1 lil I i . i III 'ill 1 1 1 1

II 1 1 1 1 i I

I

1 , I M 1 rli- ' II |

on T ' j - ' ■ ' i : - 1 1 l| " TTT • T 1 1 T " i ;

&U '1 ■ ! ' Mill

Ml 1 M HI ' i M i

1 1 II 1 il II iii i i MM

1 Ml i ■ M i ' M

i 1 1 i i i i

MV ' I'M' i ■ ■

i i\ 1 1 i 1 1 i i i

\

I ii. i

V m

V ! , ,|: ! j^V i M M ' 1 M M M

VI < *T

\ .* M M ,N k.ii I

1 1 ii' X.rl'liill , ^-

\ * "N.

1 1 1 J\ 1 ■ I ' M ' ■*** J!' 1 1 I

<\ M 1 1 ■ : : ^~ 4_ '

r 1 M iJv' i ■ > - '

T" ' i 7 M ' 1 1 X_! II 1 ! i i \/ i ' \ i ! ' : i « ; ' . ' , ' ! i ' i r" ; ! , rh L \ !

i.i i ^ "S^ <jT 1 Pkl 1 ' ^^a \

^l #111 : '^^ J*' \ u

\t\ ' ^v * ' ^v. *r\ \

41 ) ^^ ^ ^^^^T" ^^^^ ^b

L[^fcJ ' ^f^ ^^*"^*J X*

i \f\\ \ ^^ 1 ' r^v ! ' 1 *\ ! 1 ! ■ ' ' ' r*Hl r SJ. ' ' 1

1 i i \ J ' ' ^^- ' 1 . i *^V — mm~^ ' ' ■ ■ ■ ■ : - ' T ^^' ~~ ^^v, • i 1 > ' i ^ ] "

r 'S. ^^ kJ

^\ ^w

V

^y I -r r

■ i ' M*^^ ^

V f^^ ^^^^

■ I' 1 1 ' ' II 1 m ^ * j-r^ iii ' Tr H*. rt

1 | ; I i Stt ■ . i ~n — i 1 — 1 " "" i ' *"' H~!

^^^

^v

<p s. r | ' i Ml m ' M '

v. W^f, i mTI 1 1 1 i 1 1 n 1 1 T 1 1 mi

:-i-< i -L *"' I — - i ._ ; .. ' . .. , i . — - . — — -p 1

" T ■ 1 ir 3 : „r or ir I-Ilj.

J rials 15 t Ml . 25 - ^X- .60. 4o

m I i - IT i T T

F ■*!...... i , . . T1

DISTRIBUTION AND ELIMINATION OF ERRORS IN MAZE 157

increases for the first fifteen runs, and then slowly decreases
for the remaining thirty-five trials. , More errors were made
at the end of the experiment than at the beginning. In con-
sidering the individual records, the general features of the group
curve are also characteristic of that of fifteen of the animals.
In maze III, the difficult group of cul de sacs is composed of
alleys 5, 7, and 8. In maze IV, alley 3 was relatively easy
while 5, 6, and 9 were the difficult ones.

This difficulty of any cul de sac must be explained mainly
in terms of the animal's organization in reference to it, for all
of the blind alleys of mazes III and IV were highly uniform in
character. Each consisted of a single straight runway sixteen
or twenty inches long. This reduction of the difficulty of a
cul de sac to the animal's disposition toward it allows of a
common explanation for both the initial and final distribution
of errors. The disposition of an animal to enter or avoid an
alley is a result in part of the habits of turning already devel-
oped. As the maze is mastered, these habits become profoundly
modified. Returns and repeated explorations are inhibited and
several cul de sacs may be eliminated. Since the attractiveness
of a cul de sac is a function of the maze habit and this habit
is altered in the course of mastery, it is evident that the initial
and the final attractiveness of an alley must be to some extent
independent variables. With this conception the disposition to
enter an alley may actually increase as well as decrease, and
individual exceptions to the group attitude are possible.

Such an explanation is feasible and the conception possesses
some degree of a priori rationality, but any convincing proof of
the hypothesis is more difficult. Chance can not account for
the error curve of alley 5 in fig. 1, for the records of fifteen of
the sixteen animals conform to its general features. There has
been no objective change in the alley itself. Evidently it is
the attitude of the animals towards this situation that has been
altered. Nor will chance account for the distribution of errors
of any individual animal. One rat did not enter alley 5 during
the first five runs but made one entrance per trial for the re-
maining 45 runs. Another animal avoided this alley for eleven
runs, alternated between success and error for the next six
trials, and the error then became fixed. In another case this
cul de sac was avoided twice, entered once, avoided twice, and

158 HARVEY CARR

then invariably entered on all succeeding runs. The eighth
cul de sac was avoided six times, entered once, avoided once,
and then became fixed; in another case it was avoided for 28
trials, entered three times, avoided for five trials, and then
became fixed. The continual avoidance of an error for a number
of trials followed by invariable entrance can not be due to chance.
Chance may account for the first entrance but the sudden fixation
of the habit remains inexplicable. This type of behavior may
be readily explained by our hypothesis. This alley is so related
to the previous sections of the maze that the turning habits
necessitated by the latter dispose the animal to avoid it. But
these determining habits become altered with successive runs.
The avoiding disposition becomes weakened and finally sup-
planted by the opposite attitude. The alley is thus regularly
avoided for a while; there is indecision and alternation for a
short period; and this behavior is followed by sudden fixation
and invariable entrance. The fixation of this error is not due
to repeated entrances. In a sense this error was engrained in
the animal's organization before it was entered, because the
disposition is the outcome of the entire maze 4 habit developed
up to that time.

CONCLUSIONS

The temporal order in which the various cul de sacs are elim-
inated is roughly correlated sometimes with their spatial order
in reference to the food box and sometimes with their order of
proximity to the point of entrance.

The temporal order of mastery of the cul de sacs is invariably
correlated with their order representing the increasing number
of errors made in each. The ease or rapidity of mastery of
any cul de sac is thus inversely related to its degree of attrac-
tiveness as measured in terms of number of entrances. The
problem as to the order of elimination of the cul de sacs must
be explained in large part in terms of the distribution of errors.

The factors influencing the distribution of entrances among
the cul de sacs are the tendency to return, the character of the
motives actuating the animal, and peculiarities of the cul de sacs.

* This hypothesis is closely related to Peterson's conception developed in his recent
article, Completeness of Response as an Explanation Principle in Learning, Psych.
Rev., 23, 153-162.

DISTRIBUTION AND ELIMINATION OF ERRORS IN MAZE 159

The amount of returning will vary with the animal, the maze,
the stage of mastery, and the section of the maze.

The character of the motives will shift within a run as prog-
ress is made from section to section, and from one stage of
mastery to another.

The attractiveness of cul de sacs due to peculiarities of con-
struction or position must in part be interpreted in terms of
the animal's organization in reference to them. As a possible
hypothesis, it is suggested that these disposing conditions of the
organism are a function of the maze habit as developed up to
that time. With this conception the individuals of a group
may vary in their attitude toward a cul de sac and the attitude
of any individual may change as the maze is being mastered.

MOVING AND STILL LIGHTS AS STIMULI IN A

DISCRIMINATION EXPERIMENT WITH

WHITE RATS

CORA D. REEVES

The work here reported was done in the spring of 1912 under
the direction of Prof. J. B. Watson, by whom the problem was
suggested. As the work of others may be furthered I present
briefly results compiled at the close of the experiment.

Apparatus. — The apparatus consisted of two crayon boxes,
9.3 x 10.5 x 16 cm., blackened inside and out, and each with
a doorway cut in one end. To each, on the end opposite the
opening, was attached a narrow, vertical support with a hori-
zontal bar on the top of it. Suspended from this bar and imme-
diately above the door was a miniature tungsten, 2 c.p. lamp,
thus:

Figure 1. — Elevation of food-box

Rats tested. — The animals used were four male albino rats of
the same litter. They had learned the maze but were not yet
quite full grown.

160

DISCRIMINATION EXPERIMENT WITH WHITE RATS

161

Methods. — The boxes were placed upon the arc of a circle of
80 cm. radius near the end of a black-slate table of about a
meter width. The following diagram will show conditions:

Figure 2. — Plan of apparatus. 1, 2. Food-box. Bo. Boards six inches wide
which were placed on edge so as to pen off the area used in experimentation.

a, b. Chalk line along which boxes were shifted. -< indicates the position

where the rat was placed and direction of the head.

Boards were used to close off the rest of the table. The ex-
periment was carried on in a dark basement room and the black
boxes were on the side of the room and of the table away from
the light. Thus two dark objects against the dull background
of the walls rested upon the slate table. Above these were
the lights. The rats were kept in cages in the animal room and
brought in a small wire cage to the room where the experiments
were conducted. Before the rat was taken from the carrying
cage a bit of food (bread soaked in milk) was placed in a dish
in each box. A net was placed over the dish which was not to
be in the food box for the rat being tested. The food and net
were low in the dish and out of sight. This was to provide the
same sensory stimuli in all particulars except the one to be dis-
criminated. The light over the one box was started swinging

162 CORA D. REEVES

by hand, moving through an arc of more than three inches and
the light over the other box was left hanging still. The rat
was dropped upon the table at the center of the circle (see fig. 2)
or 80 cm. from each box. Care was taken to place the animal
with head in the direction indicated by the arrow. A record
of the time taken by each rat in reaching the box, and as to
whether the choice was right or left hand, was made. The
boxes were interchanged by chance. They were placed along
the arc of circle at all points and all distances apart. The same
box always had the light swinging. When the box had been
reached and entered the rat was allowed to feed if correct choice
had been made, and was then picked up and returned for another
test. Two rats, A and B, were fed at the box with the still
light; C and D at the one with the swinging light. The first
days, ten trials were made each day for each rat; then twenty
trials daily for the seven weeks the experiment continued.

Preliminary work. — The rats were put on the table and the
time taken for them to reach a single box without light for 10
trials each. During these preliminary tests no food was placed
in the boxes. The average time per trial was 32 seconds the
first day. The second day A went to the box ten times, but
B, C and D went to the box three, five, and five times respec-
tively, then quit running about before reaching the empty box.
The next two days food was put in the box; the average for
the four was 9 seconds the first day, and 4 seconds the second day.

The following two days the two boxes were placed one with
food and one without food. The average time for the rats to
reach the box was 5.5 seconds for the first day, and 3.1 seconds
the second day. ,

The following day the light was put above the food box and
not made to swing. The rats A, B, and D behaved as on days
when the light had not been there, at least, any hesitation if
present was not sufficient to change materially the average
time. The rat C, however, after two trials took a longer time
each trial, looking at the light before starting and then refused
to move when placed on the table.

The next day the light was started swinging over the one
box while left still over the other. The conditions of the ex-
periment had thus gradually been reached. The average time

DISCRIMINATION EXPERIMENT WITH WHITE RATS 163

per trial for A, B, and D on this day was 1.7 seconds, for C
4 seconds.

Time
or

TOO

1°

tc

Still

A

Moving

20

10

t*~ 2~> j" v* s*

/•*- IT* JW< ,< /* 6t ? t £ ,

100

« ,

5*^ y *> s* «* ;

* /*- /**- a^ jw /* j-x <*■ ;■

Figure 3. — Curves showing first responses of four white rats to stationary and

moving lights. Per cent of correct choices for each successive ten

trials. Average time in tenths of seconds.

164

CORA D. REEVES

The experiment. — Both A and B formed a rather persistent
habit of going to the left hand box so that the first day's records
of per cent of food choices have no significance. The days fol-
lowing, this habit was weakening and the correct choices in-
creased. The preceding curves, fig. 3, show that A was slow
in response as was also C, while B and D were habitually quicker
in response. The curves show also that there was learning
where the rats went to the moving stimulus, but with the others
less high percentages of correct choice were made. Some ten-
dency to decrease the average time for the day appeared.

On the eighth day, after two days of perfect record, the rat
C was terrorized by a stranger with a dog entering the labo-
ratory while this rat was on the table. An attempt to get
again two days' perfect records caused the work to be prolonged
for two months. After about two weeks (160 trials for each
rat) the number of trials per day was increased to 20. There
was an immediate increase in the average time per trial as shown
by the following curve, fig. 4, which is made up from the time
records of the four averaged for successive trials. The later
trials for each day were slower. Evidently though only small
portions of food were allowed each time the hunger stimulus
was lessened after a few successful trials.

Seconds

^vjeT^geJ of z.o fr'ial*

AuerAQet of ioo trials

Figure 4. — Curve showing time in seconds for response to stimulus for four white
rats while discriminating between still and moving lights during successive
trials.

DISCRIMINATION EXPERIMENT WITH WHITE RATS 165

The first part of the curve is made by averaging together the
averages of the successive 20 trials of each of the four rats; the
last part the averages are for 100 consecutive trials.

The increase of time toward the end of the experiment is not
easily accounted for. The age of the rats may be one factor.
They, however, were not old but were then only full grown.
The weather was warmer, and temperature and humidity may
be other factors. The rats A and C were from the first (shown
in fig 3) until the last days slower than B and D. For C the
time average for the last 80 trials was 30 seconds. There is,
then, no correlation between accuracy and speed of response.
The rat A was the only one that toward the close of the experi-
ment showed a reduction of the average time. Reference to
the table which follows will show that the per cent of correct
choices for this rat increases remarkably with the last 100 trials.

Table Showing Average Per Cent of Correct Choices in Discrimi-
nation of Stationary and Swinging Lights

1st 2nd 3rd 4th 5th 6th 7th

Food at stationary lights: 100 100 100 100 100 100 100

Rat A 40 40 61 57 69 69 90

RatB 57.5 52 53 64 58 63 78

Average 48.2 486 57 60 63.5 66 84

Food at swinging light :

RatC 76.5 66 68 71 76 86 86

RatD 55 68 58 68 68 77 87

Average 65.7 67 63 68.5 72.5 81 86

Averages 57 56.5 60 65 68 71 85

The slowness and difficulty of establishing this discrimination
is apparent from the table.

The error curves (fig. 5 on page 166) show the same slow pro-
cess of learning to discriminate and further emphasize the
tendency of the rats to go to the moving rather than the still light.
The use of only four individuals makes the results less certain
than would be the case had a larger number been used.

The records of rat A have frequent notes, as ' Watching,
moving back and forth, swinging or nodding while advancing."

This diagram of a path taken by A (fig. 6) shows a frequent
sort of behavior for this individual, which was being fed at the
still light. The starts toward the moving light were frequent
and were followed by a pause and a run toward the still light.
The swaying or nodding motion, rhythmic with the light, was

106

CORA D. REEVES

00

70
fco

F y TOT Qu T V t T o -r

E.TTOT Q^OTUt i»T

/ih X^- h* y*- ** X ^ 7 £ -/j*Ku^i <i J*4- i^J. iAj. i<- T 7 *- 4>*- f-ttLtMr

Figure 5.— I. Curve of error for rats A and B, fed at the stationary lights. II. Curve
of error for rats C and D, fed at the moving lights. These show the average
per cent of errors for each 100 trials.

Figure 6. — Path taken by rat A in reaching the stationary light.

DISCRIMINATION EXPERIMENT WITH WHITE RATS 167

observed in rats A, C and D. While C and D started toward
the still light at times, I have no records of paths of repeated
starts and halts with change of direction as in the case of A.

Testing results. — To determine whether some clue given by
the experimenter might not be the ground upon which the rats
discriminated, Dr. Karl S. Lashley kindly tested the rats in
my absence. There was an average of 80% of correct choices.
In order to test further the significance of the results the rats
which had been trained to go to the still light were placed as
usual but with both lights still. They went equally to either
box. A strange reaction occurred, for upon coming to the door-
way of the box whose lamp had been swinging, when the vibrisse
lightly touched the edges, rat A stopped, squealed, turned back
and went over to the box which had had the still light. Ap-
parently some sensation from contact dominated his reaction.
The same halt at the doorway occurred when the rat C was
presented with both lights swinging and he went to the box
which had had the still light. One day when the rats trained
to go to a swinging light were presented with both lights swinging
in the middle of the day's series, the rat C for the four tests
given, went to the box where accustomed' to feed, but when
the lamp which had been stationary was set swinging, while
the light previously made to swing was still, this rat went to
the box with the swinging light. D, when both lights were
swinging, went to either box. The per cent of choices of the
food box changed for this rat from 94 under standard conditions
to 40 when both lamps were in motion. In other words, they
went freely to the box which they had consistently avoided when
the light stimuli were reversed. This fact together with the
fact of the halt upon touching with the vibrisse the wrong box
seemed conclusive evidence that the lights and not some other
factor had been the effective one in making the choice when at
a distance. After this halt at the doorway I noted that the
small roughnesses of the edges of the doorway made by the saw
were, of course, not exactly alike. I suspected that by chance
or from attraction of the swinging lamp the rat C went to that
box often and was then able to track himself but the evidence
seems conclusive that the moving stimulus came to be the
stimulus depended upon in reaching the box and food.

168 CORA D. REEVES

Discussion of results. — The large number of trials (700) neces-
sary to establish this discrimination seems to indicate how small
a part the visual stimulus, has in the daily life of a rat. Had
it not been for the records of the first week and especially of
rat C (fig. 3) and the lack of any adequate explanation other
than ' ' learning to discriminate ' ' which would account for the
improvement both in choice and in time, the conclusions (after
each had had 200 trials) would have been that rats cannot dis-
criminate a moving and still light. (See table.) The rats came
in time to select the food box more accurately. In the fact that
the rats halted at the doorway of the boxes where they had
not been receiving food is an indication that they used other
criteria than the lights when they reached the food box. They
were, however, not able to select the right box when the con-
dition of movement of the lamps was changed. It seems pos-
sible that some laboratory failures to find that animals possess
as acute sensory mechanisms as have been popularly ascribed
to them may be from the fact that the problem presented was
not fitted to the animals tested. This incident will illustrate.
At the close of this series of experiments an old rat which had
been handled continually for some months was running about
the room when the writer chanced to be winding up a piece of
cord. As the end of the cord was drawn along the floor the rat
followed, patted the end, for some eighteen inches, as a kitten
would. This rat was tested several times with a cord but would
never repeat the behavior. The rat became familiar with a
new situation quickly but the stimulation afforded by a moving
object could not be doubted. The effectiveness of movement
in controlling reactions is shown in the difference in the curves
(fig. 5) and in the path of the rat as shown in fig. 6. The length-
ened average time when 20 trials per day were used instead of
10 confirms the evidence already presented that a large number
of daily repetitions is not the most advantageous method of
establishing a given discrimination.

