museum novitates

AMERICAN MUSEUM NOVITATESNumber 1253

Pubfished byTHE AMERICAN MUSEUM OF NATURAL HISTORY

New York CityApril 8, 1944

A UNIQUE CASE OF CIRCULAR MILLING IN ANTS, CONSIDEREDIN RELATION TO TRAIL FOLLOWING AND THE

GENERAL PROBLEM OF ORIENTATION'

BY T. C. SCHNEIRLA2

Were insects unable to make and tofollow trails, the problems of, Victorygardeners, farmers, economic entomolo-gists, and housewives would be consider-ably reduced. For in almost every insectorder from roaches to the social insectsthere are at least a few species that reach

or return to a food supply, food plants,home quarters, or other places by virtueof their ability to make and to follow trails.It is clear that an understanding of trailmaking and trail using is desirable forboth theoretical and very practical rea-sons.

COLLECTIVE FORAGING AND CHEMICAL TRAILS

There are many different forms of trail-ing in insects, from the type of trail that isfollowed in a relatively rigid, stereotypedmanner to routes th'at are used in a com-plicated, delicately varying manner, oftenby a single individual alone. Among theants, the simplest kind is the collectivetrail. Oti such trails the workers of a givenant colony are able to reach a food place inlarge numbers, with a minimum of delayand difficulty once the "track" is estab-lished. In this paper we shall be con-cerned with collective trails in particular.One of the least complicated ways in

which such a trail may be set up wasdemonstrated by Eidmann (1927), usinga Myrmica colony in a laboratory nest.With a new paper floor in place, one in-dividual was permitted to find sugar waterin a distant compartment of the nest, andfeed upon it. Shortly after the finder hadreturned to the home compartment, anumber of ants was observed to cross thefloor of the intervening compartmenttoward the food place, taking the sameroute the finder had used in returning,even in its detailed turnings. The simplest

1 A contribution of the Department of AnimalBehavior, the American Museum of Natural History.

2 The American Museum of Natural History andNew York University.

explanation was that the newcomers werefollowing a chemical trail left incidentallyby the first ant. Confirmation of thisinterpretation was found in a further testwhen a clean floor was substituted for theused one after the finder ant had crossedfrom the food to the home base. Thenother ants came out as before, evidentlyaroused by the returned food carrier, butthis time they were quite unable to followher path, instead wandering about andreaching the food only by chance.

Other studies throw light on how suchtrails are made. In work with a Tapinomaspecies, Santschi (1911, 1930) noticedthat in returning from the food the firstfinder rubbed her gaster against the floor,evidently an outcome of her excitementwhich resulted in a trail of some kind. In-specting the ground in her wake with ahand lens, Santschi actually saw tinydroplets in the line she had taken, suggest-ing to him that the products of the ant'sanal gland had been deposited when thegaster was rubbed against the floor. Al-though he thought this was a "deliber-ate" procedure on the part of the finder,a preferable interpretation is that the lower-ing of the gaster and opening of the anal-gland orifice occurred merely as incidental

AMERICAN MUSEUM NOVITATES

(i.e., as direct reflex) parts of the ant'sgeneral excitement, no more intentionalthan the release of sweat from the skinglands of an excited human individual.Similarly, v. Frisch and Rosch (1926)have shown that when a honeybee finds anew food place, in her excitement theproducts of the "scent glands" are re-leased, greatly increasing the chance thatother bees will find the place.The present writer has observed that

the findings of Eidmann and Santschi maybe applied directly to the case of the smallimported ant, the reddish yellow myrme-cine Monomorium pharaonis, a commonkitchen nuisance which displays a singular

ability to establish collective trails to newfood places. A pharaonis worker, uponleaving a place where she has imbibed orpicked up food, generally lowers her gasterso that its tip rubs against the substratumas she moves, thereby leaving chemicaltraces on the return path which new-comers are observed to follow, usuallywith relative ease. As a result this tiny4nt, wandering individually over rela-tively large distances around the nest,appears with surprising quickness in con-siderable numbers at any newly discoveredsource of food. This characteristic un-doubtedly has greatly assisted the almostunmatched distribution of the species.

CHEMICAL TRAILS IN THE RAIDS OF ARMY ANTS

A strikingly different mode of estab-lishing collective routes is found in theEciton species, the army or legionary antsof subtropical and tropical America. Fewants follow their trails in a more slavish andstereotyped manner than do the Ecitons.For example, the raiding columns of E.hamatum, the golden yellow terrestrialpillager, excel in the rapidity with whichnew offshot columns are formed over un-raided ground (Schneirla, 1938). If theroutes which spring up and survive longestduring a day of raiding are mapped in de-tail, the result is a tree-like figure. Thetrunk line of this trail system leads fromthe bivouac or home site, with more andmore frequent branching as the distancefrom the bivouac increases. At the verytwig end of the system, usually as much as150-200 meters from headquarters at theend of a given day, the newest trailsgrow outward into unraided ground fromwhich insect booty is snatched. Eachtwig column ends in a relatively smallgroup of ants, a few dozen of them in anarea usually not more than 25 cm. wide,behaving in a rather peculiar manner asthey push onward. Hugging close to theground with her antennae flitting close toits surface in quick spasms of wasp-likevibratory movement, each newcomerto the group meanders forward hesitantlya few centimeters over fresh terrain, then

quickly turns and rejoins the base column.In this way the trails are extended on-ward, by a relay process in which no givenant or ants have the "lead" for very longbut with many ants participating succes-sively in the trail-blazing process.

Eciton trails frequently follow vines, thetop edges of tree roots, the worn indenta-tions of peccary trails, and other topo-graphical features along which the antsmove readily in antennal contact with asurface. However, a given trail persistsand is followed by further ants because it ismarked chemically, first of all by thepioneer ants (presumably with theirglandular products) as they move excitedlyforward, each contributing a limited ex-tension of the track into new terrain.Experience with a given route is not neces-sary in the Ecitons. Any member of thecolony can readily follow the trails of eachpart of the raiding system without havingpassed over the particular route previously.Thus in special tests, when workers are re-moved from the bivouac and placed besideany trail of the day, they readily enter andrun along the route, whether or not otherants are traveling upon it at the time. Orif Ecitons are held captive overnight whenthe colony moves to a new locality, theyare able to follow any fresh trail of thenext day when they chance upon it afterbeing released. The chemical traces

2 [No. 1253

CIRCULAR COLUMN AND ORIENTATION IN ANTS

evidently remain in existence for some time,despite repeated washing of the ground bybeavy rains. This is shown by the factthat when colonies remain in the samebivouac location for as long as three weeks,a periodic occurrence in the terrestrialforagers at least, a long section of trailmay drop out of use for two or more daysafter it is originally formed, then reappearin use as part of a new raiding system.

Evidence for a chemical similarity in the trailsof different colonies of the same species is offeredby the ability of Ecitons to run a trail of anothercolony to whose zone they are transferred; how-ever, a certain chemical difference is indicatedby a noticeably more hesitant progress thanwhen on trails of their own colony. When atransferred ant encounters foragers of the strangecolony she is attacked repeatedly and finallykilled unless she flees the trail. Such facts, to-gether with the results of nestmate-recognition

tests in which the distinctive influence of "col-ony odor" has been demonstrated (Fielde,1902; Morley, 1942), indicate that ants ofdifferent colonies of the same species vary intheir iinmediate effluvia, as well as in the chemi-cal nature of the routes they make. Further-more, since preliminary tests indicate that anEciton worker is able to follow the trail of an-other species (e.g., transported E. hamatumworkers push hesitantly along a section of E.burchelli trail to which they have been trans-ported), a chemical similarity is suggested forthe routes of closely related army ant species.'The basis of trail-following thus would

appear to rest on the tendency to approachsources of the ant's own colony chemical orsimilar chemical patterns, and thus tomove forward according to the line of thischemical when on trails. What accountsfor the ability of the individual worker torespond in this highly adaptive manner?

THE ONTOGENETIC BASIS OF TRAIL FOLLOWING IN ARMY ANTS

In the case of the Ecitons it seems im-probable that the mere possession of ol-factory sensitivity brings with it this capac-

ity for trail following. After hatching,trail following requires a few days to ap-

pear, and circumstances do not justifyattributing the delay to maturation,i.e., simply to an improved sensitivity basedupon further growth of the olfactory tissuesthemselves. The fact that the callows are

able to run long distances when partici-pating in the nightly bivouac-changemovements militates against the possi-bility that it is merely muscular weaknesswhich holds them back in the daytime.The possibility that the irritating effect oflight may restrain them at first should bekept in mind; yet circumstances oppose

this as the main factor. Normally, E.hamatum workers become fairly able trailfollowers after four or five days fromhatching; however, the results of some

preliminary tests suggest that being withthe colony is essential for the improvement.After hamatum pupae had been kept away

from the colony for a period of five daysfollowing hatching, when placed indi-vidually on the raiding trails they seemedhelpless and inept in contrast to others of

the same brood that had lived with thecolony after hatching.

Normally, for some time after a newbrood of mature pupae is removed fromits cocoons by adult workers, these cal-lows (light-colored and readily distin-guished from regular workers) remain inthe bivouac without venturing upon theraiding trails to any extent. Then, after aday or two, they begin to appear on trailsin the daytime, and are observed in pro-gressively increasing numbers at pointsfarther from the bivouac. There is also anoticeable change in their behavior duringthis period. Upon the trails they moveslowly and hesitantly at first, unable tomove rapidly or to go very far in a givendirection without stopping. Repeatedlybumped and shunted about by the regu-lars, they generally pass along at the sidesof the column for limited distances,readily drifting into groups at trail-divisionpoints. Gradually their deficiency seemsto decrease, until after a few days when

1 Almost a virgin field awaits investigation in thebiochemistry of trail secretions in insects, and par-ticularly ants. To the writer's knowledge, the onlystudy of Eciton chemical is the work of Melanderand Brues (1906). These investigators found notraces of formic acid in whole body distillates ofvarious Eciton (Acamatus = Neivamyrmex) species ofthe southern United States. However, the outcomeof further tests suggested that the basic substanceprobably is leucin.

