ON THE FEEDING MECHANISM OF THE COPEPODS,CALANUS FINMARCHICUS AND DIAPTOMUS

GRACILIS

BY H. GRAHAM CANNON,

Professor of Zoology, Sheffield University.

(Received June 28th, 1928.)

(With Eight Text-figures.)

INTRODUCTION.

IN 1925, Storch and Pnsterer described at great length the feeding mechanism ofthe freshwater copepod, Diaptotmis gracilis. They maintained that food was filteredfrom a current produced by the activities of the head swimming limbs. Theiranalysis of the mechanics of this process appeared to me inaccurate so that Idecided to re-investigate the problem.

I examined first Calanus finmarchicns, a form so similar to Diaptomus that Iassumed its feeding mechanism would be essentially the same. My observations,however, on the actual currents produced were totally different from those de-scribed by Storch and Pnsterer and I decided to obtain Diaptomus gracilis itself.This I obtained through the kindness of Mr R. Gurney and I found that its feedingmechanism and the currents it produced agreed closely with those of Calanus.

My observations agree with those of Storch and Pfisterer in that I describefood particles as being retained by the maxilla from a current which is caused topass through it. I agree further that this feeding current results from the swimmingactivities of the anterior limbs. It is the swimming current that I consider theseworkers have described inaccurately and this is of vital importance to their argu-ment, as their analysis of the mechanics of the process depends primarily on itscorrect interpretation.

Part of the observations on Calanus were made while occupying the table ofthe Royal Microscopical Society at the Marine Biological Laboratory at Plymouth.

METHODS.The currents produced by the copepods were observed under the microscope

by placing coloured starch grains in the water in which they were swimming.The movements of the limbs are so rapid that it is impossible to analyse them

by ordinary methods. Storch and Pfisterer (1925, p. 347) estimate a frequency of300 beats a minute for the movements of the head swimming limbs of Diaptomus.In the specimens I observed, a rough estimate was 1000 a minute. I succeeded,

9-2

132 H . G R A H A M C A N N O N

however, in observing them with complete accuracy by using a stroboscopic sourceof illumination for the microscope.

The source of light, an ordinary "opalite" bulb, was placed behind a rotatingdisc which was pierced at equal distances by narrow radial slits whose width couldbe adjusted. The disc was 9 inches in diameter and was pierced at the edge by fourslits about 1J inches long. The best results were obtained when the slits were aboutJ inch wide. The disc was rotated by a motor whose speed could be controlledby a variable resistance.

The copepod was placed in a compressorium in a small drop of water, as deepas possible, to allow complete and unhindered movements of the limbs. It wasthen focussed under the microscope and the disc caused to rotate at graduallyincreasing speed. At first only irregular images of the animal were obtained, but,as the frequency of the flashes of light gradually approached that of the limb move-ments, the limbs appeared to move regularly and, for a copepod, very slowly.By increasing the speed still further I found it possible to obtain an image of thelimbs apparently moving as slowly as desired. At the critical point, when thefrequency of the light coincided with that of the limbs, the latter appeared to standstill, but never quite still. I assume that this means that the limb movement isnever absolutely regular. By increasing the speed of rotation beyond the criticalpoint the limbs appeared to move slowly but in the reverse direction. In makingobservations care has to be taken that the real direction of movement is beingobserved and not the reverse. This was extremely important in studying the move-ment of the maxilliped whose tip describes a rotary movement. It is, however,quite easy to settle whether the correct movement is being studied, because, inthis case, a slight diminution in the frequency of the light makes the limbs moveapparently faster, while if it is the reversed movement, the limbs appear to movemore slowly or else reverse their apparent motion.

By placing the source of light so that only half is covered by the rotating disc,it is possible to obtain the field of the microscope illuminated, on the one side, bycontinuous light, and on the other, by intermittent light. In this way I have vieweda copepod with the limbs of one side moving at their normal speed while thoseon the opposite side appeared to move extremely slowly. This is of great use instudying such a form as a copepod whose movements are apt to be spasmodic.By shifting the slide from the stroboscopic to the continuous half of the field itcan be settled at once whether or not the copepod is moving normally.

ANATOMY.

The main points in the anatomy of Diaptomus and Calanus relevant to a de-scription of their feeding mechanism may be summarised briefly.

Motion through the water is of three types: (1) a sudden rapid jerk forwards,produced by the activity of the trunk swimming limbs, (2) a series of much smallerjerks produced by spasmodic irregular movements, and (3) a steady and compara-tively slow forward movement resulting from continuous and rapid vibrations ofthe antennae, mandibular palps, maxillules and maxillipeds. The first type serves

Feeding Mechanism of Copepods 133

as a means of escape, the second is probably a means to counteract a tendency tosink while the third type produces the feeding current.