CONCLUSIONS

1. Rats can and do discriminate a stationary from a moving
light.

2. Rats show some tendency to approach a moving rather
than a stationary light.

A NOTE ON THE INTERFERENCE OF VISUAL
HABITS IN THE WHITE RAT

BINNIE D. PEARCE
From the Psychological Laboratory of the University of Texas

INTRODUCTION

The following experiments were performed at the University
of Texas under the supervision of Professor W. S. Hunter,,
from January to June, 1916. This paper should be considered
as a continuation of researches made by him and should be
read in connection with his paper on " The Interference of
Auditory Habits in the White Rat." 1 The purpose of the
present tests was to obtain data showing the strength of habit
in the white rat by measuring the effect of a habit previously
acquired upon the formation of a new habit of opposite char-
acter. The stimuli in each case were lights. The results appear
to reinforce the conclusions reached by Dr. Hunter, viz., that
a habit acquired by training does persist in the new work and
may interfere tremendously with the formation of a dissimilar
habit. My detailed conclusions will be presented at the close
of the paper.

m

Ill till III* '1

111

IIIH

A

A'

— m —

Figure 1. — Ground plan of the apparatus

The apparatus used was a T-shaped discrimination box and
is shown in fig. 1. A mazda light was placed in a small box

Jour. Animal Behav., 1917, 7, 49-65

169

170 BINNIE D. PEARCE

behind the main apparatus. Between the boxes was an aper-
ture, O, covered with a piece of clear glass and a variable number
of sheets of typewriter paper (Post Office bond). These varia-
tions and the c.ps. employed will be given below. The light
was controlled with a switch at S. The rat was expected to
react to the presence or absence of light by turning to the left
or the right as the conditions of the experiment required and
as will be detailed later. When the reaction to the stimulus
was correct, the animal escaped through an open alley (A) to
food at F ; when incorrect, an electric shock was given "by means
of the wires marked E and E' and a free exit was blocked by
means of a movable end-stop, placed in the alley A'. At each
trial the rat was introduced directly into the discrimination box
through an opening at F and the stimulus was presented im-
mediately. Punishment and reward was used throughout the
test. The following series of presentations were used, 10 trials
daily :

llrlrrlrlr

r 1 1 r r r 1 1 r 1

r r 1 r 1 1 r 1 1 r

lrrlrrlrll

EXPERIMENTAL RESULTS
I

The present experiment was begun with four untrained rats
(adults). Later four new untrained rats were added. Of these
one (No. 10) was 42 days old; one (No. 13), 37; and two (Nos.
14 and 15) 68 days old. Unless otherwise stated the results
given are for all eight rats.

On each of 3 consecutive days the animals were allowed to
make 5 preliminary runs in the box, the object being to acquaint
them with the apparatus and to accustom them to receiving
their food at F. These trials were given without light stimulus,
punishment or end-stops, save that the latter were used where
necessary to prevent the appearance of position habits.

In the regular test, habit No. 1, a correct response required
the rat to turn through the right hand passage when light was
present and through the left hand passage when the light was
absent. The light used in this first test was a mazda 32 c.p:;

THE INTERFERENCE OF VISUAL HABITS 171

shaded by one thickness of ordinary writing paper as men-
tioned above.

Table 1 shows the number of trials required by the rats in
establishing the association. The standard of learning was as
follows: Each of the last four series of 10 trials must show at
least 8 correct reactions, but the average percentage of correct
reactions for the four series must not be less than 87§%. The
trials in table 1 include all given each rat up to the 40 made at
the standard percentage.

TABLE 1

Learning Habit No. 1

Rats Trials

1

170

2

180

3

80

4

190

10

300

13

220

14

60

15

120

A comparison of table 1 with similar data obtained by Dr.
Hunter in his experiments on the acquisition of auditory habits 2
is of value in showing a greater ease in the formation of visual
habits by the white rat. My rats ranged between 60 and 300
trials with an average of 152. Dr. Hunter's rats, — from the
same stock, working in the same apparatus on the same problem,
but using sound as a stimulus, — ranged between 210 and 710
trials with an average of 423. This is a matter of great import-
ance inasmuch as the explanation would appear to lie chiefly,
if not wholly, in the different sensory channels involved. I call
to mind no prior demonstration of this fact. Extended study,
which would go far beyond this preliminary work, would un-
doubtedly reveal important differences in vision and hearing so
far as the daily life of the rat is concerned.

As each rat learned the association, control series were intro-
duced as follows :

1. No light used; no punishment. Reaction considered

right if it fits the series.

2. An 8 c. p. mazda substituted for the standard light.

Punishment used.

2 Op. cit., table 1.

172

BINNIE D. PEARCE

The object of control 2 was to determine the similarity of the
8 and 32 c.p. stimuli in terms of response. A summary of the
control tests is given in table 2. The chronological order of
records has been preserved. The per cents represent correct
responses in a given daily series of 10 trials. The low percent-
ages made with control 1 indicate the rats' dependence upon
the light as a determining stimulus. The high percentages made
with control 2 indicate that the rats sensed the light and that
it meant to turn to the right in order to secure food. The ex-
ceptions to this statement are shown in the table.

TABLE 2

Controls Used With Habit No. 1

Rats

Control 1 2 3 4 10 13 14 15

Control 1 50% 80% 40% 50% 60% 60% 50% 50%

Control 1 60 '

Normal 90 100 90 100 90 100 90 90

Normal 90

Control 1 50 . . 60 50 20 50 40 70

Normal 100 .. 90 90 60 100 90 90

Normal 90

Control 2 50 90 70 80 80 100 70 100

Control 2 80

Normal 60 100 90 90 90 100 70 100

Normal 100

Control 2 50 60 70 70 70 90 40 80

II

Training on the second habit was instituted in the case of
each rat as soon as the results of the first test had been analysed
by controls as shown above. The second habit furnished a
problem the opposite of habit No. 1. Its purpose was to train
the rats to associate turning to the left with the presence of
light and to the right for the absence of light. The 8 c.p. light
of the control tests was the stimulus here. At first it was
shaded by three thicknesses of the writing paper. But when
rat No. 3, the first rat tested on habit 2, persisted in reacting
to this stimulus as he did to the absence of light, I removed one
thickness. The purpose was to secure a light which would be
treated the same as the standard light and yet which should

It

u

u

a u

a

a

a

a «

tt

a

a

« «

«

«

a

u u

a

u

a

u «

THE INTERFERENCE OF VISUAL HABITS 173

be as different in intensity as possible from the standard. 3 The
situation is summarized in table 3. The first reactions of the
rat to the twice-shaded, 8 c.p. light were made as though this
light were the standard light of habit 1. (Reactions to the
normal stimulus should give at least 80% correct. Reactions
to darkness would all be made to the left and so would give
50% correct. Since the new series, habit 2, was the reverse
of the " normal " series, when the rat treated the stimulus as
though it were the normal standard light, he should make not
more than 20% correct.) I now knew that the once-shaded

TABLE 3
Test Correct in 10

Normal (32 c.p., once shaded) 9

" 5

3

5

5

5

4

New series (8 c.p., twice shaded) 1

32 c.p. and the twice-shaded 8 c.p. would initiate the same
responses. Furthermore, there was reason to assume, both from
the behavior just cited and from the work of other investigators,
that the lights were dissimilar enough in intensity (one being
almost equivalent to darkness) that they could be readily dis-
criminated by a rat when tested in the conventional discrimi-
nation box.

In this second test every effort was made to keep the conditions
identical with the first test save in the matter of light stimulus
and direction of turning. The results are very striking. In the
first test, the least number of trials given any rat was 60 and the
greatest 300. In the second test, one rat learned in 420 trials.
The other seven rats never completely learned the association, — ■
the trials given were 680, 760, 850, 1080, 1080, 1090, 1090, and
1160. At the close of the work these rats were improving so,
that it seems probable that they would have mastered the
problem had the training been more extended. The results of
this test are summarized in table 3, The length of the training
periods here as opposed to the learning periods with habit No. 1

3 It would be of value and interest to have animals form habit No. 2 with the
same stimulus used in habit No. 1. My choice of stimulus for the second habit
was guided by a desire to secure a procedure similar to that followed by Dr. Hunter.

174 BINNIE D. PEARCE

TABLE 4
Correct Reactions in Each Succeeding 50. Habit No. 2

Rats

Trials

1

2

3

4

10

13

14

15

50

13

5

23

13

14

9

21

12

100

18

16

17

7

11

13

17

16

150

16

20

10

13

12

12

15

27

200

13

20

18

17

16

14

20

17

250

20

17

27

18

17

9

21

30

300

12

15

23

24

18

13

21

34

350

14

19

25

14

17

15

31

31

400

10

20

26

18

18

16

32

43

450

15

20

25

16

23

26

32

15 of 20

500

17

18

18

19

25

25

30

550

19

19

25

21

28

27

41

600

15

10

22

23

26

29

41

650

17

16

22

26

21

33

34

700

17

21

24

25

22

27

27 of 30

750

22

17

28

31

27

26

800

29

26

32

28

3

26

850

22

24

29

30

of

24

900

22

32

30

28

10

950

25

27

30

41

1000

27

35

32

33

1050

30

41

27

33

1100

26

20

29

24

1150

of

of

31

of

1160

40

30

7
of
10

40

are not to be explained by variations in age (which were too
small) or in experimental conditions. The essential factor is the
interference of habit No. 1 with the formation of habit No. 2.

The following is a brief examination of representative data
secured on habit No. 2. Rat No. 3 had acquired the first habit
with great facility after 80 trials. After 1160 trials on the
second association he was making only 31 correct reactions out
of a possible 50. This rat when set upon the new problem was
in perfect condition and had shown no tendency to untoward
timidity or the formation of position habits. When presented
with the second problem, he at first reacted to the stimuli (8-
c.p. light for turning to left; darkness for turning right) as if
they had been the former stimuli (32-c.p. light for turning right
and darkness for turning left). Upon punishment he imme-
diately set up position habits from which he could be forced only
with difficulty and into which he fell again and again. Several
times he slowly approached the standard of learning; but when

THE INTERFERENCE OF VISUAL HABITS 175

he seemed about to attain it, the position habit would again
appear. This conduct was characteristic of all rats, save that
rat No. 15 did master the problem. This error-behavior need
not be regarded in its entirety as an interference phenomenon,
because it occurs in the course of all difficult problems. How-
ever, it is to be remembered that the present discrimination of
light from darkness is not a difficult problem. Table 5 gives
sample records illustrating the above factual statements concern-
ing position habits.

TABLE 5
Diary Records Showing Fluctuating Behavior in Learning Habit No. 2

Rat No. 4

April 27 8

28 9

29 5

30 5

May 1 5

2 6

15-22 8,8,8,9,6,6,7,5

Rat No. 14
May 16-28 8, 9, 8, 8, 8, 7, 7, 6, 10, 9, 9, 7, 4

III

Rat. No 15 was the only one who mastered habit No. 2. This
animal was 68 days old when first tested. He acquired habit
No. 1 in 120 trials (tables 1 and 2), was put through the controls
and immediately started upon habit No. 2. This was mastered
in 420 trials (table 4). A control similar to control 1, used in
analyzing habit No. 1, was instituted and proved that No. 15
was reacting to the stimulus (light) presented.

A third problem was then set No. 15, — a test in retention.
The rat was put back on habit No. 1, the operator again using
as stimulus the 32-c.p. light (shaded as before in habit 1). The
rat was tested for 15 days, 10 trials daily. Habit No. 2 per-
sisted and interfered with the training on habit No. 1 so that
the percentage of correct reactions never exceeded 50 for any
10 trials. By the close of the 150 trials a position habit of
always going to the left had fixed itself upon the rat with such
tenacity that tests were discontinued.

I have plotted three curves, fig. 2, which present graphically
the learning processes detailed above. The curves are con-

176

BINNIE D. PEARCE

structed as follows: The total number of trials given a rat
prior to the 40 made at the standard is divided into 10 parts.
The percentage of correct reactions in each one-tenth is then
computed and an average for all rats taken. The resulting
curve shows the progress of error elimination independently of
the absolute number of trials and is thus representative through-
out its length. N indicates the records during the 40 trials
made at the standard percentage.

%

/ / 1

7 o

y r \j></~

//
ft

\ /

40

N. y^ 1 yr

fo

/ ' _—--- —

1 f

a/ / \

1 /
1 /
1 /

/

ft
to

i

-to

\/

•

10

•

1

/o

7i

Figure 2. — Curves of learning. A — record for all rats on habit No. 1; B — record
for all rats but No. 15 on habit No. 2; C — record for rat No. 15 on habit No. 2;
D — record for Hunter's rats on learning a first habit in audition.

Curve A shows the succeeding average percentages of correct
reactions for all the rats during the formation of habit 1. It is
essentially the same as curve D. Curve D is taken from Dr.
Hunter's ' Auditory Interference ' study, table 7. It is a
normal curve of four rats on sound. The practical identity of

THE INTERFERENCE OF VISUAL HABITS 177

the two curves is interesting in view of the different absolute
lengths of time involved in the formation of the two types of
habits.

Curve B presents the results for habit 2 for all rats save No.
15. It is a curve of progress and not of completed learning.
Curve C is the record for rat 15 on habit 2. The curve is irreg-
ular in its first part. Curves A, B, and C indicate that so far
as rate of error elimination is concerned, habit interference has
been most prominent in the first six-tenths of the curves. Dr.
Hunter's results also indicated that the six-tenths point was a
turning point in the relearning process. The situation may be
but a coincidence, however.

CONCLUSIONS

1. A simple visual habit of the type here studied will inter-
fere tremendously with the formation of another visual habit
of opposite kind. This interference may practically prevent the
formation of the second habit.

2. A comparison of our data and Dr. Hunter's indicates that
the white rat learns visual habits much more readily than similar
auditory habits.

MODIFIABILITY OF THE PREFERENTIAL USE OF
THE HANDS IN THE RHESUS MONKEY

K. S. LASHLEY

Government Hospital for (he Insane

Observations on handedness in monkeys and apes has not,
in general, given evidence for the predominant use of either
hand which might be ascribed to heredity and in this respect
has not provided phylogenetic support for the view that handed-
ness in man is instinctive. Pfungst ('12) reported that in obser-
vations on over sixty individuals of different genera the instances
of predominant use of either hand could almost always be traced
to training or to previous trauma. He has not, however, reported
the details of his work. Franz ('13) found that of six monkeys
(Macacus rhesus) two were ambidextrous, three used the left
hand more frequently than the right, and one was probably
right-handed, but doubtful because of the small number of
observations which could be made. Lashley and Watson ('13),
studying a very young monkey, were unable to demonstrate the
predominant use of either hand. Yerkes ('16) made tests upon
seven apes, four of which were found to be predominantly left-
handed, one right-handed, and two probably ambidextrous.

Except in the work of Franz not enough observations of any
one animal have been reported to demonstrate such a persistent
predominant use of one hand in a variety of activities as is neces-
sary for a comparison of the condition in the animal with handed-
ness in man. During the past summer Dr. Franz placed two
young male rhesus monkeys at my disposal, suggesting a con-
tinuation of the work upon the preferential use of the hands
with a view to the permanent modification of the normal con-
dition by training. 1

Observations were made first upon the preferential use of the
hands in a number of different situations to test the effects of

1 These animals were purchased with a fund granted by the Carnegie Institution
of Washington to which acknowledgment is here made.

178

USE OF THE HANDS IN THE RHESUS MONKEY 179

the immediate environment upon the reactions. An attempt
was then made to alter the proportionate use of the hands.
As the animals were to be used in other experiments the time
available for this work was limited and it was not possible to
undertake a permanent modification of their activity: a success-
ful suppression of such an instinct would perhaps require a year
or more. Since the complete alteration of an instinctive prefer-
ence would necessitate the production of an altered predomi-
nance in the use of the hands under conditions in which train-
ing had not been given, a question of primary importance for
the problem is the possibility of transfer of training from one
situation to another, and the time available was sufficient for
a limited study of this. The observations upon the animals'
behavior in different situations formed a basis for a test of such
transfer, requiring that the training be carried out in only a
part of the situations.

The observations were made upon the animals, in part while
they were in the cages described earlier by Franz ('13), in part
while they were fastened by a strap one meter in length to a
swivel snap in the floor of the room where the monkeys were
kept. In the tests upon the use of the hands in taking food
the following seven conditions were used.

Animal fastened to the floor with the strap. —

1. Picking up food from the floor. The food was dropped
as nearly as possible in front of the animals and within easy
reaching distance.

2. Taking food from the experimenter's right hand. The
food was held directly in front of the animal and about 10 inches
from him, so that he could reach it equally well with either hand.

3. Taking food from the experimenter's left hand. The con-
ditions were otherwise as in situation 2.

Animal in the cage. —

4. Picking up food from the floor of the cage.

5. Taking food from the experimenter's right hand. His
fingers, holding the food, were thrust through the coarse wire
netting of the front of the cage so that the monkeys did not
need to reach out of the cage to get the food.

6. Taking food from the experimenter's left hand. Condi-
tions were otherwise as in 5,

180 K. S. LASHLEY

7. Taking food from a small table placed in front of the cage
and about six inches above its floor. The animals had to reach
through the meshes of the netting to get the food in this situation.

Except when holding out food in his hands the experimenter
remained at a distance of about four feet from the animals during
the observations. The animals were rather wild at first and
about two weeks were spent in handling them and accustoming
them to the experimenter before the observations on the pre-
ferential use of the hands were begun. The two were markedly
different in temperament. The smaller (No. 1) was gentle and
almost fearless, rarely hesitating to take food from the experi-
menter's hands. The larger (No. 2), on the contrary, was
exceedingly wild and fierce, constantly trying to break his strap,
when out of the cage, and striking and biting at the experimenter
when food was offered.

The tests were made at irregular times but several of the situa-
tions were included in each day's observations in order to con-
trol temporal variations in the use of the hands. With few
exceptions the results of observations on different days are in
agreement so that we may be sure that they are not due to a
temporary injury of one or other hand.