19441 3


deepening pigmentation has made theirexternal appearance difficult to distinguishfrom that of regular workers, their trailrunning likewise is close to the generalstandard.

In view of this evidence suggesting theimportance of experience, the preferablehypothesis seems to be one centeringaround the postulation of a rudimentarylearning process of the "habituation"type in which there develops an increasedreadiness to approach the colony chemical.In the colony, this chemical dominates en-vironmental stimulation from the time ofhatching; whether resting or feeding inthe bivouac the callow is stimulated by it,as indeed she was stimulated during heractive larval period. Thus it is very pos-sible that through a simple conditioningprocess, perhaps begun during the larvalperiod, the callow moves about readily inthe presence of this chemical and, moreimportant, turns toward increases in itsconcentration.' Gradually, as she movesabout more freely, the new worker driftsinto the trails in responding to the Ecitonchemical and to the movements of depart-ing foragers. Early bivouac life thus maylead directly into trail running, with theinitial hesitancy of the callow decreasingas she becomes better trained to runfreely forward in the presence of the colonychemical. This hypothesis is consistentwith a theory for social insects in generalwhich takes into account not only theoriginal equipment and physiological matu-ration of the individual, but also theimportance of early environment and earlyactivities for the development of socialreaction patterns (e.g., Wheeler, 1928a;Maier and Schneirla, 1935, chap. 7).One feature of early life which must

hasten the Eciton trail-habituation processconsiderably is participation in the bivouac-change movement of the colony. Regu-larly, when a new brood hatches, the E.hamatum colony shifts to a new bivouacsite at the end of each day (Schneirla,1938). At first the callows do not readily

1 Thorpe's researches (1939) with the ichneumonfly in particular have led him to adopt a theorybased upon the postulation of a chemically dominated6onditioning process governing the response of preda-tory adults to given prey upon which the individualhas fed as a larva.

join the drift of workers from the cluster;rather they leave the interior of the bivouacpassively when shunted along with out-.going groups of workers, and huddle nearthe exit until incidentally absorbed intothe exodus. Much confusion resultsfrom the tendency of callows to hesitate,waver, and turn around when brushed orbumped by foragers returning to thebivouac. By degrees the callows are ab-sorbed into the outgoing movement, untilafter a few hours they drain from thebivouac in a steady stream, moving alongin a thick column with few interruptions.During the early post-hatching periodcallows are seen in numbers outside thebivouac only when they join bivouac-change movements; then they are ableto progress in a regular manner only whenconditions in the column afford an en-vironment of uniform chemo-tactual stimu-lation promoting continuous movementin one direction. Running the trail underthese "hand-fed" conditions would seemto contribute fundamentally to an abil-ity to make progress in the daytime, de-veloping as foragers under shifting tactualconditions by responding to the chemicaltrail itself.The Eciton bivouac-change movement

sharpens our view of army-ant progress incolumns as a routine stereotyped followingof chemical lines. The view is quite de-fensible, since under any conditions theants are oriented mainly by the chemicalpath, and only secondarily by other fac-tors. Through visual stimulation, forexample, they are aroused in the morning(Schneirla, 1940) and are caused to shifttheir bivouacs away from bright light, yetappropriate tests on their foraging andother column activities give no evidencethat visual factors are involved in orienta-tion. On the other hand, the prevalentimportance of an olfactory basis is readilydemonstrated by simple tests of the"Finger-versuch" type (see Brun, 1914).Rubbing a finger across a trail causes aserious interruption, with the ants as-sembling on each side of the violated zone,unable to cross until the gap has beenpassed by some of them in a manner re-sembling the pioneer process of trail mak-

4 [No. 1253

CIRCULAR COLUMN AND ORlENTATION IN ANTS

ing. When a portion of trail on a leaf ismoved laterally so that each end is a fewinches from its former position, there is aninterruption which eventuates in activi-ties similar to initial trail making, withthe difference that the displaced sectionmay become part of the trail bridge if thepioneers chance to reach it. Such tests showthat a chemical trail is an indispensablebasis for the route. However, considerableliberties may be taken with sections oftrail on movable objects, as long as theends join the main trail somehow. In onefrequently repeated test, a trail section on a

long slender limb was quickly turned endfor end, so that the east end touched theplace formerly occupied by the west end,and vice versa. After a brief commotioncaused by the lifting itself, traffic soon re-sumed without much difficulty. Sincerelatively long sections of trail may bethus reversed without blocking the ants,the Eciton path appears to be a simplehomogeneous chemical track, not po-larized chemically as are the trails ofmany other collective foragers under givenconditions (e.g., Lasius species, Brun,1914).

CIRCULAR COLUMNS AND STEREOTYPY IN TRAIL FOLLOWINGCIRCULAR TRAILS IN PROCESSIONARY CATERPILLARS

The psychological limitations of insectsengaging in this type of trail orientationare suggested strikingly when the insectsform and continue to follow circular trails.In volume 6 of his "Souvenirs entomolo-giques," Fabre reported an interesting caseof this type observed in 1896, perhaps thefirst scientific record of the phenomenon forany animal.' Some processionary cater-pillars (Cnethocampa pityocampa) he hadhatched were seen to form a column aroundthe upper rim of a large vase in a circularpath 1.35 meters in circumference. Fabrehastened to assist the retention of thiscolumn by brushing away all threads re-maining on the sides of the vase after thelarvae had mounted, and by removing allexcess late comers which might haveswollen the column beyond the limits ofspace on the rim. Once started on the end-less trail, the caterpillars circled in thispath for seven days, moving more or lesscontinuously except for halts caused by lowtemperatures during the nights. These

1 A similar case of persistent circling in a Cnetho-campa species was reported by Dubois in 1899, with abrief discussion of the track itself. There is no evi-dence to indicate that Dubois did not discover thephenomenon independently.

and many other lepidopterous larvae ofsimilar behavior ordinarily follow oneanother in winding columns by virtue ofcontact with a thread or threads spun bythose first to pass over the ground. In theprocessions formed by the caterpillars ofthe American buck moth Hemilenca maia,Marshall (1904) found that removing theleading individual usually served to haltthe whole line in place for some time, untilthe "leader" was returned or until someother larvae chanced to proceed. This sug-gests that in this species only the firstindividual or the first few to pass along areactively involved in spinning, while theothers merely follow the trail. Contactwith the thread is not the sole factor infollowing the path, since when Marshallremoved a portion of thread betweenneighboring caterpillars in the body of theprocession those behind were not stopped.However, rubbing across the path with amoistened finger in addition to removingthe thread brought those following to ahalt, with a delay of a few minutes beforethe gap could be crossed. Evidently, in thebuck moth at least, the trail is followedchemically as well as tactually.

1944] 5


WHEELER'S LABORATORY CASE IN Eciton schmitti

That Ecitons may form circular proces-sions of a kind very similar to that ob-served by Fabre in lepidopterous larvaewas first reported by Wheeler (1910), withthe following description of the behavior.

In captivity Ecitons are remarkably restless,at least at certain times during the day. Partof a fine colony of E. schmitti which I kept someyears ago, exhibited this restlessness in a strikingand ludicrous manner. The colony was atfirst confined in a tall glass jar on a squareboard surrounded by a water moat. The antskept going up and down the inside of the jarin files for many hours. Finally I removedthe lid. The file at once advanced over therim and descended on the outer surface till itreached the circular base of the jar where itturned to the left at a right angle and proceededcompletely around the base till it met thecolumn at the turning point. To my surprise itkept right on over the same circumference which

was long enough to accommodate all the in-dividuals. They continued going round andround the circular base of the jar, following oneanother like so many sheep, without the slightestinkling that they were perpetually traversingthe same path. They behaved exactly as theydo on one of their predatory expeditions. Theykept up this gyration for forty-six hours beforethe column broke and spread over the boardto the water's edge and clustered in the mannerso characteristic of this and the allied species(E. opacithorax, sumichrasti, etc.). I have neverseen a more astonishing exhibition of the limita-tions of instinct. For nearly two whole daysthese blind creatures, so dependent on thecontact-odor sense of their antennae, keptpalpating their uniformly smooth, odoriferoustrail and the advancing bodies of the ants im-mediately preceding them, without perceivingthat they were making no progress but onlywasting their energies, till the spell was finallybroken by some more venturesome members ofthe colony (Wheeler, 1910, p. 265).

A CASE OF ECITON MILLING IN THE OPEN

Although I had observed in detail manyraids of the Central American species ofEciton and had frequently set up circular-trail formations with these ants in thelaboratory, under natural conditions noth-ing but short-lived or ambiguous instancesof circular-column behavior appeared untilSeptember 4, 1936, when a striking casewas discovered in the laboratory clearingat Barro Colorado Island, Canal Zone. Onthe preceding day a swarm of the smallblack E. praedator had worked its way fromthe north up the laboratory hill, crossedthe grounds near the cookhouse, and afterpassing near the west corner of HaskinsMemorial Library finally dwindled outtoward dusk in the general vicinity of theChapman cottage. At 7:30 A.M. on Sep-tember 4, Rosa, the laboratory cook, ex-citedly called my attention to an occur-rence which beyond doubt was a sequel tothe foregoing raid. There follows a sum-mary of my notes.At about 7:30 A.M. a singular milling group of

Eciton praedator workers is discovered on thecement walk in front of the Haskins Library(fig. 1). There are several hundred of them,mainly intermediates and large minor workers,the size classes most frequent in raids. Most ofthem are running in a circular column whichmoves counterclockwise, rotating about a small

circular cluster of ants huddled in the centerAlthough at this time the ring varies in widtharound its circumference and at intervalsmoves rather eccentrically, these measurementsare fairly consistent for the period 7:30 A.M.to 8:15 A.M.: outside diameter of the ring,10-11.5 cm.; width of the ring, 4-5 cm.; diame-ter of central cluster, 1-2 cm. The outer mar-gin of the ring is somewhat irregular, due to thefrequent straying of a few ants or a ragged file ofthem tangentially in the direction of rotation.Usually such deviates continue moving with thecolumn, though more slowly, and soon rejointhe principal throng. Less frequently some ofthem lose contact with the column, then rest inplace or move about hesitantly for a time, atlength blundering back where they are sweptinto the main activity once more. If an antchances to set off in the clockwise direction whenshe reenters the rotation, her movement is soonreversed in the course of a series of contacts andhead-on collisions in the throng.The column does not move strictly as a unit,

but at times it behaves as though two or moreindividual rings of ants were moving more orless independently around a common center.Although ants in files near the periphery aremoving faster than those near the center, thefirst impression that the inner files usually com-plete their revolution sooner than the outermostones is readily confirmed. Apparently as aconsequence of this difference in speed, there areobservable indications of friction between ad-jacent parts of the procession, with here andthere a file in the column deviating somewhatfrom its concentric course as contact with ants

6 [NO. 1253


moving beside it at a different rate causes someof the ants to stray. Jagged irregular gaps ap-pear intermittently between bands in the circu-lar column when one concentric part drawsbriefly away from a neighboring section. Suchbehavior, occurring at intervals in the generalmovement, may contribute to a general ec-centric wobbling that characterizes the move-ment. The deviations of sub-groups are sel-

rival to the main action. At 8:15 when I mustleave for forest observations (at the time con-sidered more important) the central cluster ofhuddled ants, at first piled two or three deep,is visibly dissolving into the movement. Thesize of the, cluster plainly decreases as more andmore .of the quiescent ants are stirred into ac-tion through repeated brushing of the hurryingthrong against them.