The uniramous antennules project laterally and act as balancers.The biramous antennae project ventro-iaterally just in front and to the sides

of the large upper lip or labrum (see Text-figs. 1 and 5). The exopodites curvelaterally and then dorsally close against the body. The endopodite projects ventrally

AY:- w

•X

Text-fig. 1. Sketch of lateral view of Calanus finmarchicus in resting position. The antennuleshave been cut off close to the body. ant. i =antennule; ant. z endop. =endopodite of an-tenna; ant. 2 exop. =exopodite of antenna; Ibr. =labrum; mdb. =mandible; mx. i =maxillule;mx. 1 ex. =exite of maxillule; mx. 2 =maxilla; mxpd. =maxilliped; s.t.l. =swimrning trunk limb.

at an angle of about 300 to the sagittal plane. The endopodites and basal part ofthe exopodites are armed with long setae spread out in a ventro-lateral fan. Theexopodite terminates in a group of three long dorsally directed setae.

The mandibules are wedged in between the distinctly bifid lower lip and themassive upper lip, being slightly overhung by the latter. The biramous mandibularpalps project ventro-iaterally and carry a fan of setae spread out over the same angleas that of the endopodite of the antenna. These setae are shorter than those of the

134 H. GRAHAM CANNON

antenna. The maxillules arise nearer the sagittal plane than the more anteriorlimbs. Their main axes project antero-ventrally overhanging slightly the mandi-bules. Both endopodites and exopodites terminate in long setae which, however,are shorter than those of the mandibular palps. Those of the endopodite projectdirectly ventrally and slightly medially, those of the exopodite project in the samedirection as those of the mandibular palp. The maxillule is armed medially withthree endites, a powerful toothed basal endite projecting obliquely forwards tothe split in the lower lip and two distal endites which project forwards, theirterminal setae lying across the mouth. On the outer side of the maxillule there isan exite armed with very long slender setae. The most ventral of these projectlaterally and then curve round posteriorly and extend as far back as the first pairof trunk swimming feet. The more dorsal, that is those next the body wall, projectalmost directly posteriorly. A section of the fan of setae thus formed would extendover an arc of a quarter of a circle.

setae of mc.*-.lluiavy

o / chamber. ^ I

ot

trunk swimming limbs'

VText-fig. 2. Diagram of transverse section through filter and suction chambers of copepod.

•The maxillae are short uniramous limbs projecting ventro-anteriorly. Theycomprise eight joints bearing long plumose setae which extend forwards to themouth. They thus form the walls of a median wedge-shaped space with the mouthat its apex (see Text-figs. 2 and 5).

The maxillipeds are cylindrical uniramous limbs arising close behind themaxillae. In the living form it is difficult, with continuous light, to see them apartfrom the maxillae. Each consists of a very short basal joint, followed by two com-paratively long joints of doubtful homology and (finally) a short flexible setoseportion of five joints. The first long joint lies close against the maxilla and reachesas far as its tip. The remainder of the limb then stretches ventro-anteriorly andlaterally, the setae spreading out in a fan close underneath the tips of the maxillules.

The trunk swimming limbs extend obliquely forwards, the apex of the anteriorpair reaching as far as the mouth. They converge to a point slightly nearer thebody than the tips of the maxillae.

Feeding Mechanism of Copepods 135

The figures in Storch and Pfisterer's paper are incorrect in several importantdetails. The setae of the maxillulary exite are figured in a parasagittal plane. Themaxillipeds are too short and project inwards instead of outwards and the swimmingtrunk limbs reach only as far as the maxillae instead of the mouth.

This arrangement of limbs results in two spaces of importance in the feedingmechanism (see Text-figs. 2 and 5). The first of these is the filter chamber betweenthe maxillae, the walls of which are the maxillae, the roof is the body wall, and thefloor the tips of the anterior trunk swimming limbs. The floor is complete laterallyexcept for a small split occurring between the trunk limbs and the ventral setaeof the maxillae. Anteriorly this space is closed by the large upper lip. Its onlyentrance is posteriorly between the maxillipeds and the first pair of swimming trunklimbs. The second of these spaces, which I call the suction chamber, is boundedmedially by the maxillae and laterally by the setae of the maxillulary exite.

SUMMARY AND CRITICISM OF THE VIEWS OF STORCH ANDPFISTERER ON THE FEEDING MECHANISM OF DIAPTOMUS.