PREFERENTIAL USE OF THE HANDS WITHOUT TRAINING

The number of times which the right and left hands were
used by the two animals in each of the seven situations is given
in table 1. There is a considerable amount of variation, de-
pending upon the different environments in which the animals
were placed, but the behavior of each animal is fairly character-
istic throughout. No. 2 was obviously right-handed, using the
right hand almost four times as frequently as the left. No." 1
showed a less marked preference for the use of either hand and
a much greater adaptability to the different situations. In
picking up food from the floor of the cage, where he was least
influenced by the presence of the experimenter, he used his
left hand in over 90% of the cases observed; a fact which indi-
cates a decided left-handedness. Where food was offered in the
experimenter's hands, however, he reached for it in 81% of the
trials with the hand homologous to that used by the experimenter.
Under like conditions No. 2 was not influenced by the hands

USE OF THE HANDS IN THE RHESUS MONKEY

181

of the experimenter, using the homologous hand in only 55%

of the cases.

TABLE 1

Preferential Use of the Hands Under Different Environmental
Conditions in Two Rhesus Monkeys

Conditions of Test

Hand used .

Monkey in cage, taking food

from —
a — experimenter's right hand . . .

b — experimenter's left hand

c — floor of cage

d — table before cage

Monkey on floor outside of cage,

taking food from —
e — experimenter's right hand . . .

f — experimenter's left hand

g — floor of room

All trials

Monkey No. 1

Number

of cases

in which

each hand

was used

R

146
46
23
59

386

3

133

796

102

155

243

41

47
164
192

944

Per cent

of cases

in which

each hand

was used

R

59.7

22.0

8.6

59.0

89.1

1.7

40.9

45.7

40.3
78.0
91.4
41.0

10.9
98.3
59.1

54.3

Monkey No. 2

Number

of cases

in which

each hand

was used

R

264
174
120
200

158
60
30

1006

70

42

57

4

40
46

267

Percent

of cases

in which

each andh

was used

R

79.1
80.5
67.2
98.0

95.2
60.0
39.4

79.4

20.9

19.5

32.8

2.0

4.8
40.0
60.6

20.6

A like predominance in the use of the hands was found in
reactions of defense. With the monkeys fastened in the middle
of the floor the experimenter touched them lightly on top of
the head with the tips of the ringers of his right hand. The
monkeys reacted to this by striking at his hand or by attempting
to grasp and bite it. The number of times in which each of the
two hands was used in warding off the touch is shown in table 2.
Number 2 was as markedly right-handed in this as in his food-
taking reactions. No. 1 again appeared to be left-handed, but
only to a slight extent.

The predominant use of the hands seemed associated with the
positions assumed by the monkeys when kept alone in different
cages. They ordinarily sat in one of the rear corners of the
cage and the corner chosen was usually that which gave the

182

K. S. LASHLEY

greatest freedom to the hand most frequently used. Thus No. 1
sat with his right side near the side of the cage, leaving his left
arm free, and No. 2 kept his right arm and side away from the
side of the cage. Interchanging the monkeys in their cages did
not seem to interfere with this custom, which seems therefore
independent of the relation of the cages to other objects.

Marked variations in the use of the hands from day to day
were noted. Not more than 25 trials in any one situation were
given on a single day and in the majority of cases one or the
other hand w T as used exclusively in each series of trials unless
some disturbing element, such as the alternate use of the ex-
perimenter's hands, intervened.

TABLE 2

The Preferential Use of the Hands in Defence from Attack by

the Experimenter

Hand used to ward off blow

Right

Left

Both

Monkey No. 1 :

Number of cases

Per cent of cases. .

70
34.2

129
76.8

122
59~5

33

19.7

13
6 3

Monkey No. 2:

Number of cases

Per cent of cases

6
3.5

TRANSFER OF THE EFFECTS OF TRAINING

When a sufficient number of observations had been obtained
to make the relative use of the hands in each situation fairly
certain, training for the exclusive use of one hand was begun.
An attempt was made to force both of the animals, when leashed
to the floor of the room, to use only the left hand in taking
food from the experimenter's right hand. The method of train-
ing was to withhold the food and to grasp or strike at the mon-
key's right hand whenever he extended it to take the food.
No. 2 did not take kindly to this training. He would snatch
at the food with his right hand, getting it as often as not, and
when punished would refuse to reach again. After about 100

USE OF THE HANDS IN THE RHESUS MONKEY

183

trials he could no longer be induced to make any attempt to get
the food and further training had to be abandoned.

Trials

400

Figure 1. — -The course of modification of the use of the right hand by training in
the case of monkey No. 1. The ordinates represent the number of times
that the right hand was used in each successive ten trials, the abscissae, the

number of trials grouped by tens. ■ — first period of training.

second period.

No. 1 learned to inhibit the use of his right hand very quickly.
After nine successive failures to get the food with his right hand
he began to use his left predominantly and only 34 failures to
get the food were required to abolish completely the use of the
right hand in this one situation. The form of the learning curve
is shown in fig. 1. After the first hundred trials the left hand
was used in all but 2% of his efforts to get the food.

When the habit seemed firmly established (after 500 trials)
the preferential use of the hands in the remaining six situations
was tested again. The data obtained are given in table 3.
In the three situations outside of the cage the training resulted
in a decrease in the use of the right hand. This is not very
certain where the animal took food from the experimenter's
left hand, since there was little room for improvement to begin
with, but it is unmistakable where food was picked up from the
floor. In contrast to this the training seems to have had little
effect in modifying the use of the hands when the animal was
in the cage, except in the last case, that of taking food from
the table. The change here from 59.0 to 4.4% in the use of
the right hand is significant.

184

K. S. LASHLEY
TABLE 3

The Effect of Training in the Use of the Left Hand in One Situation
Upon the Reactions in Other Situations. Monkey No. 1

Before training

After training

Number

Per cent

Number

Per cent

Conditions of Test

of cases

of cases

of cases

of cases

in which

in which

in which

in which

each hand

each hand

each hand

each hand

was used

was used

was used

was used

Hand used

R

L

R

L

R

L

R

L

Animal on floor outside of cage;

taking food from —

a — experimenter's right hand . . .

386

47

89.1

10.9

180

0.0

100.0

(Training in this situation)

b — experimenter's left hand

3

164

1.7

98.3

95

0.0

100.0

c — floor of room

133

192

40.9

59.1

11

141

7.2

92.8

Animal in cage; taking food

rom —

d — experimenter's right hand . . .

146

102

59.7

40.3

45

35

56.2

43.8

e — experimenter's left hand

46

155

22.0

78.0

14

62

18.0

82.0

f — floor of cage

23

243

8.6

91.4

14

59

19.0

81.0

g — table before cage

59

41

59.0

41.0

5

98

4.4

95.6

At the end of these tests training for the exclusive use of
the right hand was instituted. The situation and method were
the same as in the preceding training, with the exception that
food was now withheld from the animal's left hand. After a
very few trials he began to use his right hand and a total of
only 40 failures was required to bring about the exclusive use
of the right hand. The course of learning is shown by the dotted
line in fig. 1 . Tests for transfer of this training could be made
only for the situation of taking food from the experimenter's
left hand. The transfer was evidently complete for this con-
dition, the animal using his right hand in 49 out of V S0 trials.

DISCUSSION

With clear evidence for a transfer of training in some situa-
tions and not in others, we must ask the reason for the selec-
tion. The one character in common to the situations in which

USE OF THE HANDS IN THE RHESUS MONKEY 185

transfer occurred and differentiating them from the others was
the presence of the wire netting of the cage between the experi-
menter and the animal. When on the leash the monkey was
directly exposed to attack and in taking food from the table
in front of the cage he had to put his hands between the wires
of the front of the cage and so expose them to attack. The
stimuli from the presence of the wire netting between the mon-
key and the experimenter seem, then, to have formed the limit-
ing condition for the transfer of training.

An attempt at a further analysis of the role of these stimuli
in determining the transfer is hardly justified by our present
knowledge. In their cages the animals frequently give more
threatening reactions than when held by a leash, so it may be
that a certain physiological tone or emotional reaction, common
to a part of the situations, reinforced the habit and led to its
transfer to all of them in which it was present. The transfer
might also be looked upon as the result of an analysis of the
situations in ideational terms, but this and the concept of physio-
logical tone are, themselves, so badly in need of experimental
analysis as to amount to nothing more than a restatement of
the problem when applied as explanatory principles. We can
say safely only that the habit of using the left hand was condi-
tioned by a complex group of stimuli.

The ease with which the predominance in the use of the
hands may be altered by training will make it very difficult
to establish the existence of any hereditary predominance. We
never know the complete history of an animal or can exclude
the possibility of a severe trauma which might condition the
use of one or the other hand. Only the observation of the
predominant use of one hand in a wide variety of situations
and in situations entirely new to the animal could furnish reliable
evidence for an hereditary predominance.

The possibility that the use of the hands by human infants
may be equally easily modified by training makes the existing
data upon handedness in young children of very doubtful value.
Mrs. Woolley ('10) believed that the predominant use of the
left hand by the infant which she studied was the result of the

186 K. S. LASHLEY

carrying position, 2 but experimental evidence upon the acquire-
ment of such habits by young children is lacking.

SUMMARY

1. The rhesus monkey, as has been shown by Franz, may be
right or left-handed or may use the hands indifferently.

2. Immediate adaptation in the preferential use of the hands
in different situations may appear.

3. Where there is no decided preference, the use of the hands
may be modified very easily for a given situation by training.

4. There is a transfer of training to new situations, but these
are selected on the basis of complex stimuli.

LITERATURE CITED

Franz, S. I. Observations on the Preferential Use of the Right and Left Hands

1913. by Monkeys. Jour. Animal Behav., 3, 140-144.
Lashley, K. S. and Watson, J. B. Notes on the Development of a Young Mon-

1913. key. Jour. Animal Behav., 3, 114-139.
Meyer, M. Left-handedness and Right-handedness in Infancy. Psvcli. Bull.,

1913. 10, 52-53.
Pfungst, O. Zur Psychologie der Affen. Ber. u. d. V. Kongressf. exper. Psychol.

1912. 200-205. Leipzig.
Woolley, H. T. The Development of Right-handedness in a Normal Infant.

1910. Psych. Rev., 17, 37-41.
Yerkes, R. M. The Mental Life of Monkeys and Apes: a study of ideational

1916. behavior. Behavior Monograph, vol. 3, no. 12, iv+145 pp.

2 Max Meyer ('13) has criticized this interpretation upon inadequate grounds.
He says concerning it, " If our ancestral inheritance could be so easily modified as
Mrs. Woolley supposes, what an incentive this would be to enthusiastic educators! "
Here he evidently overlooks the fact that the great mass of habits formed by in-
fants and young children (the social inhibitions) are just such modifications of
instinctive behavor and that the modification is a process both rapid and produc-
tive of enduring results.

DIURNAL ACTIVITY OF THE EARTHWORM

FRANCIS MARSH BALDWIN

University of Illinois

This note 1 records a series of observations made to determine
whether the earthworm gives evidence of fixed periods of activity
and of quiescence. Under the conditions of our observations,
activity included (1) crawling movements either in or out of
the artificial burrow, (2) feeding and the acquisition of food,
and (3) the ejection of waste products from the body. Local
contraction of segments was not regarded.

Apparatus. — Two parallel glass plates (12 in. x 20 in.), set 3-8
in. apart, were mounted in an upright position on a base
15 in. square covered at the lateral edges and filled in with

1 A summary of work done some time ago in the biological laboratory of Clark
University, under the guidance of Professor C. F. Hodge. So far as the writer is
aware, the work has not since been duplicated with this form. The author's thanks
are due to Professor Bentley for suggestions in the preparation of the summary.

187

1SS

FRANCIS MARSH BALDWIN

moist earth. Strips of black cloth were then tacked to a frame
around the margin of the board, making a dark box to include
the plates. Holes were cut in the cloth on either side for obser-
vation, a flap of cloth serving as a cover to the hole when not
in use. Within were placed four 1-c.p. electric bulbs (two on
either side of the glass) which were controlled by switches.

Large subjects of Lumbricus terrestris were used throughout
the observations (four in the first series, and five in all the others) .
They were placed, one at a time, upon the surface above the
plates and allowed to burrow into the soil beneath (between the
plates), where their subsequent activity could be followed. The
soil was kept moist by occasional sprinkling.

In preliminary observations, records were taken at 20-minute
intervals, at various times of day and night throughout the period
of one month. From data thus accumulated, the activity of
the four subjects is summarized (in minutes) in table I, which
indicates that the worms were active about one-third of the
entire time with a fairly wide individual variation.

TABLE I

Subject

Total

time

observed

Total

time

active

Variations

from average

active time

I

18,060
18,070
16,200
16,210

7,740
6,000
3,840
4,080

2,324

II

Ill

584
1,576

IV

1,336

Averages

17,135

5,416

1,455

Records of the activity of five subjects were subsequently
made on two different days (about a week apart), the obser-
vations being continuous for each of twenty-four hours. Table
II displays the results for the two days (a and b), giving "active"
times in minutes, by quarter-days.

Although moderate individual differences in amount of activity
again appear, the average active time, when compared with the
total time of observation (1,440 min. for the whole day) closely
approximates that shown in table I. It is further to be noted
that the period of greatest activity, with but two exceptions
(I-a and Ill-b), falls within the hours 6-12 P. M.

DIURNAL ACTIVITY OF THE EARTHWORM
TABLE II

189

Subject

Days

A.

M.

P.

M.

Total
active
times

Variations
from ave.
time (373)

12-6

6-12

12-6

6-12

I

a
b

80
48

24
45

30
207

252

147

386

447

13

74

II

a
b

54
72

126
48

12
42

134
195

326
357

47
16

Ill

a
b

54
177

12
32

03
140

218
126

287

475

92

102

IV

a
b

60
66

54
108

18
60

186
134

318

368

55
05

V

a
b

'54

52

102

30
120

270
137

352
413

21
40

k

Averages

372.9

46.5

TABLE III

•

Subject

Days

A. M.

P.

M.

• Total
active
times

Variations
from ave.
time (259)

8-12

12-6

6-12

I

a
b

40
10

54
54

236
120

330
184

71
75

II

a

b

35

24

78
65

274
115

387
204

128
55

Ill

a
b

30

70

36
70

73
145

139
285

120
26

IV

a
b

18

19

69
60

138
165

225
244

34
15

V

a

b

12
60

45
50

294
135

351
245

92

14

Averages

259.4

63.0

In a similar manner, the activity of five other subjects on
parts of two days (a and b) is shown in table III. In this
series the period 12-8 A. M. was not included in the record.

Again the ratio of the average active time to the total tirrie
observed is, in this case, very close to that given in table II

190

ERANCIS MARSH BALDWIN
TABLE IV

Subj .

Days

A.M.

P. M.

Total active times

Variations from
average times
(211, 364, 156)

8-12

12-6 .6-12

I

a
b
c

54
78

144 66
66 168
11 126

264

312

137

53

52

19

II

a
b
c

84

60 65

144 156

48 132

125

384

180

86

20

24

III

a
b
c

60
104

138 ' 72

140 174

48 78

270

418

126

59

54

30

IV

a
b
c

72

78

84 48

110 162

24 132

204

350

156

07

14

00

V

a
b
c

98
98

66 30

104 156

42 138

194

358

180

17

06

24

Aves . .

211.4

364.4

155.8

44.2

29.2

19.4

(approximately 1 :4) , and again the period of greatest activity
of the subjects is found to lie between the hours 6-12 P. M.

To determine the influence of food placed at the mouth of the
burrow on the activity of the worms three series of observations
were made and the results summarized in table IV. The letters
under ' Days ' refer to the records of three successive days,
before (a), while (b), and after (c) food was placed at the mouth
of the burrow.

It appears from the above table that the presence of food at
the mouth of the burrow tends to increase the activity of the
worms, as indicated in the record of b. Activity is least upon
the day (c) following its removal.

These observations lead us to conclude that (1) in spite of
individual differences the total active time, when compared to
the total time under observation, is fairly constant, that (2) the
earthworm has definite active periods, confined, for the most
part, to hours of the night, especially to the early hours of the
# night, and that (3) food has a decided effect upon activity.

JOURNAL OF ANIMAL BEHAVIOR

Vol. 7 JULY-AUGUST No. 4

THE FEEDING OF NESTLING BIRDS *

T. C. STEPHENS

Biological Laboratory of Morningside College, and the
Iowa Lakeside Laboratory

As a basis for the discussion which will follow I wish to
inquire into the validity of field study as a method of developing
our knowledge of bird-life. In a broad sense, by the term
' field study ' ' we mean the study of the living animal in its
native habitat, under normal conditions; or under abnormal
conditions which are known and determined.

It seems to me that it must be accepted as axiomatic that
we have no other method of approach to certain problems of
behavior. Even if this is true, however, it does not follow that
the results obtained by such method are any the less subject
to scrutiny and criticism. We must demand of the field zoologist
the same accuracy of observation, the same careful exclusion
of uncertainties, and the same rigid logic that we require of
the worker in the laboratory. The observational and experi-
mental methods are open to the one who will attempt to solve
the problems of nature in the field. The laboratory worker
relies upon the same methods.

I would contend, therefore, that the reliability of field studies
stands in direct ratio to the scientific spirit of the investigator.
The value of a report based upon field studies cannot be judged
apart from the personal equation; and is this not true of every
kind of investigation? Do we not require that research carried

* Address of retiring president of the Wilson Ornithological Club at its third
annual meeting, at Columbus, O., December 29, 1915.

191

192 T. C. STEPHENS

out in the laboratory be subject to verification? And is not the
knowledge derived in the field by the process of observation or
experimentation also open to verification?

The successful pursuit of scientific knowledge depends upon
the investigator's ability to observe and interpret; which, in
turn, is largely a matter of training. Whether the observer
sees through the lens of a microscope, the field glass, or with
the naked eye, matters little. He is just as liable to err by
the one as by the other, unless he is trained to recognize the
avenues of error. Various technical equipment may be called
to the aid of the observer, but, fundamentally, credibility is
determined by the student's capability and honesty.

A given method of study must be appropriately used, and
its limitations recognized. The microscope is to enlarge an
object which is too small for direct observation. The field
glass, in effect, is to bring the object closer. Both are for the
purpose of rendering vision more distinct. It seems, therefore,
that the field glass is as legitimate an instrument for research
as the microscope. Its usefulness, however, is in the field.
The extent to which it may be applied is a detail. In the
identification of birds its use may be legitimate when the species
possesses marks by which the identification may be thus deter-
mined. When specific differentiation is based upon minor
variations in dimensions, the field glass becomes inadequate,
and it must be so recognized. On the other hand, is there not
also a danger that the taxonomist may go too far with his
millimeter rule in the differentiation of species and subspecies?