AVIF

Fig. 1. The circular column of Eciton praedator, as drawn from a photograph taken shortly be-fore 12:00 P.M. At that time the ring was approximately 14 cm. in diameter.

dom marked or lasting, yet occasionally therearises an eddy that may nearly circle upon itself,causing confusion until broken down by thepersistent force of the general rotatory motion.A sub-rotation of this kind was observed at onetime near the margin of the ring, a completedcircling within the general movement whichthreatened for a time to become a persistent

12:00 P.M.: The circular column is movingmore rapidly as a whole than in the early morn-ing, rotating as before in the counterclockwisedirection. The ring has expanded, now ap-proaching 14-15 cm. in diameter. In generalthere are wider intervals between its individualmembers, and more obvious gaps appear inter-mittently between concentric sub-groups. The

1944] 7


central cluster has disappeared altogether, andthere is now a bare central space about 1-2.5cm. wide, entered only by an occasional antwhich has momentarily turned too sharply tokeep in touch with the innermost file. Since8:15 A.M. the entire ring has shifted its positionso that the center is now about 12 cm. from itsformer place, closer to the wall of the building.At 12:10 the ring is mainly exposed to directsunlight, with only about 600 of arc on the sidetoward the library falling within the shade.At 12:15 P.M. rain sets in. Ants in the mill are

of course disturbed by the drops, particularlyon the side away from the building. This sideof the mill, although mainly protected from thedirect downpour by the wide eave of the build-ing, is still reached by a spattering of dropletswhich strikes the ants and dampens the sub-stratum. Ecitons entering the exposed areasoon hesitate, then are forced to move onward asothers push up behind. Forced to run throughthe exposed area, they tend to crowd toward thecenter of the mill with the result that the moreexposed side of the ring becomes narrower thanthe opposite side, and most of them turn rathersharply away from the rainy side when thisbecomes possible. Notwithstanding such dif-ficulties, the rotation continues.

2:00 P.M.: (Fairly heavy rain, 12:15 P.M.to 1:50 P.M.) The place occupied by the ants at12:00 is now thoroughly wet; however, thecolumn continues its counterclockwise move-ment in a new spot 15 cm. closer to the building,where the cement is fairly dry. The group didnot shift directly toward the building, but atangles in the counterclockwise direction. At2:15 P.M. the ring wobbles as it rotates, so that itsouter margin occasionally enters wet surface;in general, however, there is a small gap betweenits side and the wet cement. The entire as-semblage moves at a slower pace than at 12:00P.M. Once again there is a central cluster,which at 2:10 measures 4 cm. in diameter.

3:40 P.M.: (Sunshine after 1:50; now dark,beginning to sprinkle.) The circling proceedscounterclockwise, with the group about 8 Cm.closer to the building than at 2:00 P.M. Some 60of the ants have escaped the treadmill, only towander in a loose network of columns on theouter side (i.e., farther from the building).Their number increases as the beginning of

rain noticeably disturbs ants on the exposedside of the circle. (Heavy rain from 3:45-4:00P.M.)

8:30 P.M.: Somehow the group has dividedinto two distinct circling rings of ants. Theserings, nearly equal in size, presumably werecaused by the heavy though brief rain whichcame shortly before 4:00. Both groups rotatecounterclockwise, with their outer marginsabout 20 cm. apart; both nearer the wall thanwas the exposed margin of the single ring at3:45 P.M. In each group a few ants occupy thecentral area, where they huddle together or turnnarrowly in irregular ways. During an hour ofobservation more are shunted into the centralhuddle, drawn from ants in the inniermost file,where many have difficulty turning rapidlyenough to evade bumping by others in line.At 10:00 P.M. the two rings still rotate counter-clockwise, and in both, the central clusters havegrown in size. The rate of general movementhas become very slow.

September 5, 6:30 A.M.: On the spot ofyesterday's phenomenon little or no circlingis to be seen. The entire area is strewn withthe bodies of dead and dying Ecitons. A fewof the survivors wander about slowly, whileno more than three dozen of them form a small(ca. 7 cm. D.) and rather irregular circularcolumn in which they plod around slowly,counterclockwise. At 7:30 A.M. virtually noneof the ants are on their feet. Circling hasstopped, and various small myrmecine anddolichoderine ants of the neighborhood are busycarting away the dead.

Circumstances indicate that the Ecitonsmust have died of desiccation, after havingbeen in nearly constant circular motionover an essentially dry area for more than24 hours. They might have been able tosurvive much longer had they not shiftedtheir ring away from the wet each time rainbroke during the day. It is probable thatthese ants were marooned from the even-ing of September 3, since the raid stoppedthen and no other trace of the colony wasobserved in the vicinity after dusk on thatday.

PROBABLE INITIATION OF THE E. praedator MILL

Although this case may be regarded as aspecial instance of trailing, it is far fromsimple, as analysis soon discloses. Un-fortunately we lack direct information con-cerning the initiation of the circling;however, circumstances warrant the follow-ing supposition. It is very probable thatthe group was an offshoot of the swarm

raiding in this locality on September 3,operating at a distance from the main bodyand connected with it by only one or twoshuttle trails when a heavy rain aroseshortly before 2:00 P.M.1 The downpour

1 As measured by the station rain gauge near thelibrary, 1.12 inches fell within little more than anhour before 3: 00 P.M. on September 3. It is probablethat the ants were cut off then, since the only other

8 [No. 1253


would cause the ants to cluster, as Ecitonscommonly do when exposed to rain, andwould isolate them by flooding the area sothat the trails connecting with the swarmcould not be followed. (In the behavior ofthese ants on September 4, especially thosethat strayed away from the mill, at no timewas there any indication that shuttle trailswere in effective existence.)The circling may have begun in the

following manner. It is reasonable toassume that the isolated group or one mainbody of it was huddled in a roughly circularmass or confined to a small area during thedownpour. (Such aggregations of Ecitonscommonly form when rain breaks during araid.) Then, when the rain stopped, thetirst ants able to run freely would be thosesituated on the margin of the huddle. Mostof these individuals would tend to followaround the periphery of the group, throughtheir typical response of turning toward theside of Eciton chemical and toward the sideof continued tactual stimulation (i.e.,furnished through brushing ants on themargin of the mass). A circular trail ofEciton chemical would thus soon be formedwhich these ants and newly aroused new-comers would follow in a canalized manner.Favoring this hypothesis, there is a con-

siderable body of evidence from laboratoryobservations and tests. When light strikescaptive groups of E. praedator or otherskept in laboratory nests, after a period ofquiescence in darkness, a common re-sponse of those first set into action is to runaround the margin of the cluster, with a per-sistent circular column often resulting.Occasionally contact alone is sufficient toproduce this result, as was the case inWheeler's observations of E. schmittimentioned above. In his instance the antsestablished their treadmill around a glassjar and continued moving for 48 hours;under comparable conditions such columnsmay persist, though not operating con-tinuously, for as long as a few weeks. Ihave observed a number of cases in thelaboratory in which, for example, Ecitots

rain of the day was a light one (.05 inch in all)coming just after 7:30 P.M. Before the latter raincame, the praedator raiders (as is usual) would havemainly withdrawn into their base trails and sub-terranean avenues leading to the bivouac.

would trail persistently around the lowerborder of their cylindrical bivouac (e.g.,E. hamatum, housed in tall wire nests),form an endless column around the insidewall of a circular nest, or trek aroundobjects or clusters of ants in flat glass-covered nests. In a situation of the lasttype, an E. hamatum group consisting ofabout 40 workers and a queen would movefor hours on end in a circular path -arounda 10-cm. square of cellulose sponge in thecenter of their nesting compartment, atfrequent intervals during the 8-day lifeof the queen in captivity. This would soonbegin, for instance, whenever they werenewly exposed to light. The initial runsaround the sponge were slow and hesitant,mainly in contact with the sides and in-volving sharp turns at the corners; how-ever within two days the route had becomecircular, still leading around the sponge yetnot touching it.'The notion that the E. praedator mill be-

gan through the tendency of Eciton workersalways to turn toward the side on whichweak tactual stimulation is offered, and tomove along beside a surface which main-tains this contact, is supported by thefollowing test. When Eciton workers run-ning over a table top have settled into acourse along the outer edge of the table (acommon occurrence with E. hamatum, E.burchelli, and E. praedator), a smaller cir-cular column of them may be set up byplacing a glass jar of 10-25 cm. D. nearthe table edge so that some of the antsbrush it with antennae or legs as they

1 The workers in this case were respondin to thestimulative effect of the queen as well as to the trail,as was shown by the fact that when she was presenta compact group of them always hurried alongclosely behind and beside her (many clutching ather legs and body, or actually riding upon her), butalways with a clear space ahead of her. When shewas removed they continued to follow the trail,but now in an even column and at a slower pace. Itmay be added that normally the Eciton queen appearsoutside the bivouac only in the evening or at night,to participate in the movement of the colony to anew site. Then she follows the trail (used during theday as a raiding route) readily under her own power,accompanied by a horde of workers which greatlyswells the normal width of the bivouac-changecolumn. Far from being dragged along by workersto the new site, as many writers have assumed is thecase, she furnishes her own power and usually pullsor carries workers with her. This applies to thespecies of Eciton (sensu 8tricto) which have been ob-served (E. hamatum, E. lucanoides, E. burchelli).In the early part of the rainy season, males have beenseen following the bivouac-change route in a similarmanner. It is very possible that Eciton mating oc-curs under these conditions.