The essential points in the feeding mechanism of Diaptotnus according toStorch and Pfisterer may be briefly summarised as follows:

The activity of the swimming limbs produces a powerful antero-posterior cur-rent which runs close against the ventral side of the body. This does not agreewith "eine langsame, gleichmassige, gleitende Vorwartsbewegung" (p. 338). Apowerful current passing directly posteriorly means that a considerable amount ofwater passes backwards and hence the body causing this must pass forwards withconsiderable speed.

They point out, and lay considerable stress on the fact, that the head swimminglimbs diminish in length posteriorly. Each limb produces the maximum move-ment of water at its tip. The mandibular palp being shorter than the antennaproduces its maximum effect nearer the body wall. "So setzt also die Tatigkeitder drei Gliedmassenpaare eine verhaltnismassig hohe Wasserschicht in Bewegung,und zwar in der Weise, dass je das folgende Gliedmassenpaar die bewegte Wasser-schicht naher an den Korper herantragt." (p. 351.) That is, each limb draws thecurrent produced by the limb in front, nearer the body wall. They do not attemptto explain the mechanics of this process, which I believe to be erroneous, but whichis, however, essential to their analysis of the feeding mechanism.

Assuming, as do Storch and Pfisterer that the limbs move synchronously, thespeed of the tip of a limb will depend on the length of that limb and the arc throughwhich the limb swings in one vibration. Storch and Pfisterer do not consider thelatter factor, and, presumably, assume that each limb moved through the samearc. In this case the speeds of the tips of the limbs are directly proportional to thelengths of the limbs. The tip of the antenna moves faster than that of the mandibularpalp, and the latter faster than the maxillule. Hence the movement of the tip ofthe mandibular palp can have no effect on the layer of water set in motion by thetip of the antenna because it will be moving more slowly than the latter. If themandibular palp moved over a much greater arc in one vibration than did the

136 H . G R A H A M C A N N O N

antenna it would move faster, in which case it might conceivably draw some of thewater set in motion by the antenna nearer to the body. Actually, however, this isnot so. The mandibular palp and maxillule both appear to move over the samearc while the antenna moves through a much larger angle. Hence the layer ofwater set in motion by the tip of the antenna moves much faster than that forcedbackwards by either mandibular palp or maxillule.

According to Storch and Pfisterer then, the swimming current is drawn closeto the body by the swimming limbs. This is further enhanced by the maxillularyexite. They say (p. 351) that because of its dorso-ventral axis of rotation, andbecause it lies close against the body wall it is especially adapted to increase stillfurther the moving layer of water and draw it nearer to the body wall. The samecriticism applies here as above. Being very close to the body wall its speed ofmovement through the water is very slow compared with the tip of the antennaand it can have no effect whatever in drawing the current closer to the body wall.

The current having been drawn close to the surface of the body passes back-wards on either side, through the space between the forwardly projecting maxillaand the backward exite of the maxillule. The shape and arrangement of the latter,according to Storch and Pfisterer (p. 352), indicate that its function must be toconfine the moving layer of water, thus preventing it spreading out too quicklyand so losing force. From this it might be concluded that the maxillulary exitewas a comparatively rigid plate. Actually the setae are very fine structures, rigidonly at their bases and whip-like at their extremities. If the exite did move back-wards and forwards as these workers maintain, the pressure on the setae would forcethem outwards. They certainly could not act as a barrier confining a stream ofwater which, as Storch and Pfisterer emphasise, is of considerable strength.

This powerful backward current will suck water into it from still regions. Sucha region occurs only, according to Storch and Pfisterer (p. 353), between the basalendite of the maxillule on the outside and the tips of the maxillary setae on theinside. Why this position is chosen to the exclusion of others is not stated. Wateris therefore sucked from this region into the backwardly flowing stream and toreplace it water passes through the maxillary setae from the anterior part of thefilter space. This results in a region of low pressure close behind the mouth which,in its turn, sucks water forwards from behind and, to supply this, water fromthe ventral part of the swimming stream is sucked into the filter space behind themaxillipeds (Storch and Pfisterer, Fig. 11).

Particles carried on the stream of water drawn in in this way are retained inthe filter space, sucked forwards and eventually filtered off b} the maxillary setae.They are then combed off by the proximal endite of the maxillule and pushedforwards on to the mandibles.