It may be admissible to again raise the question whether,
in the long run, science is advanced by the excessive multiplica-
tion of named forms which are based upon such minute structural
details that the methods at the disposal of the field student
become inadequate.

The tendency toward species ' ' splitting ' ' under the technical
term of ' revision ' ' does not seem to be waning, and its bear-
ing upon the field zoologist is so pertinent that the mention
here may not be out of place.

Those ornithologists who are interested in the subject from
the practical, or economic standpoint are concerned chiefly with
the various problems which grow out of the question of the food
of different species. Usually, the end in view, from this stand-
point, is to determine the relation of birds, and of any given

THE FEEDING OF NESTLING BIRDS 193

species, to man and to other animals in which man is economically
concerned. While all of this has also a scientific value, yet the
horizon taken in from the viewpoint of pure science has a vastly
wider field than that which is involved in an economic study.

Without attempting to scan the entire horizon we may at
once select the limited field into which we now wish to inquire.
One of the subjects which has received a relatively small amount
of attention is concerning the behavior of birds, and its inter-
pretation; and particularly the behavior of birds during the
breeding period, about which there is so much to learn. It
is true that within recent years much more attention is being
given to this phase of ornithology ; but notwithstanding a rapidly
growing literature along this line, on many of the problems
there is not yet sufficient knowledge to enable us to safely
generalize.

While other causes may have played a part in the slow
development of our knowledge of the behavior of birds, it seems
not unlikely that it is to be attributed, at least in part, to some
prejudice against the reliability of field observations.

There are some problems which may be checked by laboratory
technic, and in such cases our knowledge cannot be complete
until such a check is applied. But there are some questions,
which from their nature, can only be studied in the field (with
respect to many species, at least). For example, there are many
matters of detail about the mating of the sexes, the building
of the nest, and the care of the young, etc., which cannot well
be examined in the laboratory.

So when we consider the question of the feeding of nestling
birds it seems that our most important method of study, and
perhaps the only method in some cases, is to watch the process
in nature and record the observations.

I will now review, insofar as my study of the subject permits,
the facts which have been ascertained concerning the manner
in which young birds are fed.

A considerable number of birds, among which, of course, are
the Limicolae, are precocious, and forage for themselves from
3he beginning. Smith's paper on the Spotted Sandpiper (l) 1
reported the feeding behavior of sandpiper chicks for as early
a period as I know of.

1 Numbers in parenthesis refer to bibliography, and page citations will sometimes
be given also.

194 T. C. STEPHENS

In this instance the young, which hatched during the night,
were under observation from daylight of the morning of hatching,
and the parents were not observed to carry any food to the
young. On the other hand the young left the nest within five
or six hours to feed for themselves.

If we now direct our attention to those birds which feed
their young for a period of time we find considerable range of
behavior. As a preliminary analysis of the subject we may
distinguish three modes of feeding the young, viz.;

(a) Where the young bird thrusts its bill, or entire head,
into the mouth or throat of the parent; or where the parent
regurgitates the food, in a more or less digested condition, upon
the ground, to be picked up by the young bird.

(b) Where there is insertion of neither bill, but a peculiar
intercrossing of bills to be described in detail further on.

(c) Where the parent inserts its bill into the mouth or throat
of the young in the delivery of food.

We may consider these methods in order. Perhaps the best
example of the first method is furnished by the Pelican. From
Chapman's admirable account of the feeding of young Brown
Pelicans (2, p. 97) we learn that the naked young a day or two
old takes its quota of pre-digested food from the front part of
the pouch of the parent's bill, into which it has been regurgitated.
Later, however, the nestling reaches far into the throat or gullet
of the parent, thus exhibiting what may be called a feeding
initiative.

In general this seems to be the usual method for all of the
Steganopodes, unless there should be an exception in the
Phaethontidae. Chapman has observed the same method of
feeding in the Booby (Sulidae), the Water Turkey (Anhingidae) ,
and the Man-o-war-bird (Fregatidae). The Cormorants are also
known to feed in the same way (2, p. 217). The same writer
regrets not being acquainted with the feeding habits of the
Tropic Birds (Phaethontidae) so that an inclusive statement
might be made for the order. It is therefore unfortunate that
Gross, in his most careful study of the Yellow-billed Tropic
Bird (3), was not more explicit on this point. He dismisses
the subject with the remark that " The food is transferred
from the pouch-like gullet of the adult to that of the young

THE FEEDING OF NESTLING BIRDS 195

by a process of regurgitation. This transfer of food is ac-
companied by a series of gulps, strains and wrigglings of the
head and neck on the part of both birds." (3, p. 67). Attention
will be called later to this description of the regurgitative process.

The Glossy Ibis, which belongs to the Herodiones, practices
the same method. Baynard (4, p. 109) says: ' The manner
of the Glossy Ibis in feeding [its young] is to regurgitate the
food up into the throat or mouth and for the young to put his
bill, and many times his head, down the old one's throat and
take his portion." As the young of this species grew older the
parent would disgorge partially digested moccasins into the nest,
and the young would pick up the food.

Very slightly differing from this is the account by Ward of
the feeding of the young of the Herring Gull (5). Ward (quoted
by Strong, 6, p. 37) describes the process as follows: ; The
young comes in front of an adult and with a bowing and
courtesying movement puts up its bill to that of the old one,
continuing the bowing for several minutes, resting between
times. Sometimes it took hold of the adult's bill with its own,
at other times merely touched bills. When the adult opened
its mouth the young put its bill within. Failing to get indica-
tions of food, it went to another adult, and repeated the opera-
tion, passing in succession to several, until at length it seemed
to get some favorable signs, for it remained by this one, alter-
nately begging and resting. After some time it was apparent
to me that the adult was striving to regurgitate. It would
open its mouth, stretch nearly horizontally, then bring its head
down to the ground. After a moment it would close its bill,
turn its head to one side and look at the ground over which
it had been straining, as though expecting to find something
there. Other gulls were from time to time attracted to the
scene, but were promptly chased away by this bird, who ran
rapidly at them with open beak and outstretched wings. Perhaps
half an hour after these efforts began I saw a portion of a fish
appear in its mouth, and a moment later it was deposited on
the ground, when the young promptly seized it . . . The
adult assisted in breaking it up . . . The young fed mostly
from the ground, but occasionally snatched a piece from the
bill of the adult." Strong has made similar observations on
this process in the same species (6 and 7).

196 T. C. STEPHENS

In the second mode of nestling feeding which we arbitrarily
recognize, the transfer of food is accomplished by the inter-
crossing of bills of the young and the adult, without the insertion
of either. We find this method represented in at least three
orders, viz., the Tubinares, the Herodiones, and the Odonto-
glossae.

In the case of the Laysan Albatross the bill of the young
is placed crosswise between the mandibles of the adult. Partially
digested food is then regurgitated and directed by the tongue
of the adult into the open mouth of the young 2 .

Among the Herodiones a somewhat similar method is found.
Chapman (2, p. 121) has described the feeding process in Ward's
Heron, the Florida Great Blue Heron, in the following words:
"As the parent stepped into the nest, its bill was seized by one
of the young. The young bird did not thrust its bill down the
parental throat, nor was the parent's bill introduced into that
of the young. The hold of the young bird was such as one
would take with a pair of shears, if one were to attempt to cut
off the adult's bill at the base. In this manner the old bird's
head was drawn into the nest where more or less digested fish
was disgorged, of which all the young at once partook."

The same author says the process of feeding the young of

the American Egret is identical with that just described.

Essentially the same thing is described by Gabrielson for the

Bittern and the Least Bittern (8). But in these instances the

food was not dropped upon the nest; it was, instead, passed

directly into the young bird's mouth — perhaps a more refined

and advanced process. His account for the Bittern is as follows:

"As soon as she reached the nest the young commenced jumping

at her beak, continuing this until one succeeded in seizing it

in his beak at right angles to the base. A series of indescribable

contortions followed, the head of the female being drawn jerkily

in all directions and the muscles of the neck working convulsively.

Finally the head and neck were placed flat on the nest for several

seconds and then slowly raised again. As it [the head] came

up the food came slowly up the throat into the mouth. As

the food passed along the beak, the open beak of the young

bird follow ed its course along until it slid into its mouth and

2 1 am indebted to Professor Homer R. Dill, of the University of Iowa, for the
information regarding the use of the tongue by the adult albatross in feeding the
young.

THE FEEDING OF NESTLING BIRDS 197

was quickly swallowed. The young then released its hold and
the parent stood with the muscles of the neck twitching and
jerking." (8, p. 6-4).

While the process is quite similar in the Albatrosses and
Herons, one point of difference will be noted, viz., that in the
former the beak of the nestling is placed crosswise between the
adult mandibles, while in the latter the parent's bill is placed
between the mandibles of the young.

In the Flamingo, belonging to the Odontoglossae, the feeding
process may be referred to the same general plan, but with
slight modification. The bill of the Flamingo is of such size
and peculiar shape that the usual methods are not at all per-
missible. Neither bill is inserted in the other; nor would it
be possible for the young to grasp the bill of the adult. So
we find, in the words of Chapman (2, p. 187) that " What in
effect is regurgitated clam broth, is taken drop by drop from
the tip of the parent's bill." The food is really dropped from
one bill to the other in a very dexterous fashion.

From the recent description by Gabrielson of the feeding
process in the Rose-breasted Grosbeak, it might seem that there
is in the Passercs a feeding process apparently belonging in this
category (9). It would hardly be likely that this case could be
related to those here described, but it is probably an adaptation
of independent origin.

The third method of feeding which we will take into account
consists of thrusting the food into the mouth or throat of the
nestling by the parent bird. It appears to be the common
method in the higher orders of birds, viz., the Coccyges, the
Pici, the Macrochircs, and the Passeres.

Perhaps the best analysis of this process has been given by
Herrick (10) concerning the Black-billed Cuckoo. Food, which
usually consists of insects, is placed by the parent well into the
mouth or throat of the nestling. This act is a stimulus to the
nervous mechanism of the young, and the response is a purely
reflex act which sets in motion the pharyngeal muscles, thus
accomplishing the deglutition of the food. The reflex act of
swallowing is an automatic consequence of the contact stimulus
on the throat or mouth. " This complex performance ' says
Herrick, " which represents the simplest sign language of the
hungry bird, appears as a uniform chain-reflex, and is as pre-

198 T. C. STEPHENS

dictable, and seems as mechanical as the response of the electric
bell."

He points out, however, that this reflex does not long remain
unmodified; that it may be inhibited by satiety, or accelerated
by hunger. The rapidity of the reflex is dependent chiefly upon
the hunger state; though it appears, in some cases, at least,
to be necessary that the food be placed in contact with a certain
region of the buccal cavity in order to provoke a prompt re-
sponse. In one brood of Cuckoos with which Herrick worked
he found it habitual for the adult to place its bill just within
the tip of the nestling's beak, or place the insect food across
between the mandibles of the young; in such instances the
response was slow, the parent and young often remaining thus
interlocked and motionless for five minutes by the watch, and
commonly for two minutes .On the other hand, in another
nest of the same species it was customary for the parents to
place the food deep in the throat of the young. And here,
says Herrick, " Every trial was a reaction test, and upon failure
to swallow promptly, the food was withdrawn and another
nestling was tested, precisely as in vireos, thrushes, and other
passerine birds."

The exact conditions under which the parent awaits the reflex,
or repeats the stimulus upon the same nestling, or carries the
test to another bird, is a problem which may well be further
investigated.

One of the interesting questions in connection with the feeding
of passerine nestlings is as to the practice of regurgitation by
the parents. It is quite possible that all of the passerine families
are not uniform in their method of feeding the young. It is
also quite probable that much of the difference of opinion in
the matter is due to lack of agreement in terminology. It is,
of course, important to know whether the passerine birds are
uniform in this particular form of behavior.

The first necessity is the determination of the meaning of
the word " regurgitation." It would perhaps not aid in our
present purpose to attempt to analyse the different senses in
which the term has been used by various ornithological writers,
but we may at once resort to that court of appeals, the dictionary.
The Century Dictionary defines the word as meaning " To pour
or cause to rush or surge back." The Standard Dictionary

THE FEEDING OF NESTLING BIRDS 199

gives the meaning as " To throw or pour back, as from a deep
or hollow space; cause to surge back, as some mammals re-
gurgitate food already swallowed."

These definitions seem to be eminently satisfactory, and I
venture to presume that they will find approval in the judgment
of most ornithologists. The idea of regurgitation, then, implies
the previous act of swallowing of food, with concomitant reflex
muscular activity. It seems reasonable to suppose that any
reversal in the direction of passage would just as certainly re-
quire muscular action. And there is plenty of evidence to show
that in the lower orders where true regurgitation occurs, such
muscular action is clearly present. Let me remind you of the
description previously quoted from Gross of the strainings and
wrigglings of the head and neck of the Yellow-billed Tropic
Bird in feeding the young; and of the muscular contortions
of the bittern as described by Gabrielson. The photographs
accompanying Fisher's account of the Laysan Albatross (11)
indicate a raising and lowering of the head during the process.

Let us then consider that regurgitation cannot be accomplished
without muscular effort, involving the pharyngeal muscles.

It may be objected that the position of all passerine birds in
feeding is such as to permit gravitational regurgitation; that
when the stomach and crop are higher than the head, the food
may run down into the buccal cavity. Bearing in mind the
collapsible nature of the esophageal walls when at rest, gravi-
tational regurgitation does not seem to be a tenable hypothesis,
and is not entitled to serious consideration until substantiated
with some concrete evidence.

For the Coccyges Herrick's account of the life history of the
Black-billed Cuckoo (10) is very complete. He makes no
explicit statement as to regurgitation but the inference is easily
drawn that he did not observe it.

In the case of the Pici the evidence for regurgitation seems
to be good. Mr. William Brewster (12, pp. 233-235) gives
a very complete description of the process of feeding the nestling
of the Flicker. The parent approached the nest with the
mandibles shut, no food visible; when the parent's bill was
thrust into the nestling's throat there was a pumping movement
accompanied by corresponding twitching of the tail and hinder
parts of the body, and a slighter movement of the wings. As

200 T. C. STEPHENS

many as four young were usually fed at each visit. Similar
testimony is given by Airs. Miller (15, p. 18), by Baskett (13,
p. 110), and by Burns (14, p. 54).

Passing on to the Macrochires we find that a considerable
number of observations have been recorded.

In 1890 Mr. Brewster (16) described very vividly the feeding
of the young of Hummingbirds in the following words : "Alighting
on the edge of the nest, her tail pressed firmly against its outer
side in the manner of a woodpecker, her body erect, she would
first look nervously around, then thrust at least three-fourths
of the total length of her bill down between the upraised open
mandibles of the young bird. Next she would shake her head
violently as if disgorging something; then, with their bills glued
tightly together, both birds would remain, for the space of
several seconds, perfectly immovable save for a slight, rapid,
pulsating or quivering motion of the mother's throat. The
actual contact of the bills lasted once for four seconds, once
for six seconds, and twice for eleven seconds, the time being
taken by a stop watch." ..." The close and prolonged con-
tact of the bills, the shaking of the mother's head, the subsequent
quivering of the mother's, and, above all, the fact that after
sitting on the nest for nearly an hour, she fed the young a second
time without once leaving the tree in the interim, convinced
me that the method of feeding was by regurgitation."

This testimony is confirmed by Shoemaker (17), for the same
species, who says: ' Very soon the mother bird appeared, and
after a wary approach, alighted upon the edge of the nest and
thrust her bill far down the throat of the young bird. I could
see her throat move as she regurgitated the food. She left
her bill in the little one's throat for about six seconds."

Such direct testimony cannot well be questioned without a
careful re-examination of the circumstances. It should be borne
in mind, however, that the length of the bill in the Trochilidae
is a factor which must be taken into account. Even if the food,
which probably consists of insects, were held in the mouth, or
buccal cavity, we might suppose that some muscular action
would be necessary to force it out along the mandibular tube
and into the throat of the young bird. In Mr. Brewster's
account it does not add to the proof that the parent did not
leave the tree between feedings; since it might easily be assumed
that the tree abounded in food material, i. e., insects.

THE FEEDING OF NESTLING BIRDS 201

We may also take into account the recorded observations
of the feeding of young of the Nighthawk (belonging to the
Caprimulgidae) which is in the same order as the Hummingbird,
viz., the Macrochires.

Herrick watched the Nighthawk feeding its young, and says
(18, p. 134) of the mother approaching the young ones: ' She
is loaded with fire-flies, and as her great mouth opens you behold
the wide jaws and throat brilliantly illuminated like a spacious
apartment all aglow with electricity. With wings erect and
full-spread the old bird approached to within fifteen inches of
my hand, making an electric display at every utterance of her
harsh ke-ark. Then standing over her young, with raised and
quivering wings, she put her bill well down into his throat and
pumped him full. His down-covered wings were spread and
a-quiver. In this position they remained interlocked and silent
for one or two minutes."

Here is a case where, notwithstanding the interlocking of
the parent and young for a brief period, the feeding cannot be
regarded as regurgitation because the fire-flies were plainly seen
in the mouth and throat as the adult approached. The question
naturally arises now, may not other cases where the interpretation
of regurgitation is based upon interlocking and slight quivering,
be very similar to that of the Nighthawk? I am not aware that
Herrick interprets this feeding process in terms of regurgitation.

Incidentally, we may note here the clear evidence of the fact
that two families, at least, of the Macrochires feed as do the
passerine birds, viz., by the insertion of the adult bill into the
mouth or throat of the young. It is well to note this because
of the occasional assertion that certain of the Macrochires, e.
g., the Micro podia 1 ae, feed the young by taking the nestling's
bill into the parental mouth.

We may now attempt to consider some of the evidence with
reference to regurgitation in the Passeres. Perhaps the most
extensive paper upon the subject is one by Mrs. Wheelock,
published in 1905 (19). While this paper gives the author's
observations in greatest detail, her conclusions are best sum-
marized in the preface of her book on " The Birds of California,"
in which it is stated ..." that young of all Macrochires,
woodpeckers, perching birds, cuckoos, kingfishers, most birds
of prey, and many sea birds are fed by regurgitation from the
time of hatching through a period varying in extent from three

202 T. C. STEPHENS

days to Jour weeks, according to the species. Furthermore, that
birds eating animal flesh or large insects give fresh (unregurgi-
tated) food to their young at a correspondingly earlier stage of
development than do those varieties which subsist on smaller
insects or seeds. Also that exclusive seed eaters are usually
fed by regurgitation so long as they remain in the nest."