1944] 9

10AMERICAN MUSEUM NOVITATES

pass. Ants that touch the jar lightly inthis manner tend to leave the table-edgetrail and follow the jar around its circum-ference, and a circular column resultswhich in the end may absorb nearly allthe strays on the table. The contactmust be light, or the response is to turnaway from the stimulated side, resulting inan eddy of the table-edge column awayfrom tlre jar, but with no other change inbehavior. If the jar is pushed too far intothe initial column, so that the ants actuallycollide with it in passing rather than merelybrushing its surface, they turn away andthus many get off on the vertical edge of thetable, perhaps ending by the entire grouptrailing away on the floor.

These results indicate the great impor-tance of positive response to weak contactin the initiation of circular columns, sup-porting our hypothesis concerning the"spontaneous" formation of such a columnin the open. In that instance, as suggestedabove, a turning toward Eciton chemicalmust have been involved as well. Such acombination of weak contact and Ecitonchemical facilitates the formation ofcolumnns of this type; we have observedthis repeatedly in the laboratory. Yet thefact that light contact alone will sufficeto initiate proceedings suggests that it isnot so much a matter of the nature of thestimulus as it is the intensity of stimulationthat arouses the approach-turning response.

ANALYSIS OF THE PERSISTENT CIRCULAR COLUMN

After the column has been active for atime, contact with the surface no longer isessential, as is indicated by the case citedin which an E. hamatum group trailed in acircle around a square sponge. In a morestriking demonstration, after Ecitons haveformed a column around a glass jar on aglass substratum, removal of the jar leavesa regular circular procession which usuallyis able to maintain its formation for a con-siderable time. The chemical trail which isformed as the ants follow the surface be-comes sufficient alone to canalize thecolumn. Thus in a number of instances,Ecitons ceased trekking around th6 out-side edge in running around the top sur-face of a square wooden block, graduallyshifting to a course that cut the cornersmore and more, until finally they followeda circular course which neared the edge onlyat the center of each side.

It seems clear that other factors mustcomplicate such behavior, beyond chemicaland tactual stimulation as described. Aninteresting instance of freedom from theoriginal contact which led to new complica-tions was furnished by an E. hamatumgroup first observed running around asimple partition in their shallow glass-covered nest. After several hours it wasnoticed that they were actually describinga parallelogram around the partition, the

acute angles extending beyond the ends ofthe partition in the direction of turning(i.e., counterclockwise). (See fig. 2.)What happened was that after slavishlyfollowing the wall tactually for an hour ortwo, running speed gradually increased,and some of the ants began to overrun inrounding the corner. This brought theminto contact with the side wall of the com-partment, which they followed for a shortdistance before they turned back towardthe partition (evidently responding tochemical stimulation). As their runningspeed increased, the ants steadily increasedthe distance they ran along the side wallbefore turning back toward the partition,and the path of their return to the partition(at first noticeably concave inward) cameto be a straight line which brought themback to the partition at a point graduallycloser to its opposite end. Finally the antsran a course parallel to the compartmentwall but not at all in contact with it, afterrounding the end of the partition; turningfrom this line in an arcing course thatstraightened so that they neatly roundedthe opposite end of the partition. At lengththey traced out the parallelogram withblunted corners shown in figure 2, in whichthey were near the partition only at itsends. In this rather complicated example,original contact with the partition and

[No. 125310

1944] CIRCULAR COLUMN AND ORIENTATION IN ANTS

Fig. 2. Parallelogram course established by a group of Eciton hamatum workers around a parti-tion in a shallow observation nest. The solid line represents the path eventually followed by theants; the arrows indicate the direction of travel at a given time. (See text.)

newly encountered contact with the sidewalls were dispensed with as a chemicaltrail formed; however, as increasing satura-tion of the substratum facilitated faster run-ning, this augmented the influence of a

centrifugal-force effect generated in round-ing the ends of the partition, accounting fortheir modifying the chemical trail to formthe short sides of the parallelogram. The

turn back toward the partition throughwhat became the long side of the parallelo-gram may be regarded as a result of turn-ing toward chemical stimulation when themomentum effect had weakened suffi-ciently.The centrifugal-force factor, disclosed

in the running of geometribal figures ac-cording to laboratory conditions, is to be

11


reckoned with when one studies trailing be-havior under natural conditions. For ex-ample, its influence is disclosed whensharp turns around objects gradually shiftoutward as the trail develops, until the antsfinally run at a distance from the previ-ously "hugged" surface in rounding thecorner. Or at other times, when a columnat first turns off to one side promptly afterhaving reached the end of a stick alongwhich the trail leads, the course shifts intime so that finally the turn is made in awide arc at a distance from the end of thestick. This type of modification commonlyoccurs along the course of trails used for aconsiderable time during the day. Depend-ing upon: the slope of the incline whichis mounted or descended, the momentumwhich may develop through passage alonga fairly straight section of trail according tothe smoothness of the terrain, and the run-ning speed which may be in force in ap-proaching and rounding a corner, numerousfeatures in the initial and the later form ofthe route may be understood. In studyingtrail formation, one can never afford toneglect the simple mechanical principlesunderlying animal locomotion.'Now we are in a position to attempt an

explanation of how the E. praedator mill waskept going for more than 24 hours in theopen. It is clear from the facts that thisbehavior involved more than the merefollowing of a circular chemical track;the shifting of the entire mill through adistance of about 30 cm. within less than 12hours dismisses that notion. However, atany given time the ants were responding toa ring of chemical, as is shown by thepartial disruption of travel introduced byoccasional "Finger-versuch" tests (e.g.,rubbing clean glass cloth over a portion ofring area vacant at the moment). In thisphenomenon the chemical factor seems tohave been merged with other influences in arather involved manner.The persistence of this circular column

1 For example, one such effect, the role of gravita-tion, has been studied by Barnes (1930) in relationto the lateral oscillatory excursions of Lasius inrunning. The influence of centrifugal force as con-sidered here commonly leads Ecitons into behaviorsituations resembling those described as "centrifugalswing" in the maze behavior of Formica species(Schneirla. 1929, 1941).

may be understood in terms of the interac-tion of three component factors commonlyinfluential in normal trail movements.These are: (1) a pressure, tactually exertedaccording to the principal direction ofmovement in the group, which minimizesvariability of movement and tends tocoerce "forward" progress in the group;(2) a drainage factor, the composite effect ofchemical stimulation from the trail andtactual stimulation (as well as chemical)from travelers ahead and beside the givenant; (3) a centrifugal-force factor, theinertia of movement which tends to carrythe ant away from the circle on a tangentwith the circumference. The combinedoperation of these three factors holds thecircular column together and accounts forthe principal variations in its behavior.The term "pressure" for our first factor

is used here in the sense it was used whendealing with the phenomenon of organizedswarm movement in Ecitons such as E.burchelli and E. praedator. "Its basis is theproneness of the Eciton worker to turnaway from the side of strong or repeatedtactual stimulation and to reverse herdirection of progress if such tactual stimu-lation is encountered in head-on fashion"(Schneirla, 1940, p. 425). Thus throughforcible bodily contact, an Eciton that hap-pens to run against the preponderant direc-tion of movement in a group is soon broughtinto line. If she happens to lag when in ahurrying column, forcible tactual stimula-tion from the rear spurs her onward. Thisfactor must have played a major part inaccounting for the fact that one direction oflocomotion (i.e., counterclockwise) pre-vailed in our praedator mill. From labora-tory observations we may say that a vari-able combination of circumstances (i.e., a"chance" situation) caused the majority tohit upon this direction of movement at anearly point, and that deviates were gradu-ally forced into the same direction through''pressure" until there was no longer anypersistent clockwise movement in thegroup. A somewhat similar phenomenon isinvolved in a complex manner when anomadic movement begins to developtoward evening in a raiding system of the

12 [No. 1253


Y

N1- N%/ N

/ \

/,

le//

/

N

'NN. __

Fig. 3. Sketch representing the shift of the E. praedator ring as a result of wetting by rain in thearea indicated by wavy lines. 1-7 Indicate the initial courses of concentrically traveling ants; la-7atheir respective courses upon entering the area of disturbance; lb-7b their respective courses upon

leaving this area. 7a Indicates how individual ants are reoriented when they turn back from thewetted surface. X, the vector of pressure upon the upper portion of the ring; Y, shortest distance towall under the eave; Z, direction in which the circle shifted in response to the rain. (See text.)