' Out of one hundred and eighty cases in every instance
where the young were hatched in a naked or semi-naked condi-
tion they were fed in this manner for at least three days. In
some instances the food was digested, wholly or in part; in
others it was probably swallowed merely for convenience in
carrying, and was regurgitated in an undigested condition."
There seemed to be no definite relation between the duration
of the period of regurgitative feeding and the length of time
required for full development of the young.

Mrs. Wheelock bases her paper in The Auk upon records of
one hundred and eighty-seven broods (not species), in all of
which the observations began on the day of hatching. The
following families are represented in her records: Fringillidae,
Turdidae, Mimidae, Icteridae, Sittidae, Hirundinidae, Vireonidae,
etc.

Her method consisted in watching the nest at distances vary-
ing from ten to forty feet, though in some cases the distance
may have been shorter; and in sampling the contents of the
crop, immediately after feeding, by the insertion of a feather,
and withdrawing such matter as adhered to it.

An adequate review of this important paper would be out
of place in the present connection. The writer believes that
the conclusions stated in the paper must be substantiated by
repeated observations, since certain subsequent study has not
been wholly confirmative.

As was suggested in Mrs. Wheelock' s paper more notes have
been published with reference to regurgitation in seed-eating
birds. Thus, Wood (21) recently makes the statement that
regurgitation occurs in the Goldfinch, although the evidence is
somewhat circumstantial. Bergtold (22), in a very compre-
hensive and exhaustive account of the life-history and behavior
of the House Finch, states that in that species the young are
fed by regurgitation until they leave the nest (22, p. 59). A
detailed description of the act of regurgitation is omitted, so

THE FEEDING OF NESTLING BIRDS 203

that we are unable to judge what the author's conception of
the regurgitative process is.

Miss Stanwood (23) incidentally states that the Olive-backed
Thrush feeds by regurgitation. McAtee (24) states that Gros-
beaks have been observed to feed the young by regurgitation;
the same author elsewhere (25, p. 421; and 26, p. 342) takes
very positive ground in favor of regurgitation, especially in
the Fringillidae, but without offering any evidence.

We may now consider the negative testimony. Jones (27,
p. 42) states that regurgitative feeding is never practiced by
the Common Tern {Sterna hirundo).

Judd, in a very excellent account of the food of nestling birds
(28, p. 412), while not explicitly discussing regurgitation, says
that the first meal of the nestlings of the crow blackbird often
consists of plump spiders of soft texture. Likewise, on page
425, the same author says that the first meal of the Crow
usually is a young grasshopper, a plump spider, or a soft cutworm.

Jones (29, p. 69) reports the observations made by his students
at the nestside of the Field Sparrow, the Song Sparrow, and
the House Wren. He says: " There was no evidence that any
of these birds fed by regurgitation. In the case of the sparrows
this was clearly proved, but what might have happened in the
case of the wrens can only be surmised. At any rate, the food
was uniformly brought dangling from the bill and was not
swallowed before being delivered to the nestlings. This was
the case with the first feedings of both the Song Sparrows and
the House Wrens."

Bigglestone (30) made a careful study of the nest life of the
Yellow Warbler. All of the young were under observation
within a few hours after hatching; but the last egg was under
observation during hatching, and from that time onward the
young bird was under constant observation till it left the nest.
No regurgitation was observed, although this was one of the
chief objects of the entire study.

A similar study was made of the Catbird by Gabrielson (31),
in which two of the nestlings were under observation from the
moment of hatching onward. Here also nothing suggestive
of regurgitative feeding occurred.

During the summer of 1914 two similar nest studies were
completed. The first, which has not yet been published, was

204 T. C. STEPHENS

undertaken by Mr. Jay Kempkes, who worked with the Western
Meadow Lark (Stumella ncglecta) at the Iowa Lakeside Labora-
tory, on Lake Okoboji. Here also the results were negative,
for no regurgitation was observed, although the birds were
constantly under observation while in the nest; and in this
instance one of the eggs was watched through the hatching
process.

The second of these studies deals with the Rose-breasted
Grosbeak (Zamelodia ludoviciana) , and was carried on by Mr.
Gabrielson at Marshalltown, Iowa. This report was published
in a recent number of the Wilson Bulletin (9). The question
of regurgitation was made the chief object in this study, because
of the statement in one of the bulletins of the Biological Survey
of the Government (24, p. 75) that the Grosbeaks feed their
young in this way. The result of this very critical field study
of the Rose-breasted Grosbeak was negative so far as regurgi-
tative feeding was concerned.

While, doubtless, this review falls short of covering all of the
literature on the subject, it may be sufficient to indicate that
comparatively few studies have been undertaken on the passerine
birds having expressly in view the question of regurgitative
feeding. The writer believes the evidence is against this method
in the passerine group. . There seems to be a field here for much
interesting and valuable work.

The question will arise, What are to be the criteria in such
a problem? The scientific attitude of mind is, no doubt, the
first essential qualification on the part of the observer. It is
perhaps no more important in any kind of ornithological inquiry
than in this close and precise field work. Certain conditions
must be observed. Among other things observation at close
range is essential; and close range in this connection should
mean two to four feet, with the nest not above the level of the
observer's eye. The study should begin, preferably, with the
hatching of the egg. Definite observation must be made as to
whether, or not, the food is carried visibly in the parent bird's
beak. It is true that McAfee (26, p. 342) claims that the
visibility of the food, held in the mandibles of the bird as it
visits the young for feeding, is no disproof of regurgitation.
But as we have reviewed the numerous instances where un-
questioned regurgitation occurs, we have not found it customary

THE FEEDING OF NESTLING BIRDS 205

for such species to carry food in the mandibles. And in such
a proper restriction of the term "regurgitation" as I have made
in this paper, and as I believe the majority of ornithologists have
been accustomed to use the term, it seems to me very improbable
that birds which practice regurgitation would alternate with
solid food. Food apparent in the bill may not disprove that
the mouth is full likewise, and that the surfeit may extend
well into the gullet; but that the removal of this surfeit from
the throat or gullet is to be interpreted as regurgitation is quite
beyond the limits of good terminology. I believe, therefore,
that food visible in the bill is very good evidence against
regurgitation.

One other point to be taken into account is the presence or
absence of action of the pharyngeal muscles. While this may
often be a little obscure and uncertain, especially for inexperienced
observers, it nevertheless may be regarded as important circum-
stantial evidence. If muscular activity is clearly present a
presumption is established in favor of regurgitation; but if no
muscular activity can be determined such a presumption cannot
be claimed, to say the least.

In conclusion I may be permitted to suggest that we may,
in the future, find that there is greater significance in the
comparative study of the manner in which nestlings are fed
than has hitherto been recognized. Among the lower orders
of birds we find a method of feeding the young which we will
have little difficulty in regarding as primitive. In the higher
birds, such as the Passeres, we find developed a distinctly com-
plex performance. Between the primitive and complex pro-
cesses alluded to, we find feeding methods which may be regarded
as transition stages. Without, at this time, attempting a too
rigid arrangement of these feeding processes, as to their natural
and phylogenetic sequence, we may at least be justified in the
conclusion that the problem deserves much further study.
Whether any importance might be attached to such a quasi-
physiological process in the determination of affinities it would
be premature to speculate. But that this may be no more
than a random suggestion, only future investigation can
determine.

206 T. C. STEPHENS

REFERENCES

(Arranged in order of citation)

1. Smith, Arthur F. Notes on the Spotted Sandpiper. Wils. Bull, XXVI,

1914, pp. 81-86.

2. Chapman, Frank M. Camps and Cruises of an Ornithologist. New York,

1908.

3. Gross, Alfred O. Observations on the Yellow-billed Tropic Bird (Phaethon

americanus Grant) at the Bermuda Islands. Auk, XXIX, 1912, pp. 49-71.

4. Baynard, Oscar E. Home Life of the Glossy Ibis {Plegadis aulumnalis Linn).

Wils. Bull., XXV, 1913, pp. 103-117.

5. Ward, H. L. Notes on the Herring Gull and the Caspian Tern. Bull. Wise.

Nat. Hist. Soc, IV, 1906, pp. 113-134.

6. Strong, R. M. On the Habits and Behavior of the Herring Gull, Larus

argentatus Pont. Auk, XXXI, 1914, pp. 22-49, pp. 179-199.

7. Strong, R. M. On the Habits and Behavior of the Herring Gull, Larus

argentatus Pont. Report Smith. Inst., 1914, pp. 479-509.

8. Gabrielson, Ira N. Ten Days' Bird Study in a Nebraska Swamp. Wils.

Bull., XXVI, 1914, pp. 51-68.

9. Gabrielson, Ira N. Field Observations on the Rose-breasted Grosbeak.

Wils. Bull., XXVII, 1915, pp. 357-368.

10. Herrick, Francis H. Life and Behavior of the Cuckoo. Jr. Exp. Zool,

IX, 1910, pp. 169-233.

11. Fishfr, Walter K. On the Habits of the Laysan Albatross. Auk, XXI,

1904, pp. 8-20.

12. Brewster, William. A Brood of Young Flickers (Colaptes auratus) and How

They Were Fed. Auk, X, 1893, pp. 231-236.
13 Baskett, J. N. . . . Nidologist, II, p. 110.

14. Burns, Frank L. A Monograph of the Flicker {Colaptes auratus). Wils.

Bull, XII, 1900, pp. 1-83.

15. Miller, Olive Thorne. First Book of Birds.

16. Brewster, William. Food of Young Hummingbirds. Auk, VII, 1890, pp.

206-207.

17. Shoemaker, Frank H. A Late Nest of the Ruby-throated Hummingbird.

Proc. Nebr. Ornith. Union, II, 1901, pp. 34-38.

18. Herrick, F. H. The Home Life of Wild Birds. New York, 1905.

19. Wheelock, Irene G. Regurgitative Feeding of Nestlings. Auk, XXII, 1905,

pp. 54-71.

20. Wheelock, Irene G. The Birds of California.

21. Wood, J. Claire. The Food of Nestling Goldfinches. Wils. Bull, XXV,

1913, p. 96.

22. Bergtold, W. H. A Study of the Finch. Auk, XXX, 1913, pp. 40-73.

23. Stanwood, Cordelia J. The Olive-backed Thrush (Hylocichla ustulala

swainsoni) at His Summer Home. Wils. Bull, XXV, 1913, pp. 118-137.

24. McAtee, W. L. Food Habits of the Grosbeaks. Bull. 32, U. S. Bur. Biol.

Survey, 1908.

25. McAtee, W. L. Field Studies on the Food of Nestling Birds. (A review).

Auk, XXXI, 1914, pp. 420-421.

26. McAtee, W. L. Correspondence. Wils. Bull, XXVII, 1915, pp. 339-344.

27. Jones, Lynds. A Contribution to the Life History of the Common (Sterna

hirundo) and Roseate (S. dougalli) Terns. Wils. Bull, XVIII, 1906, pp.
35-47.

28. Judd, Sylvester D. The Food of Nestling Birds. Yearbook, U. S. Dep't.

Agric, 1900, pp. 411-436.

29. Jones, Lynds. Some Records of the Feeding of Nestlings. Wils. Bull, XXV,

1913, pp. 67-71.

30. Bigglestone, Harry C. A Study of the Nesting Behavior of the Yellow

Warbler (Dendroica ae. aestiva). Wils. Bull, XXV, 1913, pp. 49-67.

31. Gabrielson, Ira N. Nest Life of the Catbird. Wils. Bull, XXV, 1913,

pp. 166-187.

Sioux City, Iowa.

A STUDY OF THE REACTIONS OF CERTAIN
BIRDS TO SOUND STIMULI

ROLAND F. HUSSEY

I. INTRODUCTION

This problem was undertaken at the Biological Station of
the University of Michigan, under the direction of Professor
R. M. Strong, between the dates of July 10 and August 21,
1916.

The Biological Station is located on Douglas Lake, in the
western part of Cheboygan County, Michigan. The region
immediately about the camp is covered with a fairly dense stand
of young aspens, birches, and pin cherries, with occasional oaks
and maples. This is the type of vegetation which has succeeded
the original forests of pines and hardwoods in this cut-over
and burned-over country. A few pines remain, chiefly along
the lake shore, but there are no hardwoods in the immediate
vicinity of the Station. Another important type of vegetation
is found in the dense cedar bogs, where " arbor vitae, balsam,
tamarack, and spruce form a nearly impenetrable jungle."

The bird life naturally shows a marked ecological concentra-
tion of species in these widely different habitats. In order
that observations on any species might be made as nearly contin-
uous as was desired, it was thought best to work with nesting
birds as far as possible; and accordingly on July 5 search was
begun for suitable subjects. But although birds are really
fairly plentiful in the aspens during the early summer, the
abundance of nesting sites furnished by so extensive and so
uniform a habitat as the aspen association made the discovery
of nests very difficult. In fact, it was not until the morning
of July 8 that a nest was found located so that work could
successfully be carried on.

The nest referred to was that of a hermit thrush. It was

placed on the ground in a small cluster of oak seedlings, about

a hundred feet from the shore of Douglas Lake. Work was

begun here on July 10, and was continued until the young birds

2 207

208 ROLAND F. HUSSEY

left the nest. By far the greater part of the summer's work
was done on these hermit thrushes, because the nest was easily
accessible from the Station, and because the hermit thrush
proved a good subject for experiments of this sort.

Another nest, that of a robin, was discovered some days later,
within the limits of the camp. Although for more than two
weeks efforts were made to test the auditory reactions of this
bird, it was found to be too timid for practicable experiments.

In .addition some work was done in connection with this
problem on birds found in the field, particularly on some of
the birds characteristic of the cedar bog, and with some shore
birds. Experiments were also made upon some young cedar
waxwings which had been taken from their nest and placed
in a cage for special study ; but by the time that my experiments
were begun the waxwings had become so accustomed to the
presence of various people and to noises of all sorts that no
positive results could be obtained.

II. METHODS AND MATERIALS

The experiments with the nesting hermit thrush were for the
most part made with the aid of a small white observation tent
which was pitched so that the apertures for observation were
within six feet of the nest; and a narrow path from these aper-
tures to the nest was cleared of grasses, etc., so as to admit
the making of photographs. The tent was first pitched on the
morning of July 10, and it was taken down the same day; on
July 12 it was replaced and was left standing until all the work
possible on the hermit thrushes had been completed.

The observations on the birds of the cedar bog and on the
other birds experimented with in the open field were made
without the use of any special shelter. Great care was taken
to avoid other than auditory stimulation.

The sounds used were made as varied in character as possible.
Those tried on the hermit thrush included shouting, singing,
whistling both with the lips and with a " double-tone " metal
whistle, chirping with the lips, clapping the hands, rapping on
wood and on metal, and rustling papers and birch-bark. To
test the reactions to sounds of different pitch a mandolin was*
used. The same sounds were also used in the case of the other
birds, but such as involved sudden movement, as clapping the
hands or rapping on wood or on metal, were not tried to any

THE REACTIONS OF BIRDS TO SOUND STIMULI 209

considerable extent. There is a possibility that auditory stimuli
may sometimes have been accompanied by visual stimuli in
the case of the thrushes when the giving of the sound required
motion on my part. My blind was pitched between the nest
and the morning sun, and the tent- walls were very thin, so that
movements within it were often perceived vaguely from the
outside.

The reactions to all these stimuli were more or -less similar,
differing in degree only. In every case they consisted of move-
ments of the head and of the bill, of raising the wings, and of
winking. The eye-wink, however, is of doubtful value, and
was considered only in the case of the thrushes. The interval
between the giving of the stimulus and the perception of the
reaction to it seemed fairly constant for those birds which reacted
at all, being about three-fourths of a second, as nearly as could

be determined.

III. OBSERVATIONS

1. On the nesting hermit thrush. July 10, 9:10-10:50 a. m.,
1 :30-3 :20 p. m. The bird was not much alarmed, even at the
very first, by my stirring about in the tent, but flew when, at
9:20, I caused the side of the tent toward the nest to flutter.
It returned in two minutes, and took its usual position on the
nest, facing the tent obliquely, with its head turned toward the
woods, to which the bird always went on leaving the nest, rather
than toward the lake. Only rarely, and then usually in response
to sound stimuli, did it turn its head directly toward the tent.
I never saw it approach the nest except from the side toward
the blind.

The first time that I released the shutter of my camera the

bird showed decided* alarm, but this was of short duration,

less than three seconds. Rapping with a pencil on the metal

cover of my note-book caused the bird to turn its head from

side to side, as if it were trying to locate the source of the sound.

A short blast on the whistle caused the same reaction, apparently

of about the same intensity; when this experiment was repeated

the same result was obtained, but after the tenth trial the bird

seemed to have become accustomed to it, and the only response

it made was merely to assume a more alert attitude. 1

1 Similar observations were made by Strong on gulls studied from a blind.
Strong, R. M., 1914. On the Habits and Behavior of the Herring Gull, Larus
argentalns Pont. The . Auk, vol. 31, Nos. 1-2, pp. 22-49, 178-199; plates 1-10,
19, 20, also Smithsonian Report, 1915, pp. 479-509.

210 ROLAND F. HUSSEY

Even so faint a sound as that made by my pencil in writing
on the note-book page seemed to cause the bird some uneasiness.
Singing caused very slight apprehension, not nearly so much as
did the sound of the waves thrown up on the shore by a passing
launch. At the second release of the camera shutter the bird
showed less concern than was evident the first time; but when
I moved the camera in winding the film, it was alarmed to the
point of standing in the nest, and it even walked off while I
was rewinding the shutter curtain. It was gone only five
seconds.

In the afternoon a companion accompanied me when I returned
to the blind. The bird stayed on the nest when we entered
the tent; but when I moved my finger in the aperture it showed
great alarm, and left the nest when I flashed a mirror there:
it was gone seventeen minutes. On its return I waited for
several minutes before attempting any experiments. I noted
that the bird was always on the alert, particularly for sounds
from the tent; however, it was not insensible to other sounds,
for several times it took food, usually large ants, from the edge
of the nest, apparently in response both to auditory and to
visual stimuli; the thrush seemed to be made aware of the ant
through an auditory stimulus, and then to discover it through
the visual sense. The bird seemed also to start visibly at
sudden noises from outside the tent; but on this date it was
concerned chiefly with those which were made within the blind.