19441 13

/


swarm raider E. burchelli (Schneirla, inpress).Through the effect of pressure, in particu-

lar, one prominent change in the action ofthe praedator ring may be understood.When rain spattered upon one side of theassemblage, causing confusion and ac-counting for a noticeable faltering of antsentering that area, the circular column waskept in being especially through the in-fluence of a forceful mass contact from theless exposed and more smoothly operatingpart of the mill. Ants that hesitated andturned when struck by rain were therebyforced to turn "forward" again and some-how rush through the damp area. Evi-dences of another special pressure effectwere observed soon after the rain broke,an effect which seemed to contribute to theshifting of the entire circular mill towardthe wall (i.e., farther under the wide eave ofthe building) during the downpour. Mostof the ants running through the exposedarea (i.e., the side away from the wall) notonly speeded up noticeably, but also tendedto turn short away from the side of disturb-ance instead of remaining in their respectiveconcentric lines. Presently there wereindications that the strong pressure effectthereby generated, concentrated upon theinner part of the circle when ants enteredthe drier area in the clockwise direction,was literally pushing the less exposed partof the mill toward the wall. By causingnumbers of ants to run more widely out-ward while passing through the marginalsection on that side, this unilateral pres-sure produced an eccentric torque or"twist" in the circling, toward the wall andcounterclockwise (fig. 3). As a result ofthis effect and its various secondary conse-quences through the group, the entire millgradually shifted its position at angles(i.e., in the counterclockwise direction, andnot straight toward the wall) farther underthe eave of the building. As new portionsof surface closer to the building becamechemically saturated by ants forced outcentrifugally on that side, the entire millcould move into them, but most important,it was the dynamic interrelationship of theEciton units in the mass that permittedthe ring to retain its essential pattern

throughout the crisis. Then, after the rain,it regained its earlier symmetry.'Although pressure is by no means the

only agency accounting for persistent"forward" motion and the minimizing ofdeviations, the "push from behind" playsan important part in most of the groupactivities of these ants and should not beneglected here. Another function of pres-sure in the mill, a lateral component operat-ing centripetally, will be considered pres-ently.Our second factor, "drainage," also holds

a basic place in all Eciton mass movements,especially in the mass organization of swarmraiding (Schneirla, 1940, p. 425). This in-volves the propensity of the Eciton workerto turn toward the side of weak tactual andweak chemical stimulation, or toward thesein combination. This positive response toweak tactual stimulation has been illus-trated in the case of a circular columnstarting around the edge of a glass jar.General observation supports the view thatgentle contact with another Eciton aheadelicits turning toward the side from whichthe stimulation comes; from the samesource and from special tests comes sup-port for the tendency of Ecitons to turntoward Eciton chemical, favoring the sideof greater concentration.For example, a large strip of bristol board

was prepared in an Eciton-chemical gradient byregulating the passage of workers across it tohave a maximum of crossings at one end and aminimum at the other, with graded frequenciesbetween. Then in succession 40 E. hamatumworkers were released at the edge, 20 at the"maximal-saturation" end and 20 at the mini-mal-saturation end, and were permitted tocross freely. More than 15 of the formercrossed at the same end of the card; whereasthe latter group crossed at points extendingnearly the full length of the card. Thus therepresentative response of ants entering theweakly saturated zone was to turn into partsof the gradient in which Eciton chemical- wasmore concentrated.

1 Although the movement of the ring out of sun-light during late morning was not observed as it oc-curred, it seems very probable that this shift dependedupon events somewhat similar to those described.The distorting effect-in that case would depend uponthe irritating action of bright light upon the Ecitons,causing them to speed up when entering the exposedarea and turn away from the dazzle as promptly ascircumstances permitted. Such responses are com-monly observable in the everyday behavior of E.praedator swarms which are partially exposed to bril-liant illumination (e.g., in sun-fleck areas).

44 [No. 1253


The evident influence of the third factor,centrifugal force, is disclosed in figure 1, inwhich one or more lines of ants may be seentaking a path tangential to the rotating ringin the direction of turning. In the outer.portion of the ring, where the ants run

centrifugal force increasingly effective asrunning speed increases, Eciton pathsaround square or irregularly shaped objectsapproximnate a circle more and more closely.Through its effect, a hamatum path arounda linear partition (influenced by initial

.e ~~~~N,

//

//

//

I.,

4-1-9e

IfI I

I;

ol

/6& /11 /

-, %\

< ~~~/}_- _,9

N

I

II

/

raX /,b

I '"-\e, ~

I". 1

Fig. 4. Sketch to represent factors influencing the movement of ants in the E. praedator ring.1, An ant moving on the margin: a, centripetal vector; b, centrifugal vector; r, the resultant eir-cular course. 2, An ant moving in one of the inner lines: a, centripetal vector; jS, centrifugal vec-tor, oy, pressure from outside and rear; p, the resultant circular course. (See text.)

most rapidly, the influence of this factor ismost directly apparent in behavior, in thatthere is a persistent tendency for themarginal ants to "fray out" from the ringin the forward direction. We have men-tioned laboratory cases in which, with

contact with end walls) became modifiedinto a parallelogram (fig. 2). In such casesthe interaction of our second and thirdfactors is recognizable: a centrifugal-forceeffect leading the ant to run onward afterturning a corner, and a response to Eciton

1944] .15


chemical when the centrifugal effectweakens.To take stock, ants in the mill are kept in

motion in the forward direction by a pres-sure from others just behind, an effectwhich is mutually exerted throughout thegroup. Upon this basis, the dynamic pat-tern mainly responsible for survival of thecircular movement itself may be expressedmost simply in terms of a combination ofdrainage and centrifugal-force effects. Toillustrate this interactive set of influences,let us consider the case of an Eciton movingat the margin of the ring (fig. 4). Oc-casional contact with ants moving along onher left (i.e., inside, in the counterclock-wise mill) together with a turning towardthe side of distinctive Eciton chemical,would account for a centripetal vector inher movement (fig. 4, 1 a). At the sametime, depending upon running speed, hermomentum would account for a centrifugalforce which would tend to impel her directlyahead on a tangent with the circle (fig. 4,lb). (As we have indicated, the effect ofcentrifugal force is directly evidenced attimes in actual tangential departures ofmarginal ants from the ring, while the effectof the "drainage" factor is shown in theappearance of hesitancy and turning backtoward the circle after short tangential ex-cursions of this kind.) The circular courseof marginal ants may be considered the re-sultant of these two vectors (fig. 4, Ir).Although the case seems more compli-

cated for travelers on lines within the outerborder of the mill, we may assume that thesame drainage and centrifugal-force patterngoverns the circular course of all. However,for ants moving in interior files, there areindications that pressure effects are super-added. In their case, the centrifugal-forceinfluence (fig. 4, 2,B) is counteracted notonly by a contact-chemical drainage effect(fig. 4, 2a) from ants turning away on theleft (i.e., "inside," in the present case),but also by pressure effects through fre-quent and at times forcible contact withants running on the right side (i.e., "out-side"). The circular course (2p) may beconsidered the resultant of these threevectors. One feature we have mentioned,the indications of "friction" between fast

moving outer lines and adjacent slowermoving inside lines, seems to illustrate theinvolvement of a bilateral tactual patternfor ants within the body of the mill.Assuredly, centripetal pressure must not beexcluded here as a regulator of movement.We have noted that the circular mill ex-panded between 8:15 A.M. and noon. Itwill be recalled that a rise in temperatureduring this period would increase the run-ning speed of all ants in the group, thusheightening the centrifugal-force effecttending to widen the circling of the re-spective concentric rings of ants.' But inaddition, in accounting for the expansionof the mill, it seems necessary to considerthe possibility that "inside" ants, by exert-ing an increased centrifugal pressure,facilitated the wider turning of outsidelines; and that outside lines, in arcing morebroadly, thereby reduced centrifugal pres-sure and permitted freer action to thosewithin.Now we may consider the Ecitons caught

in the innermost line, near the hub of thewheel. We have remarked the fact that inearly morning, near the end of the rain,and toward evening a central cluster wasformed. At other times the central areawas essentially bare, and when empty itwas also larger. These were the times whenhigher temperatures accounted for a fasterrotation of the mill as a whole, whichchecks with our suggestion that the fastermotion of individuals increased thecentrifugal-force effect and thereby spreadthe ring. However, when the ring con-tracted, individuals in the innermost linewere forced by crowding (i.e., lateral pres-sure) of ants in the next outer ring to turnvery sharply. This meant that some of thecentral ants were buffeted and crowded soforcibly that they were caused to circle inplace. Turning variously at differentangles these ants interfered with oneanother, an organized circling of the inner-most members thereby became impossible,and they presently were thrown into acluster. Very probably the relative slug-

1 Shapley (1920) and others (Miller and Gans,1925; Pratt, 1925) have shown for various speciesthat the speed of ants traveling an established pathincreases with the temperature up to a given maxi-mum.

16 [No. 1253


gishness of the ants at such times (i.e., inlower temperature) contributed to theclustering tendency. Similarly, in normalforaging, when the coming of dusk slowsthe Ecitons, the nucleus of clusters at trailjunctions is formed by disoriented antscaught and knocked about in the hectic

traffic at such points. We may say thatEcitons in the center of the narrowedpraedator mill were passively tossed aboutby a movement in which they could nolonger participate as synchronized units,and thus in a sense assumed the role offlotsam in the center of a whirlpool.

MILLING IN RELATION TO TRAILING UNDER NATURAL CONDITIONS

Instances of this kind in animal behaviorare arresting not simply because of theirbizarre character, but also because theymay be utilized to review the factors ofnormal behavior in relief, as it were. Inthis example we have seen the operation,under special conditions, of the principalfactors in everyday Eciton mass inove-

ment, i.e., pressure and drainage, togetherwith the centrifugal-force effect. Studies ofother Eciton species have shown how thesefactors are involved in the formation anduse of ordinary -raiding trails (Schneirla,1938). Furthermore, the same conceptsmake understandable the integrative proc-

esses involved in the surprisingly directadvance of the relatively enormous E.burchelli swarm (Schneirla, 1940) and thesmaller swarm of E. praedator. Despitevery complex and at first sight utterlyrandom eddying and cross-current move-

ments among sub-groups in the swarm, byvirtue of describable relationships of thepressure and drainage factors the wholebody of ants is capable of a fairly straight-forward advance through vegetation andover broken forest terrain. In the wakeof every swarm there is a long fan-shapednetwork of columns (consolidation fan),which narrows posteriorly into a singlecolumn communicating with the bivouac.It is a relatively constant pressure, operat-ing through this system and applied more or

less evenly across the rear of the swarm byants arriving from the bivouac, whichenables the raiding mass to hold its laterallyextended formation and at the same timecontinue its advance.Why ants behind the swarm do not con-

tinue to roam throughout the wide lanesaturated chemically by the swarm, but in-stead confine themselves to a progressively

narrower channel at distances farther inthe rear, is explicable in terms of gradualchanges in chemo-tactual response withfurther use of the ground. A turning ofEcitons toward others, thus as a rule keep-ing to circuitous columns and not spreadingout widely, is noticeable in the rear of theswarm where forces are reduced in num-bers. The center of the area is favored pre-sumably on a chemical basis. Because ofthe way the swarm moves ahead, spreadingout alternately on the flanks but with afairly consistent crowding in the center,there is evidently a reduction in chemicalsaturation from the center toward bothsides in the area pillaged by this body.Ants coming up in the rear evidently re-spond to this gradient by turning morereadily toward the center of the wide lanethan toward its borders. Their move-ments on the ground by favoring the centralpart increase the steepness of the gradient,thus provoking still more precise turnscentrally by ants advancing behind themin the fan. At the rear of the fan, wherethe gradient must be steepest, all lateraltrails disappear from use, as a rule, andonly a narrow central route remains. Inthe behavior of Ecitons following thisprincipal trail between bivouac and raid-ing front, there are many indications thatthe route is responded to as the region ofgreatest intensity in a bilaterally extendedchemical gradient. A crowding away fromthe edges toward the center of the path isnoticeable in moments of traffic congestion,especially when the route is new, and thefastest moving ants generally usurp themidline at the expense of burdened ones orslower moving individuals such as theclumsy major workers. Displaced laterallyfrom the trail a few centimeters or more, in

1944]- 17


the ensuing rambling of the Eciton there isa noticeable drift toward the main path,even if it is empty at the time. Displacedon the original ground, ants return to thecentral route far more quickly than whentransferred to new ground with the trail(e.g., a section of it on a narrow leaf)located at a comparable distance.Now although Ecitons are undoubtedly

among the simplest of ants, psychologicallyconsidered, highly stereotyped in theiractivities, yet to the writer's knowledge atrue circular mill has been observed in onlythis one instance under natural conditions.The burchelli or praedator swarm as it ad-vances exhibits countless instances of vor-texes, whirls, and eddying movements ofshort columns in the mass. Also, pushingout from the forward and lateral margins ofthe swarm are "pseudopodic columns"which frequently take eccentric courses,even looping back toward their originpoints at times. When the consolidationfan takes form on ground first swept by theswarm, in its complex anastomosis there arealways numerous remnants of circlingcourses formerly apparent in the swarm(Schneirla, 1940, fig. 1), which stand outfor a time as loops and circular trails in thenetwork. At times, on these circuitousportions of the system, one observes thatgiven ants or trains of ants may travel oneor more rounds in a loop before there is achange. However, in these cases, as whencomparable circuits happen to form in thetree-like trail systems of E. hamatum, thecomplexion of events soon modifies thecircular path, so that true milling does notoccur.

Actually, although all the chief factorsunderlying our instance of milling arereadily recognized in trail following undernatural conditions, circumstances operat-ing in the forest environment militatestrongly against any possibility of Ecitonsbeing caught in a persistent local circlingmovement. Either the pressure of antsentering over connecting trails changesmatters, or, if few newcomers arrive, theshuttle trails provide escape routes. The

raiding system of any Eciton species is acomplex network of chemical trails con-necting with the bivouac over well-traveledtrunk routes, providing slight opportunityfor any section of traffic to operate inde-pendently for long without being modifiedthrough its connections with adjoining sec-tions. For this reason, no part of a raidingsystem can remain very long in a given pat-tern of action. In the case of our praedatormill the exception occurred, but thereanother unusual condition happened to befulfilled: the ants were cut off from theircolony when on or near very homogeneousterrain, the cement walk. Here the sub-stratum was tactually neutral, presentingno irregularities which could intrude uponthe symmetrical operation of pressure,drainage, and centrifugal-force influencesaround the ring. Afield, few if any situa-tions ever approach very close to neutralityin this sense. To be accurate, the cementsurface on which our praedator groupchanced to rotate must, be considered aquasi-artificial situation.We find essentially the same idea in

Wheeler's (1928b) interesting discussion ofpolymorphic specialization in ants, appliedto the worker major of the harvesterPheidole."Some authors regard the soldiers, the highest

high-brows of our ant series, as monstrous, orpathological forms on account of the excessivedevelopment of their crania. Certain factsmight seem to lend support to such an opinion.If the soldier of Ph. instabilis be placed on itshead on a perfectly smooth, hard, horizontalsurface, the insect may be quite unable to rightitself and may even die standing on its head.But this is a typical laboratory experiment.In its natural environment the soldier neverencounters such surfaces. Closer study showsthat all these supposedly monstrous forms arereally exquisitely specialized and adapted for thefunctions they have to perform in the life oftheir respective colonies. The soldiers of theharvesting Pheidoles and Pheidologetons areneeded not only as seed-crushers, but those ofthe latter genus have another very differentfunction. Several observers have seen groupsof the minute Pheidologeton workers sittingquietly on the huge heads of the soldiers andriding to and from the nest. The soldiers ofthe insect-eating Pheidoles dismember the toughprey before or after it has been carried into thenest... ." (p. 19).

18 [No.;1253


RELATIONS TO MILLING IN OTHER ANIMALS

PARR'S ANALYSIS OF MILLING IN SCHOOLING FISHES

After the foregoing analysis of the Ecitonmill had been sketched, in a search of theliterature for cases elsewhere in the animalseries my attention was drawn to a similarphenomenon in schooling fishes, describedand analyzed by A. E. Parr (1927). Acomparison of the Eciton mill in figure 1

from the known properties of the animal.Of course the Parr analysis takes priorityas a7 theory of milling behavior in infra-mammalian animals; however, mention ofit has been delayed until this point in orderto emphasize the fact that a second in-vestigator, working independently with,~~~~~~~~~~~~~~~~~~~.

Fi.5.Sethofteicla is il esrbd yPar (e tx.

with the sketch of one level of Parr's fishmill in figure 5 indicates a marked re-semblance in these patterns of invertebrateand vertebrate behavior. Even morestriking are the similar outcomes of analy-sis by the two investigators. Each of usevaluates his phenomenon as essentially onein the mechanics of animal locomotionin relation to available sensory patterns;each develops an explanation inductively

very different material, arrived at a theorywhich, notwithstanding the different set-ting, bears a surprising similarity to theParr "bio-mechanics" explanation.The theory offered by Parr to account

for the formation and maintenance of amilling group of fish may be represented inr6sum6 as follows. When a school of fish(e.g., mackerel) meets a deflecting surfaceand begins a sharp turn, individuals first

1944] 19


to head in the new direction are presentedwith a potent visual stimulus from thosedirectly opposite and traveling in theoriginal direction. This stimulation domi-nates over that from adjacent fish whichhave participated in the turn and, areswimming on the outside in the new direc-tion. Since the stimulative influence ofcompanions swimming in the same direc-tion is already adjusted to, the new stimu-lus is "much stronger" because it repre-sents a more rapidly shifting pattern.Hence fish on the inside of the wheel con-tinue to turn toward the inside. Thenucleus of a mill is thereby formed throughthis mutual visual stimulation whichkeeps fish in the interior file turning towardoppositely swimming fish (fig. 5, 1 and 2).Soon individuals in files next the newcenter become influenced by the centralindividuals turning away from them in-teriorly, and they respond by turningtoward that side, thus beginning to wheel

on the same principles as those causing thefirst circular movement of the central file.In each case, a similar turning of the ad-jacent outward companions in the school(fig. 5, 3, 4, and 5) prevents their stimuluseffect on the given individual from oppos-ing that of the adjacent interior file. Themill is maintained on the same basis:the circular movement of the central fileacquires the nature of a continuous auto-matic process, and the tendency to turn asoriginally transmitted outward is main-tained as each fish continues to respond tothe inward turning of its adjacent innercompanion. Once the mill is formed, theoutermost file exerts a certain condensinginfluence upon the moving group, since theperipheral members are influenced solelyby a one-sided attraction toward com-panions turning away on the inside, andthus act as a wall restraining any indi-viduals happening to stray outward frominterior files.

A COMPARISON OF MILLING IN FISHES AND IN ARMY ANTS

These phenomena of circular milling inschooling fishes and in army ants are

strikingly similar, as a comparison of therespective analyses above suggests. Inparticular, in both cases a tendency to turntoward a dominant centripetally operatingstimulation seems to be paramount. How-ever, there are a number of differencesworthy of consideration. One is in thedominant sensory mechanism of the milling,which in the schooling fishes thus farstudied (Parr, 1927; Breder and Nigrelli,1935; Shlaifer, 1942) has been identified as

vision. In connection with his study ofmobile aggregations in the herring Jenkin-sia, Breder (1929) has pointed out thatfishes schooling on the basis of vision ap-proach only as close to one another as thedistance at which clear optic resolutionoccurs. At shorter distances a negativeresponse is given. In contrast, the army

ants maintain their formation in depend-ence upon a pattern of diverse tactuo-chemical stimulative factors. This un-

doubtedly admits certain differences in thedetailed mechanism of circling, such as the

influence of a tactual "pressure" from be-hind and from the next outer file in theEciton mill, not demonstrable in the fishes.Although from Parr's analysis this factor ofproximal stimulation would seem negligibleor absent in the visually based fish mill, it isbarely possible that it may acquire at leasta limited importance in the exceptionallydense mills formed by the more tactualcatfishes (Ameirus). When in schools, asParr (1927) states, young catfish arespaced at intervals governed by the span oftheir tactile barbels.

This difference in the dominant sensorymodality may be involved in accounting foran apparent difference in the initiation ofmills in Ecitons and in fishes. As far as ourcases go, the circling in visually dominatedfish appears to begin centrally and growcentrifugally, whereas through variousrelevant considerations we were led to re-gard the praedator mill (not observed in itsorigin) as having begun peripherally anddeveloped centripetally. Of course, it isquite possible that more extensive studiesmay show that the mode of initiation, cen-

20 [No. 1253


tral or peripheral, is not an invariable mat-ter for either animal. Of far greater im-portance than this is the above-mentionedsimilarity in the maintenance of the twomills. Both may be understood as cases ofpredominantly unilateral stimulation of in-volved individuals, with Ecitons turning instereotyped-manner to the side of light con-tact and more effective Eciton chemical,the mackerel toward the side of visualstimulation.