The day was clear and warm, and the bird panted con-
siderably. I found that usually it would close its bill when I
made a sudden sound, and would turn its head and wink its
eye, while to sounds from outside the tent the response con-
sisted usually in merely turning the head. " The bird sometimes
turned its head at our conversation, but usually it paid no
attention to it; once it seemed quite disturbed at the noise
made by tearing birch-bark. My companion left after half an
hour in the tent.

The slightest movement of the tent-wall near any of the
apertures was sufficient to attract the bird's attention, but it
seemed only slightly concerned by the movements of the canvas
elsewhere. A sudden gust of wind sounding in the pines over-
head caused the bird to start slightly, and soon afterward it
started noticeably at a junco's song from the same trees. Then,

THE REACTIONS OF BIRDS TO SOUND STIMULI 211

later, when I tapped on my note-book cover with my pencil,
the bird looked about over its head as if trying to discover the
source of the sound.

During the afternoon I tried the effect of song on the bird,
but I obtained no positive results. I did notice, though, that
if the song were suddenly broken off and two or three sharp
raps given on the tent pole, the bird reacted more vigorously
than to the raps alone. And a suddenly ended loud sound
seemed to startle it more than a sudden loud sound did ordinarily.
These reactions were noticed several times during the afternoon,
but they became much less marked on successive repetitions
of the experiment.

July 12, 9:25-10:40 a. m., 1:10-3 p. m. The bird seemed
very uneasy at noises made outside the tent, particularly at
the loud cawing of some crows. After about half an hour of
quiet, I blew a loud blast on the whistle, which produced only
a slight reaction. The squeaking sound made by kissing the
back of the hand vigorously produced no visible reaction. The
higher pitched of the two whistle notes, when sounded alone,
seemed to produce a slightly more vigorous reaction than did
the lower one alone. But no whistle blast produced so strong
a reaction as did the sudden loud call of a cuckoo from the
near-by aspens.

The visual stimulus produced by a moving object seemed very
much stronger than did any auditory stimulus; when I showed
my fingers at the aperture the bird showed very decided alarm.
As the experiment was repeated the reactions became much less
pronounced.

In the afternoon I took a mandolin to the tent; and, as with
the whistle, the notes of higher pitch seemed to produce slightly
more vigorous reactions. However, the differences in the reac-
tions were so slight that it was very difficult to make such
determinations. The notes used ranged over about two octaves,
from the open G-string of the mandolin to B two octaves higher.

I found that a chord produced a more vigorous reaction than
did any single note; however, the bird quickly became ac-
customed to the chord, and after half a dozen trials merely
showed increased vigilance. As on July 10, I found that if
a sound was interrupted by another of very different character,
a more vigorous reaction was produced than by mere suddenness

212 ROLAND F. HUSSEY

or loudness of sound. After the bird had become accustomed
to the mandolin and to the whistle so as to react to them only
slightly, I tried the effect of interrupting the mandolin chords
by loucf whistle blasts, and found that the bird reacted markedly,
though with decreasing vigor as the experiment was repeated.

I experimented also on the combination of auditory with
moving visual stimuli. As stated above, the sudden appearance
of my finger at the aperture seemed to startle the bird, but the
reaction decreased rapidly in vigor on successive trials. The
same was true of the reaction produced by sounding the open
G-string of the mandolin. If, however, I showed my finger
at the aperture, then, after a short pause, moved it suddenly
and at the same instant sounded the G-string loudly, the bird
reacted vigorously; and the reaction was as strong at the thirtieth
trial as at the first.

July 13, 8:45-10:20 a. m., 1:10-3:20 p. m. The work in the
morning was similar to that of the two days previous, and the
reactions observed were similar, though less vigorous on the
w T hole. The combination of auditory and visual stimuli was
tried again, and the same results were obtained by combining
the movement of my finger with a loud whistle blast as with
a mandolin sound; although fifteen trials were made, there was
little, if any, diminution in the vigor of the reaction, which in
this case consisted usually in turning the head and closing the
bill momentarily.

In the afternoon I approached the nest from the side opposite
the tent. I crept up over a thick growth of bear-berries without
taking particular pains to move quietly, so that I am sure that
the bird must have heard me. The bear-berries screened me
from the nest, so that until I came within three feet of it I
could not see the brooding bird. I found that the bird seemed
not at all on the alert for any disturbing sound from the tent,
nor had it seemed to notice my approach. I then blew a loud
blast on the whistle, but it showed no sign of fright, and merely
turned its head slowly until it caught sight of me, when it
immediately left the nest.

I then concealed my watch carefully about six inches from
the nest, and immediately entered my tent. When, after fifteen
minutes, the bird returned, it was very plainly made uneasy
by the ticking of the watch; it turned its head from side to

THE REACTIONS OF BIRDS TO SOUND STIMULI 213

side, and looked for it when once its approximate location had
been determined from the sound. After about half a minute,
however, the bird gave up the search and paid no more attention
to the watch.

The bird now seemed constantly on the alert for sounds from
the tent, having seen me enter it; very slight reactions were
noticed, however, to whistling, shouting, and clapping the hands,
while rapping on wood and chirping with the lips produced no
visible reactions.

July 16. The three eggs hatched this afternoon. I made no
observations.

July 17, 8:40-9:30 a. m., 1:10-3:10 p. m. There was no sign
of either parent during the fifty minutes I spent in the tent
in the morning. In the afternoon one of the parents was at
the nest when I approached, and stayed while I entered the
tent. On this date, for the first time, I obtained a response
to the chirping sound. To it the bird responded by turning
the head, often as much as 90 degrees, and by winking. Although
the interval between winks varied between such wide limits as
two seconds and thirteen seconds, I think that the wink can
safely be considered as a part of the reaction, since in each
case it accompanied the turning of the head, about three-fourths
of a second after the sound stimulus was given.

To the higher-pitched of the two whistle notes the bird again
reacted slightly more vigorously than to the lower; the reaction
here consisted in turning the head and closing the bill. The
reaction to this sound was on this date much less vigorous than
that to the chirping.

I tried a new vocal sound successfully; a rapid, rolling
" rr-r-r-rr." To this the bird responded on six out of seven
trials, showing alarm not merely at the beginning of the sound,
but as long as it was continued.

The experiment of changing from one sound to another of a
different sort gave results this day on only three trials out of
seven. However, the bird seemed more responsive to all other
tests than on the last day that observations were made before
the hatching of the eggs.

July 19, 9-10:40 a. m. The young birds' eyes were just
opening when I began my observations. Any small sound from
the tent seemed to stimulate the nestlings to raise their heads

214 ROLAND F. HUSSEY

as if for food; and I noticed that in this connection such sounds
as rustling paper or birch-bark were stronger stimuli than were
sharper sounds such as tapping or chirping. When the parent
returned it stood in the nest over the young, and was continually
on the alert. The usual sound stimuli were tried — tapping,
whistling, clapping the hands, etc. — and the usual reactions
were secured. Interrupting a sound by another of a different
sort produced vigorous reactions twice in three trials. For the
first time since July 10 the bird seemed to be made uneasy by
the sound of my pencil; when I started writing it gave a slight
reaction, and again when I stopped writing. This was the
first day since July 10, however, when the wind was blowing
from the tent toward the nest.

When I made my first photographic exposure this morning,
the parent on the nest had ceased momentarily to watch the
tent and was bending over the young. The sound of the camera
aroused it to watchfulness again. Subsequent releases of the
camera shutter did not seem to attract the bird's attention
at all.

July 20, 9:10-10:50 a. m. While the parent was absent I
tried the effect of a loud whistle blast on the nestlings. At
the first trial one of them raised its head, but five subsequent
trials produced no effect.

When a parent returned I made some experiments to determine
to what extent reactions were inhibited when the bird was at
the nest. While it was approaching the nest and when within
a few inches of it, I tried various sounds, but without effect.
When this bird had finished feeding the nestlings, I gave a
loud whistle blast, and the bird started visibly. Later, when
a parent approached the nest again, and was still five or six
feet away, I tapped my note-book cover with my pencil, with
the result that the bird turned and ran off some fifty feet. When
it returned a minute later I repeated this experiment, and the
bird flew away and was gone several minutes. After several
feedings I made some further experiments, with the following
results: when the parent was coming to the nest and was
within a few inches of it, or when at the nest-side and engaged
in feeding the nestlings, it seemed to pay little or no attention
to sounds from the tent; but after the young were fed and the
nest was cleaned, it seemed always to notice them. Yet it

THE REACTIONS OF BIRDS TO SOUND STIMULI 215

was by no means as responsive (on this date) to any sounds
from the tent as to the very faint lisping sounds with which
the young asked for food.

July 21, 8:30-10:45 a. m., 1:20-2:30 p. m. One of the
nestlings was lying in the nest, when I entered the tent, with
its head raised and bill opened so that the lower mandible lay
on the edge of the nest. I clapped my hands several times at
considerable intervals, and each time the young bird responded
by raising its head nearly to a vertical position and closing the
bill slightly. When a parent returned I repeated the experiments
of the day before, and again I found that the reactions were
greatly inhibited when the bird was at or near the nest, and
that the bird when at the nest seemed to pay much more
attention to the sounds made by the young than to any sounds
I made. In the afternoon I repeated experiments on the nest-
lings, but the only response obtained was similar to the one
just described.

July 24, 8:05-11 a. m. During the first fifteen minutes of
the period the young were very quiet. I observed that whenever
I made any sudden loud noise, one of the nestlings invariably
raised its head, opened its eyes, and (apparently) watched the
tent for a few seconds. If the sound were repeated, it did not
show any further alarm, but merely continued to watch the
tent for a somewhat longer period than if the sound were not
repeated. I also noticed that the nestlings never seemed to
be aware of the approach of the parent, whether it ran or flew
to the nest-side; and it was my experience that its flight was
very noisy, with loud whirring of the wings, when it approached
the nest. In direct contrast to this was the case of some cedar
waxwings which came to my attention during the summer;
here the nestlings seemed always to be aware of the parent's
approach at a distance, and for some time were in a state of
expectancy.

Except in the cases mentioned in the preceding paragraphs
I was never able to obtain any marked response from the nest-
lings to sudden and loud sounds. This seems to me to be
directly contradictory to a statement of Lloyd Morgan (Animal
Behavior, p. 49) to the effect that young birds " show signs of
alarm at any sudden and unaccustomed sound." After I had
been in the tent twenty minutes and the young birds had become

210 ROLAND F. HUSSEY

entirely quiet, I repeated the experiments with various kinds
of sounds, as whistling, rapping on wood and on metal, clapping
my hands, etc., but I was unable to detect any positive reactions.

I also continued the observations on the inhibition of reac-
tions in the parent when near the nest, with the same results
as before.

July 25, 8:25-10:30 a. m. The observations made from the
tent this morning were substantially the same as those of the
day before, except that for the first time I was sure that both
parents came under my observation. I was not able to detect
any difference in their behavior.

At 9 :45 I left the blind and withdrew for about ten minutes.
On my return I waited until a parent had finished feeding the
nestlings and had left the nest, and then, instead of going to
the blind at once, I took advantage of some natural cover to
conceal myself. Neither of the parent thrushes seemed to pay
the slightest attention to me on their subsequent visits to the
nest, nor did either of them respond visibly to any sound that
I made, though I tried shouting, whistling, and chirping loudly.
But after I reentered the tent (at 10:15) I found that both birds
reacted markedly to these same sound stimuli.

2. Observations on a young hermit thrush. In the late after-
noon of July 26 a young thrush was taken from this nest for
the purpose of studying its reactions to sounds under laboratory
conditions. On this date it gave no evidence of reactions to
sounds or of the formation of associations between sounds and
other events. On the two days following there was no oppor-
tunity to study this bird.

On July 28, a very warm and sultry day, the bird became
sick, due doubtless to the heat and to the change in environment
and feeding; it was transferred to a more airy cage. The next
day the temperature was still higher and the bird was still sick ;
yet it was very active, and made violent efforts to escape from
the cage, with the result that it so nearly exhausted itself that
it was practically unable to open its eyes. It was on this date
that the first observations were made which showed any ability
on the part of the' bird to associate sounds with other events.
For instance, after the bird had been fed four times, it was
noticed that when the door of the cage was rattled the bird

THE REACTIONS OF BIRDS TO SOUND STIMULI 217

turned toward it at once and begged for food. If food were
not forthcoming, this reaction became less and less vigorous
on successive trials, and finally ceased altogether until the bird
had been fed again. In the afternoon of the same day the
thrush was taken outside the cage; and even under these condi-
tions, and when at some distance from it, the bird continued
to orient itself toward the source of the sound and to beg for
food when the cage door was rattled. During all of these
experiments the bird kept its eyes tightly closed.

Further experiments were planned, but the thrush died on
July 30.

3. Observations on birds of the cedar bog. On August 13
I went down into a dense cedar bog about two miles from the
station, and, while in the more open parts, I ran across a flock
of chickadees, golden-crowned kinglets, and brown creepers,
numbering perhaps thirty individuals. I tried the effect of
whistle blasts, of clucking with the mouth, and of shouting,
all at a distance of less than fifteen feet, without obtaining a
positive reaction from any of these birds; nor did this surprise
me, when the fearless habits of these birds were considered.
When I clapped my hands, however, a kinglet did fly to a farther
tree; but this was undoubtedly due rather to the effect of the
visual than to the auditory stimulus. Later I tried the effect
of a whistle blast on a black-throated green warbler, with more
positive results. The bird raised its wings as if to fly, but soon
settled back and then continued its search for food. Subsequent
trials produced no result.

4. Observations on young cedar waxwings. As was stated
in the introduction, these waxwings were so accustomed to
sounds of all sorts before my experiments were begun that
almost no positive results were obtained. In fact the only
thing definitely established was that these birds were more
responsive to sound stimuli before being fed than afterward.

5. Observations on certain shorebirds. On August 21, I made
a series of observations on some solitary sandpipers, least
sandpipers, and kildeers which were feeding on the beach near
camp. I tried whistle blasts, shouts, and percussive sounds,
both vocal and clapping my hands. The only reaction I observed
among the sandpipers was given by a solitary sandpiper, which

218 ROLAND F. HUSSEY

raised its wings when I clapped my hands. To the whistle
blasts the kildeers responded at first by running a few steps
along the shore, but later they gave no reaction to this sound.
To duckings they did not respond, but when I clapped my
hands the whole flock took flight. The visual stimulus un-
doubtedly was the stronger in bringing about this reaction.

Later I made similar experiments on a solitary kildeer, without
obtaining any positive results.

IV. SUMMARY

1. The hermit thrush was constantly on the alert as long as
I was in the observation tent, and while I was there it seemed
very often to be more sensitive to sounds from without the
tent than to those which I made inside.

2. If a continued sound were suddenly interrupted by another
loud sound of a very different character, the bird reacted more
vigorously in most cases than to either sound alone.

3. The hermit thrush seemed not particularly sensitive to
sounds when no one was in the tent.

4. An auditory stimulus produces a much stronger reaction
in birds when it is reinforced by a moving visual one, and a
moving visual stimulus when it is reinforced by an auditory
stimulus.

5. The hermit thrushes under observation very soon became
accustomed to various sounds and in a short time ceased to
react visibly to them unless the sounds were reinforced by
other stimuli.

6. The hermit thrush was very much more sensitive to sound
stimuli early in the period of experimentation than later, and
there was a secondary maximum just after the hatching of
the eggs.

7. The reactions to sounds on the part of the hermit thrush
are very much inhibited when the bird is at the nest or within
three feet of it, but this inhibition is much less after the young
are fed than before.

8. The adult hermit thrush is apparently more sensitive to
notes of higher pitch than to sounds of lower pitch.

9. Very young birds were more influenced by such sounds as
the rustling of paper or of birch-bark than by sharper sounds
such as tappings, whistling, etc.

THE REACTIONS OF BIRDS TO SOUND STIMULI 219

10. I found the young of the hermit thrush during the first
ten days after hatching very little influenced by sound stimuli.

11. A thirteen-day old hermit thrush very quickly learned
to associate certain sounds with feeding.

12. Shore birds, chickadees, golden-crowned kinglets, and
brown creepers observed in the field were not visibly influenced
by sound stimuli. A slight response was obtained from a black-
throated green warbler.

13. From a rather unsatisfactory study of young cedar wax-
wings it appears that they are more sensitive to sounds before
being fed than afterward.

CHOICE OF FOOD IN AMEBA

A. A. SCHAEFFER

Zoological Laboratory, University of Tennessee

CONTENTS

PAGE

Introduction 220

Reactions to two particles of different constitution lying close together 220

The effect of mechanical stimulation 226

Discussion of experimental results 231

Choice of food 231

Selection of food in stentor and Paramecium as compared with that in ameba. . 246

Summary 250

Bibliography 251

Explanation of plates 252

INTRODUCTION

In a previous paper ('16a) I outlined the problem of choice
of food among animals as I conceive it, and in this paper I wish
to discuss the power of choice in ameba as exhibited in the
various series of experiments recorded in several of my previous
papers, as well as in experiments which are recorded for the first
time in this paper.

The problem of choice of food has turned out to be very
intricate and difficult; much more so than was at first suspected.
It is rendered especially difficult because previous experience in
sense perception plays a very important part in selection. Choice
thus becomes in large part a developing or historical process.
Series of individual acts of choice are the smallest units which
can be considered in analyzing this problem. Individual acts
of choice are frequently quite meaningless, and even contra-
dictory to each other, when removed from their historical set-
ting. This fact is of the greatest importance in this connection
and must not be lost sight of.

Before a discussion of choice of food is entered upon, I wish
first to describe two series of experiments which have particular
bearing on this matter. One series has to do with choice expressed
by an ameba when encountering two particles of different compo-
sition lying very close together, while the other series deals with
the effect of mechanically agitating various kinds of particles
in close proximity to the ameba.