Other differences between the milling ofthese two animals deserve mention here.An obvious difference is that although fishschools and milling when it occurs com-monly operate in three dimensions, theEciton mill is necessarily a two-dimensionalaffair which is more rigidly localized in oneplane of space. This suggests a furtherprevalent difference which very probably isimportant for the frequency with whichmilling occurs under natural conditions,that is, the relative heterogeneity of the

two-dimensional working surface of theants, in sharp contrast with the greathomogeneity of the aquatic medium ofpelagic fishes. A homogeneous environ-ment tends to accentuate and to facilitatestereotypy in behavior, by permitting thepeculiarities of the animal to establishrelatively rigid patterns which dominate itsbehavior in given monotonous ways overconsiderable periods of time. In contrast,notwithstanding the fact that the propertiesunderlying Eciton behavior are in them-selves highly conducive to stereotypy(Schneirla, 1940, pp. 447 ff.), the varyingsituations of a tropical-forest environmentmilitate strongly against the persistence ofmonotonous patterns such as the circular-trail formation. It is otherwise with fishessuch as the herring, which not only possessthe organic prerequisites for stereotyped be-havior, but also happen to live in an en-vironment which admits the predominanceof these properties in given patterns forconsiderable periods of time.

CONTRASTING PATTERNS IN ANT ORIENTATION IN RELATION TO MILLING

Actually the mechanism of Ecitonorientation, although complex in its ownterms, represents nearly the extreme ofsimplicity in ant way-finding. It bears aclose relationship to the stereotyped modesof orientation which characterize the in-vertebrates in particular, reaction typeswhich, although highly involved, havebeen studied very successfully through theexperimental examination of direct rela-tionships between the energy pattern ofstimulation and the dynamics of locomo-tion.' The Eciton system of orientationreduces essentially to a generalized chemo-tactually controlled pattern, apparentlywithout much learning involved beyondthe initial approach response to Ecitonchemical. Contrasted with the cases ofmany other collective foragers, especiallyamong the social insects, this army antsystem appears highly rudimentary.

Formicine orientation on its higher levelsexemplifies the condition which Maier and

1 For a survey of evidence on these basic forms oforientation, see in particular the valuable work ofFraenkel and Gunn (1940).

Schneirla (1935, chap. 7) have termed "pro-visional orientation," in that the animal'sbehavior in a given field of stimulation isdominated in appropriate and plastic waysby circumstances which initiated the trip,and by a capacity to modify the routethrough previous experience in the situa-tion. For example, the common garden antLasius niger americanus runs its trails ac-cording to the prevailing circumstances ofchemical, visual, tactual, and muscle-sense (kinesthetic) stimulation, improvingthrough learning in successive runs overthe same general route. Since Lubbock'spioneer studies (1882), it has been knownthat Lasius niger follows its trail not on achemical basis alone, but also according tothe direction of light rays. Although thereare grounds for believing that the Lasiusforager may learn to move according tothese sensory patterns, there remainlargely unanswered a number of questionsconcerning the ant's ability to adapt hermovements to changing outer conditions.One such question concerns how this ant

1944] 21

22AMERICAN MUSEUM NOVITATES

(together with a number of other species ofsimilar foraging activities) is able to use agiven route through a considerable periodwith light direction one of her orientingcues, notwithstanding a gradual change inthe angle of the sun's rays through theperiod. Part of the answer seems to liein their ability to utilize particular serialtactual cues such as the edge of a boulderor stick in a given section of terrain. Fur-thermore, in certain localities they maylearn to depend upon slope of the groundas one cue, and Brun (1914) showed thatLasius adapts differently to such propertiesaccording to the nature of its relativelybroad chemical trail. Also, if the ant re-turns repeatedly from feeding on nectar,there arises an end-for-end (i.e., polarized)difference in the chemical trail to which shebecomes able to respond (for example, shereadily differentiates the two directionsafter being displaced from the trail). If,however, larvae (Lasius larvae, experi-mentally placed at a distance) are beingcarried back to the nest, she becomes moredependent upon the non-chemical cues(e.g., direction of light) and becomesseriously disoriented if light, tactual cues,or slope of the path is changed. In con-trast, Ecitons, as we have pointed out,follow their foraging trails mainly on thebasis of a generalized ability to movetoward a common chemical product. Intheir case an ability to learn a particularroute according to direction of light orother special cues existing on the givenpathway seems to be absent, as far as thewriter's tests have shown.Thus the Ecitons represent the primitive

type of collective foragers among ants,which follow their trails in a slavish, stereo-typed manner without the involvementof a special learning of the route, whileLasius represents the more highly special-ized type of collective forager which iscapable of learning a route in dependenceupon prevailing sensory conditions. Underappropriate conditions Ecitons follow acircular trail persistently, but we may ven-ture to say that no particular success wouldmeet an attempt to elicit such behaviorwhen Lasius individuals are the subjects.The contrast is even greater if we com-

pare simple collective foragers such asEciton with true individual foragers suchas ants of Formica or Camponotus species.These ants are capable of learning indi-vidual routes to foraging areas in differentdirections from the nest, and can getthrough despite terrain changes (e.g., dig-ging up the ground, laying down new sur-faces) which would completely blockEciton or most collective-foraging ants.Field tests show that a Formica individualmay set off from a food place toward thenest (50 yards or more away), first travel-ing mainly in dependence upon a clump oftrees (visual stimulation of the ocelli),then pass through a low stretch of terrainmainly in dependence upon direction oflight (stimulation of the compound eyes),then shift mainly to chemical cues upon abroad "ant road" a dozen or more yardsfrom the nest (getting through on a visualbasis if the ground has been disturbed),and finally reach the entrance to the nest byutilizing tactual cues, slope of ground, anda chemical gradient in complex combina-tion (see Brun, 1914; Schneirla, 1929).Iin the laboratory, Formica incerta sub-jects are quite able to master a fairly com-plicated maze even when diffuse illumina-tion reduces visual cues to a decided mini-mum and chemical cues are rendered highlyunstable through frequent shifting andchanging of the alley linings (Schneirla,1943). Attempts to secure a mill phenom-enon comparable to the Eciton praedatorcircular trail could never succeed with antsof these species.Thus among ant species one finds a rela-

tively wide range of foraging patterns,from a stereotyped following of masschemical trails to a highly specializedlearning of individual routes largely inde-pendently of any chemical tracking. Ofcourse, a detailed study of phenomena suchas the Eciton circular trail shows that eventhe "elementary" types of trail-followingbehavior are far from a simple matter ofrunning along a line of chemical. However,in Eciton species the milling phenomenondiscloses susceptibilities for stereotypedbehavior which are truly representativeof characteristic psychological limitationsappearing in their normal foraging activi-

22 [No. 1253

194]CIRCULAR COLUMN AND ORIENTATION IN ANTS

ties. We have ventured to predict theimpossibility of obtaining a true case ofmilling behavior in more advanced collec-tive foragers and in individual foraging

species of ants, mainly because of theircapacity for variable behavior in wayfinding which largely excludes such activi-ties.

CONCLUDING REMARKS

Studies of mass behavior phenomenasuch as milling make their chief contribu-tion in the light cast upon the generalproblem of levels of integration in socialbehavior. It is not the pattern in and foritself, but its relation to other properties ofindividual and group behavior in the sameanimal, which merits special investigation.The milling pattern arises when animals

moving within the same area yield in com-mon to the influence of routine stimulus-response mechanisms arising through grouplocomotion. These characteristics markit as a very rudimentary form of grouporganization. In themselves, circularcolumns and other types of milling aresimply monotonous group activities oflittle immediate adaptive significance savefor the expenditure of energy, standingvery low in the general scale of group-behavior patterns.As for the eventual adaptive value of

milling, in different cases it may be posi-tive (e.g., by increasing protection fromenemies), negative (e.g., by displacingessential behavior functions), or may haveno special significance. The biologicalvalue of such group behavior, a problem ofconsiderable importance, must be workedout in terms of the general life economy ofthe respective animal forms in which itappears. In the present connection inter-est attaches to the nature of milling and itsrelation to other behavior resources in thegiven animal.

Analyses of milling in two animals haveshown us that in such group functions thecomponent individuals are dominated byaction-engendered stimulus effects to whichthey respond in very reflex-like ways.Studied in itself, obviously, this type ofgroup behavior does not necessarily pro-vide us with a true picture of the givenanimal's capacities for more versatileadjustments under different conditions.It is apparent that the mere occurrence of

the milling pattern may be a misleadingsign of psychological inferiority, since itmay well obscure capacities for plastic,opportune adjustments not dominated bythose native factors which account formilling.

It happens that the army-ant circularcolumn really typifies the nature of theanimal, even though Eciton milling is un-doubtedly rare under field conditions. De-tailed study of Eciton behavior underforest conditions shows that the individualis subject to fixed and prevalent, intrinsic(i.e., native) limitations canalizing its be-havior quite narrowly. However, in itscomplex environment, the animal is pro-vided incidentally with innumerable possi-bilities for variation in behavior throughthe heterogeneity of forest and jungle ter-rain. Such diversity in surrounding stimu-lation permits a limited repertoire ofstereotyped individual reactions to func-tion in the group setting in highly adaptiveways. Any beginning of monotonous be-havior, as we have seen, consequently iscertain to be interrupted in some manner,so that it becomes merely one of a series ofdiversified events in the Eciton raid. Thusit is environmental change that frees thisanimal from the direct consequences of itsnative individual limitations.The frequency with which milling and

similar rudimentary group functions ap-pear in any given animal must be held as afact of uncertain significance, until thelimitations of the animal and its potenti-alities for new adaptive behavior are betterknown. For example, the fact that millingoccurs with considerable frequency inschooling fishes is, of course, no certainindication that they stand psychologicallybelow army ants, in which we have foundthe potentialities for such behavior great,although milling itself is highly exceptionalunder natural conditions. To be sure, thecapacities of some fishes such as the herring

1944] 23


do appear to be rudimentary and highlylimited, yet there are other schooling fishesin which milling occurs frequently, al-though at the same time the resources forvariable and complex behavior aredefinitely wider than in the herring (Brederand Nigrelli, 1935).On its face, then, simplified group be-

havior such as milling suggests merely acapacity for a psychologically reduced be-havior under the given conditions. Withrespect to the general problem of socialbehavior, it is a reasonable proposition thatthe appearance of such elementary grouppatterns may disclose psychological limita-tions which are binding lifetime limitations,or may indicate merely a special propensityfor routine mass activities under particularconditions, with a possible capacity forhigher type adjustments when occasionwarrants. Unfortunately this principle isfrequently neglected in sociological writ-ings.