220

CHOICE OF FOOD IN AMEBA 221

REACTIONS TO TWO PARTICLES OF DIFFERENT CONSTITUTION

LYING CLOSE TOGETHER

Globulin and silicic acid. 1 — A grain of globulin and a grain of
silicic acid were laid close together in the path of an Amoeba
proteus 2 — figure 1. The test objects were laid some distance
ahead so that all the changes in behavior due to the test sub-
stances could be observed. The ameba moved directly forward
for a considerable distance, then sent out on either side several
pseudopods which were, however, soon retracted — 3. As the
ameba moved further forward toward the test substances, it
broke up into three pseudopods, the middle one of which pointed
directly toward the silicic acid, though it did not move into
contact with the acid — 5. The right-hand pseudopod was
retracted while the ameba moved on through the pseudopod
on the left, gradually swinging around the silicic acid, and
coming into contact with the globulin from the opposite direc-
tion. The globulin was ingested without the formation of a
food cup, while the silicic acid was carried to the back of the
ameba where it remained a short time.

Silicic acid and egg albumin. — The ameba whose reactions to
'globulin and silicic acid were described above, later reacted
positively to, but finally avoided, a capillary tube containing a
solution of egg albumin — 14. After this test a grain of silicic
acid was laid in front of the tube of albumin — 26. The ameba
moved toward the silicic acid until within about fifty microns
of it when there were formed several side pseudopods, of which
the one on the right became the main pseudopod. But this
one soon curved to the left and toward the silicic acid, and
then kept on moving in this direction until within thirty microns
of the acid when the tip of the main pseudopod turned slightly
to the right and moved past the test object — 35. But as soon
as the tip of the ameba was well past the silicic acid, the ameba

1 For method of preparation of the test substances mentioned in this paper
reference should be made to my former papers, '16b, '16c.

2 For a description of the species of ameba mentioned in this paper refer to my
paper '16d, in Science. In previous papers I spoke of 'raptorial' and 'granular'
amebas. Raptorial amebas are of the species Amoeba dubia Schaeffer, while the
granular amebas refer to two species: Amoeba proteus Pallas emend. Leidy, and
Amoeba discoides Schaeffer. From my notes I am unable to tell in all cases to which
of the latter two species the granular amebas belonged. This is due to the fact
that discoides was not discovered to be distinct from proteus until after these
experiments were made. In this paper therefore, the label A. proteus includes also
the species discoides.

222 A. A. SCHAEFFER

turned toward the open end of the tube and a small pseudopod
was also sent out toward the tube — 38. The ameba, however,
moved away from the tube of albumin through a pseudopod
sent out on the right. This experiment demonstrates very
clearly that, although the mere presence of a solid object im-
mersed in a solution of some other substance serves to attract
an ameba, the main factor in determining the movements of an
ameba is either the increasing density of the diffusing molecules
or ions as the source of diffusion is approached, or some other
physical disturbance propagated radially from the source of
diffusion.

(Figures 41-50 represent a control experiment showing that
silicic acid by itself does not cause a positive reaction.)

Globulin and soluble egg albumin. — In an experiment similar
to the preceding, on the same ameba, but in which a grain of
globulin was used instead of silicic acid — 51, the ameba moved
away from the tube of egg albumin after the globulin was eaten —
65. This experiment, when studied in connection with the
preceding experiments upon this ameba, shows very strikingly
the nice discriminations in the selection of food which this
animal is capable of making.

Carbon and lecithin. — The lecithin bore Merck's label and was
made from eggs. A small amount of it was smeared on a carbon
grain which was then laid in the path of an A. proteus — 101. The
ameba moved forward into contact with it, and then moved
off through a pseudopod which had been forming on the left —
107. The carbon-lecithin was then shifted — 109. The ameba
moved toward the test object, then turned toward the left, then
moved forward into contact with it again, and then moved on
leaving the carbon-lecithin behind. A new piece of carbon,
with fresh lecithin was then placed near the ameba — 115. The
ameba moved forward into contact with it and then passed on.
A small pseudopod was thrown out posterior to the carbon-
lecithin, and later another anterior to it, but the ameba finally
moved away without attempting ingestion.

Globulin and lecithin. — A grain of globulin was smeared with
lecithin and placed in the same ameba's path — 123. The ameba
moved forward a short distance toward the test objects, then
turned to the left, avoiding them.

Globulin and hematin. — A grain of hematin and a grain of

CHOICE OF FOOD IN AMEBA 223

globulin were laid close together in the path of an A. proteus —
67. As the ameba moved forward a pseudopod was thrown out
on the left toward the globulin-hematin. A food cup was formed
and both substances were ingested. There was no period of
rest, the ameba moving on in the same direction. The hematin
remained in the ameba's body for a considerable time there-
after. A little later another grain of hematin and one of glob-
ulin were laid in the ameba's path — 78. The ameba moved
into contact with them and carried them up on its back — 82,
and then formed a food cup extending upwards and ingested
them. The ameba remained comparatively quiet for a short
period, then moved on in the original direction. Another grain
of hematin was then laid with a grain of globulin some distance
ahead of the ameba — 87. The ameba moved forward toward
the test objects and when within about seven microns of them —
91, 92 — the tip of the main pseudopod spread out and formed
a food cup over them just as if they were moving organisms
like flagellates. The food cup was completed and the hematin-
globulin taken into the protoplasm as on the two previous
occasions. The ameba quieted down for a few minutes, then
moved off about thirty degrees to the left of the original direction.

No attempt was made by the ameba in any of these experi-
ments to separate the globulin from the hematin and to ingest
the globulin only, leaving the hematin behind. But it is to be
noted that in all these experiments the globulin grain was much
larger than the hematin. The most remarkable incident is the
ingestion in a food cup — 93, 94 — where the food cup was formed
before the ameba had come into actual physical contact with
the hematin or the globulin.

Globulin and uric acid. — A large grain of globulin and a small
grain of uric acid were laid close together in the path of the
ameba used in the experiment with hematin and globulin recorded
above — 129. The ameba moved forward into contact with the
globulin-uric acid and ingested them without the formation of
a food cup. The ameba remained quiet for about five minutes,
after which it moved off in the original direction.

Globulin and carbon. — A small grain of carbon was laid on a
larger grain of globulin in the path of an active A. proteus—
138. The ameba moved forward in Y-form with the carbon-
globulin lying between the prongs. The right prong finally

3

224 A. A. SCHAEFFER

became the more active and in its forward movement dragged
the retreating left prong into contact with the globulin-carbon —
147. A pseudopod was then sent out against the test substances
which upset them so that the carbon lay between the ameba and
the globulin. A large food cup was then formed over both
substances — 149. The carbon grain was soon pushed out of the
food cup while the globulin was taken into the protoplasm.
Just how this happened could not be determined, but it was not
" accidental."

Another grain of carbon together with one of globulin of
about the same size, were laid in the path of an A. proteus — 151.
The ameba moved forward to the right of the test substances —
153. A side pseudopod was then sent out to the left. It passed
the carbon on the left. A small pseudopod was then sent out
into contact with the globulin grain — 157. A large food cup was
then quickly formed over the globulin-carbon — 158 — but pre-
sently the carbon was pushed out while the globulin was taken
into the protoplasm. The details of the separating process again
could not be observed.

A grain of globulin was then placed on a large grain of carbon
in the path of a large ameba belonging to the proteus species —
159. As the ameba moved forward a pseudopod was thrown
out on the left — 160, but it was soon withdrawn — 162. The
main pseudopod flowed on past the carbon-globulin and then
turned sharply to the right — 164. A small Y-shaped pseudopod
was sent out toward the test substances as the ameba moved
away, but it was retracted before it had covered more than
half the intervening distance — 164. The grains of carbon and
globulin were then shifted — 165. The ameba moved past them
a short distance — 168. The tip of the main pseudopod then
turned to the left and moved into contact with the test sub-
stances. The carbon-globulin grains which were sticking together
slightly, were rolled around a few times by the ameba. A food
cup was then formed with the test substances lying, not in the
food cup, but in the mouth of it — 172. Half a minute later
another food cup was formed over them, but the ameba did not
fuse the free ends of the cup until four minutes after forming
it — 172 . The ameba then quieted down for about thirty-minutes,
during which time but very little movement could be observed;
nevertheless the globulin and the carbon became separated from

CHOICE OF FOOD IN AMEBA 225

each other, the carbon lying in the anterior half and the globulin
in the posterior half — 178. The ameba then moved off in the
direction opposite to the original, but so that the carbon should
at once occupy a position at the posterior end. A few minutes
later the carbon was excreted. Ten minutes after the ameba
had enclosed the carbon and the globulin in the food cup, another
normal food cup was formed while the ameba was lying quiet,
but there was no solid substance in it, nor was there anything
in the vicinity to cause its formation. The stimulus producing
it must have been internal.

This is a remarkable series of experiments. A degree of pre-
cision in food discrimination is disclosed in these observations
which was altogether unsuspected. The separating process in
the first two experiments is not understood. Not all the details
necessary to an understanding of the process were observed.
The ameba reacted like a higher animal might if reduced to
ameboid form. The last experiment clearly indicates that the
ameba was ' aware ' of the presence of a particle that was
not food lying close to one that was food. The way in which
the first food cup was formed clearly shows this. The long
delayed fusion of the free edges of the food cup indicates the
effect of the disturbing carbon. The long rest of thirty minutes
also is unusual. The change in the direction of locomotion was
doubtless caused by the position of the carbon. And the for-
mation of the empty food cup during the resting period was in
some way incited by the disturbing carbon, but just what con-
nection there might be between the two remains unknown.

Gluten and glass. — In the path of an A. dubia that had just
previously ingested pieces of agitated glass, was placed a piece
of glass and a piece of gluten lying close together — -179. The
ameba moved into contact with the gluten and the glass — -
184-186 — and then carried them up on its back, but there was
no attempt to ingest either of the substances.

Globulin and Coleps hirtus. — A small mass of metallic iron was
temporarily attached to a grain of globulin and slightly agi-
tated with an electric coil. The Amoeba dubia near which the
globulin lay, nevertheless treated it with indifference. A coleps
then came along, and while hovering over the globulin, both
the globulin and the coleps were ingested by the ameba in a food
cup thrown out from the posterior end — 190. Five minutes after

226 A. A. SCHAEFFER

the formation of the food cup the coleps stopped moving, and at
the same time the water disappeared from the food vacuole.
The coleps became separated from the globulin, and about
twelve minutes after the food cup was formed the globulin was
excreted.

The results of these experiments leave no doubt of the fine
sense of discrimination which ameba is capable of exercising.
Not only does ameba discriminate before coming into contact
with objects, but also after they are in the food cup and even
after they are imbedded in the protoplasm. The experiment
recording the ingestion in the same food cup of globulin and a
coleps indicates this, for after these substances were imbedded
in the protoplasm a coleps was preferred to globulin.

The experiments with egg albumin and globulin, and egg
albumin and silicic acid, show clearly that the presence of a
substance in solution only is not' sufficient to attract ameba,
nor to cause ingestion, but that the substance must be actively
diffusing from a definitely localized region. 1

THE EFFECT OF MECHANICAL STIMULATION

It became evident as the feeding experiments went on that
movement of food objects is an important factor in ingestion,
and sometimes indeed a determinining factor. A number of
experiments were then projected to see especially what the
effect of water vibrations is upon ameba, and whether vibrations
of themselves are capable of causing a definite change in behavior.

It is to be regretted that but few drawings can be presented
in illustration of the gradual change in behavior that is pro-
duced by mechanical stimulation; but it was impossible to make
a series of drawings for each experiment, as was done in the
other cases, for the drawing hand of the experimenter was em-
ployed in vibrating the needle. Consequently with the excep-
tion of one experiment — 194-197 — which is represented by
memory drawings made immediately after the experiment, only
the final stages are illustrated by camera lucida figures. The
different kinds of objects used in these experiments are fairly
representative. Several kinds of insoluble indigestible sub-
stances, as well as food substances, were employed.

3 There is some evidence here that two particles of different degrees of attrac-
tiveness when encountered separately, are reacted to, when lying close together,
as one (the more attractive) being attractive and the other (the less attractive)
repulsive.

CHOICE OF FOOD IN AMEBA 227

The effect of vibrating food particles. — A small fragment of an
ameba was placed in the path of an A. proteus. Pseudopods
were sent out toward the fragment, but all of them were soon
withdrawn — 198-205. The fragment was then allowed to roll
down the side of the ameba, whereupon ingestion followed at
once — 206. It is certain that the mechanical stimulation pro-
duced by the rolling fragment hastened ingestion.

An A. dubia was then tested ten times with grains of globulin,
but only one was eaten, the seventh. On the eleventh trial a
piece of globulin was laid against a newly formed pseudopod
and agitated with the point of a glass needle. A food cup was
promptly formed and the globulin apparently ingested — 207.
Five minutes later the ameba moved on, leaving the globulin
behind in about the same place — 210. It had not been com-
pletely surrounded by protoplasm. The formation of the food
cup in this case was caused by the vibration of the globulin.

Another A. dubia that reacted rather indifferently toward
globulin, although the globulin was finally eaten, and quite
indifferently toward ovalbumin, iron and fibrin, was tested with
a grain of fibrin agitated with a glass needle. A food cup was
formed and the fibrin apparently ingested at once in a food
cup containing a large quantity of water — 211. The ameba
became quiet for a minute and a half and then moved off in
the original direction, leaving the fibrih behind, four minutes
after the formation of the food cup — 212-216. The fibrin was
probably not completely surrounded by protoplasm. The for-
mation of the food cup was caused by the vibrations of the
fibrin.

The effect of vibrating alga filaments. — One of the most illu-
minating experiments bearing on the general problems of mechan-
ical stimulation is the following: A raptorial ameba which had
reacted indifferently to capillary tubes filled with peptone solu-
tion was gently stimulated mechanically by the free ends of
three thin oscillatoria threads wrapped around a glass needle.
Stimulation at the posterior end produced a prompt reversal
of streaming. When stimulated at the newly formed posterior
end, streaming was again reversed. In this way the direction
of the protoplasmic current was reversed eight times in suc-
cession, the current always moving toward the point of stimu-
lation, without bringing the ameba away from its original location.
Figure 217 shows the position of the ameba while the proto-

228 A. A. SCHAEFFER

plasmic current was changed four times. The sides of the
ameba were then stimulated at various places with the alga
threads and at each point a pseudopod was thrown out. If
the stimulating was done properly, food cups were begun at
these pseudopods. Partial or complete food cups could be
formed at will, depending upon the character and continuance
of the stimulus. In short, all the ordinary items of positive
reacting, from simple movement toward a stimulus to the prompt
and complete formation of a typical food cup, could be produced
by varying the intensity, frequency, and character of the stim-
ulus. The mechanical stimuli proceeding from the alga threads
were probably the only ones concerned in producing the reac-
tions described, for the bottom of the dish was strewn with
broken oscillatoria filaments, over which the ameba frequently
flowed without changing its behavior. It was only when the
filaments were agitated in contact with the ameba, or nearly in
contact with it, that the ameba responded with a positive reac-
tion. The efficiency of this sort of stimulation lies doubtless in
the fact that the great flexibility of the alga threads permits
one to simulate to a high degree the movements of small living
organisms.

An experiment similar to this one gave almost identical results.
A large binucleate A. proteus that had reacted indifferently to
globulin grains or eaten them imperfectly was stimulated with
the alga threads and the response was the formation of a com-
plete food cup — FCi, 220. The ameba was stimulated again
and another food cup was started, FC 2 . A grain of globulin
was then dropped into the partly formed food cup and upon
further stimulation the food cup closed completely. The glob-
ulin was retained for over two hours, when observation was
terminated. The first food cup which was formed, it is inter-
esting to note, remained almost undiminished in size for over
two hours. It remained much longer than if it had contained
food.

In another experiment of a similar nature the glass needle
was used without any alga threads or anything else on it — 194.
The point of the needle was agitated near, but not in contact
with, an A. proteus. A pseudopod was promptly started in
the stimulated region — 195, and it attained to considerable size
and finally formed itself into a food cup, which was, however,

CHOICE OF FOOD IN AMEBA 229

not completed — 197. The needle at no time touched the
ameba. This experiment, in connection with the preceding
ones, shows that a purely mechanical stimulus is sufficient to
set in operation the ameba 1 s feeding mechanism.

The effect of vibrating insoluble indigestible particles: Glass. —
A clean piece of glass was laid near an A. dubia and slightly
agitated with a clean glass needle — 226. The glass particle was
promptly ingested in a typical food cup with a large amount of
water. The ameba did not undergo a period of rest but kept
on moving forward and in six minutes the glass was excreted —
232. It had been completely surrounded by protoplasm. Five
minutes later the same piece of glass was again agitated with
a glass needle — 234. Again the glass was ingested in a typical
food cup. Three and one-half minutes later the glass was
excreted — 237. Another trial with the same piece of glass pro-
voked the formation of a food cup until it was about three-
fourths completed, then it was retracted — 238. Two further
trials produced visible responses, but no complete food cup was
formed. The ameba became increasingly indifferent toward the
artificial stimulation with each succeeding trial.

Silicic acid. — An A. dubia was stimulated with oscillatoria
threads until a food cup was partly formed, then a grain of
silicic acid was placed in it and stimulation with the alga fila-
ments resumed — 240. The food cup closed up apparently com-
pletely, for three minutes after the closing of the cup the silicic
acid was actively thrust out as a piece of undigested food material
might have been thrown out — 244.

Carbon. — A fragment of purified carbon was laid on the back
of a raptorial ameba — 245. The carbon was raised up by the
protoplasm immediately underneath it, but very soon there-
after it rolled down the side of the ameba. A large pseudopod
was then sent out over the carbon but no distinctive food cup
was formed. When the ameba was forcibly rolled over, it was
found that the carbon adhered to the ameba. No attempt had
been made to surround it after the psuedopod was laid over it.

Indigotin. — A piece of indigotin was laid in front of an A.
dubia and agitated with a glass needle — 247. The indigotin
was eaten by means of a typical food cup containing a large
quantity of water. The ameba moved on in the original direc-
tion. Nine minutes after the indigotin was ingested, it was

230 A. A. SCHAEFFER

left behind. It is uncertain whether the indigotin had been
completely ingested.

The experiments recorded in this section demonstrate con-
clusively that amebas respond positively to mechanical stimula-
tion. Food objects when mechanically agitated, are more readily
eaten than when lying quiet, and in many cases food objects are
not eaten at all unless they are in motion. Further, insoluble
objects, like glass, are not eaten when lying quiet, but when
agitated properly they are readily eaten. It is not necessary
for a solid object to come into contact with the ameba in order
that a food cup may form; vibrations of the water are quite
sufficient to produce the feeding reaction. A chemical stimulus
is therefore entirely unnecessary in order to set off the feeding
reaction. Movement is not only a contributing factor but it
is in itself an efficient factor in calling forth the feeding process.

It will be noted that when vibrated particles are eaten, they
are seldom retained for more than a few minutes, whether they
are of food value or not. The reason for the speedy excretion
of such particles is that the chief, if not the only quality of the
particle which mcited the feeding process, movement, disappears
when the food cup closes, for mechanical vibration is then no
longer possible. Since the most attractive quality suddenly dis-
appears, the feeding process stops before it is entirely completed.
The formation of a food cup resembles to some extent a reflex
act: when an agitated particle of glass is eaten, the formation
of the food cup continues for some time after the vibrations have
ceased. The formation of food cups is due to a racial habit
which is guided to a greater or less extent by circumstances
attending the formation of a cup at any particular time. Since
circumstances vary greatly, one observes great diversity in the
behavior toward particles which produce positive reactions.

Whether the swallowed particle is retained or speedily excreted
is largely independent of the character of the feeding reaction.
The food cup may have been perfectly formed and yet the par-
ticle inciting it may be speedily excreted; or on the other hand,
the food cup may have been very imperfect, but the particle
may be retained. The retention of a vibrated particle in the
ameba seems to depend on qualities other than those which
caused ingestion. What such qualities are cannot be definitely
stated. The composition of the particle seems to have relatively

CHOICE OF FOOD IN AMEBA 231

little to do with it, for food particles like globulin are frequently
thrown out at once, while glass is always thrown out ; and choles-
terin, presumably an indigestible substance, was in one case
retained for over and hour and a half. The condition of hunger
in the ameba is one of the most important factors. But what-
ever the factors are which control retention, it is certain that
they are not the same as those conditioning ingestion.

DISCUSSION OF EXPERIMENTAL RESULTS

Choice of food. — What is choice? This is a very troublesome
concept in science. We need some such word as choice in
describing the reactions of animals in order to avoid hopelessly
confusing circumlocutions, and yet it seems impossible to give
a satisfactory definition of choice from an objective point of
view. Since I wish to discuss particularly the phenomena of
choice of food in this paper, I shall undertake to explain how
I use this word.

Choice has been used to label a process by which a certain
state of matter in a system is brought about. More restrictively,
it has been used to designate exclusively the end result of such
a process, the state of matter after the process has ceased. It
is obvious, without illustration, that these two meanings of the
word are quite distinct from each other and that they can con-
veniently be qualified by using, respectively, the phrases " pro-
cess of " and " result of " choice. The word choice is also some-
times used to describe such processes as, for example, the action
of a magnet upon a mixture of sand and iron filings. This is an
unfortunate use of the word and two objections may be urged
against this usage: first, an equally intelligible description of the
action of the magnet can be given without the use of the word
choice; second, choice might with equal propriety, be applied to
every movement of matter in the universe. It is therefore a
hindrance to clear thinking to use the word choice as descrip-
tive or nominative of such processes as these.

Again, the concept of choice is frequently restricted to pro-
cesses observed in organisms, that is, to conscious processes.
This view, although of the greatest interest, is very difficult to
consider from our present objective experimental point of view,
since it would be necessary first to determine whether an animal
whose power of choice is in question, possesses consciousness.

232 A. A. SCHAEFFER

The concept of choice is thus seen to occupy a difficult posi-
tion in science. If a process in organisms can be described as
a sequence of events, we may adopt the view that we do not need
the word choice in the description of these events; if, on the
other hand, a process of choice cannot be described as a sequence,
we cannot from the point of view of objective science use a meta-
physical concept — consciousness — as a directive force to explain
the phenomenon, but must conclude that some of the events in
the chain of sequences are still unknown.

Perhaps the chief objection to the use of the concept of choice
is that it insistently calls to mind a relation existing between, for
example, an ameba and some particles which it senses. The
relation between the ameba and the particles is made to occupy
a more important position in the mind than either the ameba or
the particles by themselves. The concept is strongly metaphysi-
cal, for the instant one seeks for a relation between phenomena
one steps into the borderland of metaphysics. A thousand and
one other concepts without w r hich we could not possibly get
along in science are just as truly metaphysical, but for various
reasons we have fallen into the habit of not noticing this
characteristic.

Again, the concept of choice originates almost wholly from
subjective material, and in ordinary usage it has first of all a
subjective meaning. It is supersaturated with anthropomorph-
ism. Let us say that a man chooses an apple from among
some oranges. How would these phenomena be related objec-
tively without connoting a subjective process? A kinemato-
graphic record even, without any words at all, would be certain
to suggest a subjective process. But although we ordinarily
thus think of the subjective process, nevertheless if the same
events should be made known to us in a paper on animal or
human behavior, we would disregard the subjective meaning
and center attention only on the objective phenomena. It is,
in short, the meaning we are after and the point of view, not
the words merely.

To some investigators of behavior such words as selection,
choice, etc., are anathema. They would avoid them as they
would the plague. But they themselves do not wholly suc-
ceed in avoiding - them. How would the idea: The ameba
selects its food — be rendered on this view? This idea could
conceivably be presented, as in the case of the man selecting

CHOICE OF FOOD IN AMEBA 233

an apple from among some oranges, by a kinematograph with-
out the use of a word. But would the general impression on
the mind be any different, i.e., would it be devoid of subjective
elements? It is just as anthropomorphic to say that the ameba
eats globulin and leaves carbon, as to say, the ameba expresses
choice between globulin and carbon (eating the former and not
the latter).

Briefly then, the word choice serves for two purposes. First,
it is used as a hypothesis, as a tool of research. It postulates
a certain kind of relation which is to give direction to investiga-
tion. All except haphazard investigation proceeds in this man-
ner. Second, its use is indispensable for brief and intelligible
description; but any subjective connotation which the word may
call forth should be disregarded by the reader unless the author
specifies definitely otherwise. It is in this latter sense that the
word is used in the descriptions in this paper.

In a former paper ('10) I demonstrated that the ciliate stentor
is capable of exercising very nice discrimination among the
various particles which are carried to its mouth by its cilia.
Not only is the discrimination between food and indigestible
particles (excepting carmine) very precise, but even among food
particles themselves, certain organisms being eaten and others
rejected. This is particularly noticeable when the stentors are
partially satiated.

Now ameba discriminates with equal or perhaps greater pre-
cision. Excepting carmine, no indigestible substances are eaten
unless agitated. Of the various things eaten, living organisms,
such as flagellates, coleps, etc., come first in point of preference.
Then comes globulin, grain gluten, carmine, and tyrosin. Less
attractive than these are: aleuronat, fibrin, lactalbumin, oval-
bumin and keratin, and peptone in solution. Digestible sub-
stances that are eaten only occasionally are soluble egg white,
solid egg white, uric acid. No starch, lecithin, silicic acid,
carbon (excepting in one case), sodium chloride, iron, choles-
terin, etc., were eaten unless agitated. Sand grains were never
eaten under my observation, and I have never seen " numerous '
sand grains inside of amebas from wild or laboratory cultures.
It will be recalled that sand grains are frequently referred to in
text books and elsewhere as " typical " contents of the ameba's
body.

How can these preferences be explained? As far as can be

234 A. A. SCHAEFFER

told at present there is no single quality possessed by all the
substances which ameba eats, which might be looked upon as
the basis of selection; nor do the refused substances possess a
common quality which causes amebas to leave these substances
alone.

Excepting carmine, all the substances which ameba readily
eats are (presumably) digestible, though only two were actually
tested in this respect : globulin and grain gluten. But a number
of substances which are digestible, such as for example gelatin,
egg albumin, fibrin, etc., are eaten only very occasionally or
not at all. Of the lifeless solid substances, those which are the
more soluble (excepting gelatin and egg albumin, which are
very soluble, and globulin which is said to be insoluble) were
the more readily eaten. But it is possible that in the case of
globulin impurities were present, or that the salts in the water
caused slight solubility. The attractiveness of carmine and of
lead oxide, and the general indifference toward zein, fibrin,
keratin, etc., might be due to the solubility of the former and
the insolubility of the latter. Tyrosin is the only rapidly
dissolving substance eaten. It is very attractive, but it seems
that its rapid rate of solubility is slightly disagreeable or injuri-
ous. But there are still several items of behavior that cannot
be explained even if an essential part of the basis of selection
among lifeless substances should be the rate of solubility. Thus,
why should the soluble proteins: egg albumin and gelatin, be
passed by with indifference? Or what explanation could be
assigned for the fact that solids such as tyrosin and carmine
call forth feeding reactions very readily when these substances
are in the actual process of dissolving, but if a capillary tube
filled with solutions of these substances, is presented, the feeding
reaction is not produced? It is clear that neither solubility
as such nor the rate of solubility determines in all cases whether
a lifeless substance shall be eaten or rejected.

The opinion has become quite general that the chemical con-
stitution of substances determines whether an animal will eat
them or not. This view, however, has no experimental support;
but on account of its general acceptance, especially as applying
to unicellular forms, it may be profitable to examine it in some
detail.

Just how can a substance affect a sense organ? Apparently

CHOICE OF FOOD IN AMEBA \ 35

only in two general ways: its molecules or ions may combine
chemically, temporarily or permanently, with a part of the sense
organ; or, some sort of physical energy radially propagated
from the sensed particles may produce physical changes in the
sense organ. If a substance affects a sense organ by. uniting
with it chemically, such sensations as might result therefrom
should be related in some way with the chemical constitution
of the substance, and the sensations should vary in some way
as the composition of the stimulating substance varies. Such
a sense organ would mediate true chemical sensations. It
would seem that all the elements of at least a few compounds
should be represented in consciousness, by a chemical sense
organ, when they are tasted or smelled. But this never happens
in man. The taste or smell of a substance of a given concen-
tration, on homogeneous sense organs, is always a simple sensa-
tion, such as proceeds from a simple stimulus. There is no
evidence from the psychological side, as it is recognized that
there is none on the physiological, that organs of taste or smell
sense the chemical composition of substances. But on the
other hand, there is some evidence that physical qualities are
sensed by the distinctive organs of taste and smell as is shown
by the possibility of stimulating the taste buds by electricity
so that there are produced sensations of sweetness, acidity, etc.,
depending upon the particular buds stimulated. It is possible
therefore to stimulate a sense organ of taste by means of
chemicals in solution as well as by electricity, a physical stimulus.
But the effect of a chemical on a sense organ is not at all the
same thing as a chemical effect on a sense organ.

But this distinction is not always clearly drawn.

Metalnikow, in his very interesting and extensive paper on
the feeding habits of Paramecium, as an important example,
does not make very clear the distinction between reactions to
physical and to chemical qualities, although he asserts that
the chemical nature of a substance determines whether it will
be eaten by Paramecium. Let us therefore examine the exper-
imental results and arguments of Metalnikow, in the light of
what was said in the preceding paragraphs, to see whether he
is justified in stating that Paramecium selects its food upon a
chemical basis.

Paramecia eat powdered glass, sulphur, chalk, aluminum,

236 A. A. SCHAEFFER

sepia, as well as egg yolk, milk, olive oil, and other digestible
substances (see p. 398, table 1.), but he says:

Toutes ces experiences nous montrent avec certitude que
l'absorption par les Infusoires des corpuscles suspendus dans
l'eau et la formation des vacuoles digestives depend de la nature
chimique de la substance dont sont formes ces corpuscles. Sans
aucun doute les Infusoires sont capables de distinguer les
differentes sortes de nourriture qui peuvent se trouver a leur
portee.

II est interessant de verifier a present si les infusoires peuvent
absorber les differentes substances toxiques insolubles dans l'eau.
A cet effet, les Infusoires ont ete nourris de differents sels in-
solubles de mercure et d'arsenic, et j'ai ete etonne de voir que
ces sels sont assez vite absorbes par les infusoires qui arrivent
meme a former quelques vacuoles digestives; ces infusoires
perissent ensuite bientot. Mais il ne faut pas oublier qui meme
l'homme et les animaux superieurs ne sont pas toujours capables
de distinguer les substances toxiques, si ces substances sont
depourvues de mauvais gout (pp. 401-402).

It is evident that Metalnikow speaks here of selection by the
feeding mechanism, and not of histonic selection, that is by the
digestive mechanism of the tissues.

It is difficult to see how paramecia can be said to distinguish
between substances on a chemical basis when indigestible and
insoluble substances as well as digestible and soluble, are eaten
to about the same extent.

Choice based upon chemical qualities is qualitative; each
chemical substance must be conceived of as acting specifically.
According to this conception, when the sense organ is stimulated
the quality of the sensation is exactly expressible; it cannot
vary excepting as to its amplitude or continuance. There can
be no doubt as to the exact nature of the stimulating object.
We would therefore expect an animal capable of choosing on
this basis, to choose with great precision. We would certainly
expect much more precise selection even if the power of selection
was very crude, than is shown in Metalnikow's first table. For
according to this table glass and sulfur are eaten apparently
as readily as milk or starch and eighty per cent as freely as
olive oil (table 1, culture C).

Organismal selection in Paramecium seems therefore to be

CHOICE OF FOOD IN AMEBA 237

extremely crude; in fact it can hardly be said to exist at all
under ordinary conditions.

But even if Paramecium discriminated between digestible and
indigestible substances, the supposition that it did so on a
chemical basis is open to question. For it is assuming a great
deal when it is said that Paramecium becomes aware in some
way of the diverse chemical structure of starch, milk, various
bacteria and bacilli, leukocytes, yeast cells, egg yolk, olive oil,
albumins, etc., while these substances are passed back to the
mouth by the cilia of the gullet. Now in order that Paramecium
may become aware of the chemical nature of these several
substances, new compounds must be formed in the sense organs
of the Paramecium (the cilia?) involving a permanent or tem-
porary union of a part or the whole of the molecule of starch,
albumin, oil, etc., with some structure in the sense organ, to
form, in each case, a definite and characteristic compound
which would then serve in some way to set into operation the
ingesting mechanism. What the chemical equipment of a sense
organ would have to be in order that compounds might be thus
formed with all the different substances which Paramecium eats,
it is difficult to conceive. Metalnikow does not attempt to
give an explanation of how selection on a chemical basis might
operate; nor to my knowledge do any of the other writers on
protozoa, although several refer more or less casually to the
selection of food as based on its chemical nature. That it is
unnecessary to explain choice of food upon a chemical basis
in Paramecium or in any of the other protozoa whose feeding
habits have thus far been investigated, will be shown by examining
the process of selection in ameba.

Recently Lund ('14) published an important paper on the
feeding reactions of the large ciliate, bursaria. In this paper
Lund discusses the selection of food in bursaria, especially as
to whether it rests on a chemical or on a physical basis, and
in this connection refers to a previous paper of mine treating
of similar questions about the feeding behavior of stentor. It
seems however, that Lund did not quite understand my con-
ception of the difference between selection of food on a chemical
basis and on a physical basis, doubtless because I did not go
into the details of the difference as I conceive it. I have
examined his paper carefully but, although he insists that

238 A. A. SCHAEFFER

bursaria selects its food on a chemical basis, I have nowhere
found a discussion in his paper which sets forth just what he
conceives to be a chemical basis as distinguished from a physical,
in so far as the selection of food is concerned. I have, therefore,
found it of interest to examine his paper in some detail, for
it is absolutely essential that the difference between selection
on a chemical and on a physical basis be clearly drawn, before
we can say that an animal selects its food on either basis.

Lund fed bursaria with yolk of hen's eggs untreated, and
stained with various chemicals, employing essentially the methods
I used in my work on stentor ('10), and Metalnikow ('12), on
Paramecium.

Yolk treated with various dyes is eaten less readily than
untreated yolk. If NaOH, HC1, saffranin, etc., are added to
the culture solutions in varying quantities, the yolk is eaten
in decreasing quantity in proportion as the toxic substance is
increased in concentration. Temperature also has a similar
effect; with a rise of temperature from 0° to 35° C, there is a
corresponding increase in the amount of yolk eaten. Beyond
35° C. the temperature becomes injurious. In addition to these
experiments, Lund mentions some observations to the effect
that carmine, india ink, aluminium, carbon black, cinnibar, etc.,
are eaten by bursaria (p. 43).

It appears then that bursaria stands between stentor and
Paramecium (and close to Paramecium) as far as concerns its
capacity to discriminate in feeding.

From the results of his experiments, Lund expresses agreement
with Metalnikow ('12) in so far as the basis of selection is
concerned.

But it was shown in the preceding pages that Metalnikow's
conclusion that food is selected on a chemical basis is not in
agreement with all the facts; and the same criticism to which
Metalnikow's conclusions are open, also apply to Lund's. It
will therefore not be necessary to go over this ground again.
It should be noted however, in this connection that Lund dis-
misses with a single paragraph, as if they did not bear at all
on the question of selection, all the observations he made on
the eating of chemically inert substances, such as aluminium,
carbon, etc. But it is precisely these observations which most
strikingly contradict his notion that selection is made on the
basis of the chemical composition of the substances. How can

CHOICE OF FOOD IN AMEBA 239

one explain the selection of insoluble substances such as carbon or
aluminium, which are eaten, on a chemical basis''!

The experimental results which form the basis of Lund's
conclusion admit of another interpretation. The fact that more
and more yolk was eaten as less and less of NaOH, HC1, saffranin,
etc., was present, does not of course indicate active selection;
the presence of toxic substances affected the general bodily
condition so that feeding was to some extent suspended. It
cannot be inferred that the bursarias ' ' tasted ' ' with the food
selective mechanism the toxic substance, and on this account
refused to eat less as this substance was more and more con-
centrated. The experiments with temperature show a feeding
gradient that corresponds almost exactly with that obtained in
those experiments where toxic substances were employed in
varying concentrations. Clearly therefore, there must be other
evidence to show that selection is based on chemical constitution ;
for the observation that chemicals in varying concentrations
prevent feeding to a greater or less extent, does not mean that
these chemicals were 'tasted' by the mechanism of food selection
any more than that the various temperatures, or the electric
current, or the mechanical stirring, were tasted. In other words,
the feeding mechanism in bursaria, as in many other animals,
is affected by stimuli affecting the general bodily condition, as
seems to be the case when stimulated by agitating mechanically
the medium in which the bursarias live, or by passing electric
currents through the medium, etc. In these cases