In mammalian groups, the milling typeof behavior has been featured in folklorepresumably because of its spectacularcharacter, as in descriptions of massedbison rushing in a frenzied stampedefrom prairie fire, sheep jumping fencesblindly in column, and other instances ofpell-niell surging by a horde of animals.This scarcely represents the normal andtypical state of affairs in an adequate way.The scattered literature (Alverdes, 1927;Murchison, 1935) suggests that responsesto the movement of other animals keep thegroup together and thus provide a basis forits activities, yet the simple stimulus-and-response relationships of mass formationappear to occupy a subordinate role in theordinary activity patterns of any mam-malian aggregation. It is mainly at times ofsudden and great excitement that thehighly simplified miffing pattern dominatesgroup behavior to the exclusion of otherfactors. For the most part, group activi-ties in lower mammals appear to beplastically adaptive and rather finely ad-justed to the given situation, as is evidentfrom careful studies such as Darling's(1937) work with red deer and Carpenter's(1934) with howler monkeys. In general,the milling pattern seems to be a highly

subordinate aspect of herd behavior inlower mammals.

It is a matter of considerable interest thatan animal as limited in its capacity for newlearned behavior as the army ant seems tobe may largely escape the non-adaptiveconsequences of great susceptibility forcircular-pattern mass activities only byvirtue of the incidental fact of environ-mental heterogeneity. In contrast, the farsuperior resources of mammals for trialand error learning are available, in additionto mere ecological variety, for admittingappropriate group adjustments that largelyhold the milling type of behavior in thebackground. This should be the case intheir group activities in the natural en-vironment, where opportunities to learnnew patterns are not subject, to the limita-tions that hold in barnyard and stock pen.

It may be observed that the point ap-plies even more emphatically to the socialpotentialities of man himself. Yet fromthe point of view of what may be termedthe "herd instinct" school of sociologicaltheorists, certain writers (e.g., LeBon,1917; Trotter, 1915), operating on a sub-jective and non-experimental basis, haveinsisted upon the inevitability of emotional-ized, individually degraded, regimentedpatterns in the group behavior of mankind.We find Trotter carrying the notion of"herd impulse" to the extreme of assumingthat human subservience to the attitudesof the group is inborn, with the childinstinctively responsive to what Trottermysteriously calls the "voice of the herd."This idea is comparable to that other hoarymysticism, the "gregarious instinct," whichaccounts for animal aggregations bypostulating an inborn tendency to aggre-gate.On whatever level of the animal series we

work, a scientific approach requires us firstto examine the causes of monotonous pat-terns such as milling when they appear,then to study their relation to the potenti-alities of the animal for other types ofbehavior. From this point of view, it isapparent that the "herd instinct" schoolhas erred not only in evading the questionof how such behavior originates, but alsoin underestimating the resources of higher

24 [No. 1253


animals for group versatility under newconditions. Particularly in the case of man,when milling behavior becomes undesirableor obnoxious, it is well to consider thecanalizing, straitjacket influence of socialinstitutions fostering group tendencies forsuch behavior. An enlightening study ofsuch influences as prevalent in contempo-rary Germany has been carried out byFreeman (1940).

It may be observed that while army antsare constitutionally susceptible to the pre-dominance of circular-column behavior andcan be freed from it only by the incidentalfact of environmental variation, man isby no means susceptible in the same sense,with his cortical basis for versatile cor-rective patterns which under encourage-ment may reduce milling to the minor r6leof an occasional subway rush.

BIBLIOGRAPHY

ALVERDES, F.1927. Social life in the animal world. New

York, Harcourt, Brace and Co.BARNES, T. C.

1930. The effect of gravity on the oscilla-tions in the path of an ant (Lasiusflavus nearticus Wheeler): A studyof random movements. Jour. Gen.Psychol., vol. 3, pp. 318-324.

BREDER, C. M., JR.1929. Certain effects in the habits of school-

ing fishes, as based on the observationof Jenkinsia. Amer. Mus. Novi-tates, no. 382, pp. 1-5.

BREDER, C. M., JR., AND R. F. NIGRELLI1935. The influence of temperature and other

factors on the winter aggregations ofthe sunfish, Lepomis auritus, withcritical remarks on the social behaviorof fishes. Ecology, vol. 16, pp. 33-47.

BRUN, RUDOLF1914. Die Raumorientierung der Ameisen

und das Orientierungsproblem imallgemeinen. Jena, Gustav Fischer,pp. 1-234.

CARPENTER, C. R.1934. A field study of the behavior and

social relations of howling monkeys.Comp. Psychol. Monogr., vol. 10, pp.

1-168.DARLING, F. F.

1937. A herd of red deer. London, OxfordUniversity Press.

DUBoIs, R.1899. Sur la soie de la chenille procession-

naire du Pin maritime et sur lamani'ere de la faire filer au fur et amesure de sa production. Ann. Soc.Linn. Lyon, new ser., vol. 46, pp.

125-127.EIDMANN, H.

1927. Die Sprache der Ameisen. Rev. Zool.Russe, vol. 7, pp. 39-47.

FABRE, JEAN-HENRI1903. Souvenirs entomologiques. Paris,

Delagrave, vol. 6.FIELDE, ADELE M.

1902. Notes on an ant. Proc. Acad. Nat.Sci. Philadelphia, vol. 54, pp. 599-625.

FRAENKEL, G., AND D. L. GUNN1940. The orientation of animals. London,

Oxford University Press.FREEMAN, E.

1940. Conquering the man in the street.New York, Vanguard Press.

FRISH, K. V., AND G. R6SCH1926. Neue Versuche flber die Bedeutung

von Duftorgan und Pollenduft fUr dieVerstandigung im Bienenvolk. Zeit-schr. vergl. Physiol., vol. 4, pp. 1-21.

LEBON, G.1917. The crowd. Translation of Psycholo-

gie des foules. London, T. FisherUnwin.

LUBBOCK, J.1882. Ants, bees, and wasps. London,

, Kegan, Paul, Trench and Co.MAIER, N. R. F., AND T. C. SCHNEIRLA

1935. Principles of animal psychology. NewYork, McGraw-Hill Book Co.

MARSHALL, W. S.1904. The marching of the larva of the Maia

moth, Hemilenca maia. Biol. Bull.,vol. 6, pp. 260-265.

MELANDER, A. L., AND C. T. BRUES1906. The chemical nature of some insects,

secretions. Bull. Wisconsin Nat. Hist.Soc., new ser., vol. 4, pp. 22-36.

MILLER, D. F., AND M. GANS1925. Some observations on the reactions

of the ant Cremastogaster lineolata(Say) to heat. Jour. Comp. Psychol.,vol.5, pp. 465-473.

MORLEY, B. D. W.1942. Observations on the nest odours of

ants. Proe. Linn. Soc. London, 154thsess., pp. 109-114.

MURCHISON, C. (ED.)1935. A handbook of social psychology.

Worcester, Clark University Press.PARR, A. E.

1927. A contribution to the theoretical analy-sis of the schooling behavior of fishes.Occas. Papers Bingham Oceanogr.Coll., no. 1, pp. 1-32.

PRATT, K. C.1925. Thermokineties of Crematogaster lineo-

lata (Say). Jour. Comp. Psychol.,vol. 5, pp.265-270.

1944] 25


SANTSCHI, F.1911. Observations et remarques critiques

sur le mecanisme de l'orientation chezles fourmis. Rev. Suisse Zool., vol. 19,pp. 303-338.

1930. Nouvelles experiences sur l'orienta-tion des Tapinoma par s6cr6tionsdromographiques. Arch. Psychol.,vol. 22, pp. 348-351.

SCHLAIFER, A.1942. The schooling behavior of mackerel:

a preliminary experimental analysis.Zoologica, vol.27, pp. 75-80.

SCHNEIRLA, T. C.1929. Learning and orientation in ants.

Comp. Psychol. Monogr., vol. 6, pp.143.

1938. A theory of army-ant behavior basedupon the analysis of activities in arepresentative species. Jour. Comp.Psychol., vol.25, pp. 51-90.

1940. Further studies on the army-ant be-havior pattern. Mass organizationin the swarm-raiders. Ibid., vol. 29,pp. 401-460.

1941. Studies on the nature of ant learning.I. The characteristics of a distinctiveinitial period of generalized learning.Ibid.,vol. 32, pp. 41-82.

1943. The nature of ant learning. II.The intermediate stage of segmental

maze adjustment. Ibid., vol. 35, pp.149-176.

[In press.] Studies on the army-ant behaviorpattern. Nomadism in the swarm-raider Eciton burchelli. Proc. Amer.Phil. Soc., vol. 89, no. 1.

SHAPLEY, H.1920. Thermokinetics of Liometopum apicu-

latum Mayr. Proc. Natl. Acad. Sci.,vol. 6, pp. 204-211.

THORPE, W. H.1939. Further experiments on olfactory

conditioning in a parasitic insect.The nature of the conditioning proc-ess. Proc. Roy. Soc. London, ser. B,vol. 126, pp. 370-397.

1939. Further studies on pre-imaginal olfac-tory conditioning. Ibid., ser. B, vol.127, pp. 424-432.

TROTTER, W.1916. Instincts of the herd in peace and war.

New York, Macmillan and Co.WHEELER, W. M.

1910. Ants, their structure, development,and behavior. New York, ColumbiaUniversity Press.

1928a. The social insects, their origin and evo-lution. New York, Harcourt, Braceand Co.

1928b. Foibles of insects and men. NewYork, Alfred A. Knopf.

26 [No. 1253

museum novitates

Documents