With regard to the maxillipeds Storch and Pfisterer state " Die Maxillipedenbewegen sich nur in einer sehr geringen Amplitude von hinten aussen nach vorneinnen und umgekehrt" (p. 347), but can offer no explanation of their function(P-355)-

The description and mechanical analysis of the feeding mechanisms as given

Feeding Mechanism of Copepods 137

by Storch and Pfisterer thus depend on two critical points, the production of apowerful antero-posterior swimming current by the head swimming limbs, andthe suction of that current closer to the body by the more posterior of those limbs.The latter I have shown is mechanically impossible. The former rests on aninaccurate observation as no such current exists.

FEEDING AND SWIMMING CURRENTS.

In a ventral or dorsal view of either Calanus or Diaptomus swimming slowlythrough the water, there can be seen, very readily, two large swirls1, one on eitherside of the body in the angle between the antennule and the axis of the body (seeText-fig. 3). The centres of the swirls are indefinite but occur usually in Calanus

Text-fig. 3. Diagram of ventral view of Calanus finmarchicus slowly swimming to showwater currents.

about the middle of the total length of the body, and in Diaptomus further forwards.They rotate in such a way that the water nearest the body moves backwards. Theyare continuous underneath the body so that, in side view, a marked swirl is obviousventrally (see Text-fig. 4). Dorsally also there is a swirl, but this is much lessmarked than ventrally. The copepod thus moves steadily forwards in the middle ofa vortex of moving water, which may be termed the " swimming vortex.'* A certainamount of water is drawn into the vortex from in front, and a certain amount passesout posteriorly, and this represents the motive force of the copepod when swimming

1 The term "swirl" is used to indicate a rotary movement of water. I have not used it assynonymous with " vortex." If a vortex is viewed laterally, it will appear in the microscope, pro-vided that the plane of the image approximately bisects the vortex, as two separate swirls rotating inopposite directions about the annular axis of the vortex.

138 H. GRAHAM CANNON

in this fashion. However, the drift towards the body anteriorly is not very markedwhen compared writh such a form as Hemimysis where water streams towards the headfrom all directions, and is thrown out posteriorly in a powerful swimming stream.

A second smaller vortex occurs inside the swimming vortex, rotating in theopposite direction. In a ventral view it can be seen as two swirls at the sidesof the anterior ends of the suction chambers. In side view it cannot be seenin the median plane as it is interrupted by the forwardly projecting swimmingtrunk limbs, but a little to one side it can be seen to extend forwards betweenthe maxillae to the bases of the anterior swimming limbs and then pass ventrally

Text-fig. 4. Diagram of lateral view of Calanus finmarchiciis slowly swimming to show water currents.

and backwards. This vortex is the "feeding vortex." It cannot be traced over thedorsal side of the body.

The inner part of the swimming vortex probably represents the "powerfulantero-posterior current" described by Storch and Pfisterer for Diaptomus. Itis certainly powerful but it is a vortex and this accounts for the fact that the copepodmoves forwards comparatively slowly. According to Storch and Pfisterer, a con-siderable amount of water is transported in an antero-posterior direction, and inorder to take its place the body would have to move forward with correspondingspeed. If, however, the water which moves backwards at the level of the tips ofthe swimming limbs, the level at which the velocity must be greatest, moves atthe same high speed which Storch and Pfisterer describe, but is part of a vortical

Feeding Mechanism of Copepods 139

movement, there is comparatively very little actual transport in an antero-posteriordirection and hence the body moves slowly forward. The energy of the limbs isexpended, not on transporting water backwards, but in maintaining a vorticalmovement in the water and very little energy is required for this. This may accountfor the extreme rapidity of the limb movement. They have very little resistance toovercome, simply the viscous drag on the vortex, and hence are able to vibrateat a speed which is very considerable for a Crustacean limb.

The feeding swirl represents the suction through the maxilla described forDiaptomus.

The swimming vortex results from the vibrations of the antennae, mandibularpalps, and distal part of the maxillules. The feeding vortex is a necessary resultantof the swimming vortex, but is increased by the activities of the maxillularyexite and maxillipeds. This was demonstrated clearly in a specimen of Calanuswhich was nearly dead. All its limbs had ceased moving except the antennae andmandibular palps, and the antennules had not flexed backwards as happens whena copepod dies. The swimming vortex was almost as large as normal and therewas a very pronounced feeding vortex. This does not agree with Storch andPfisterer's account of Diaptomus which regards the maxillule as of vital importancein producing the feeding current.

LIMB MOVEMENTS AND CURRENT PRODUCTION.

When Calanus and Diaptomus are swimming slowly forwards the frequency ofthe head swimming limb movements is remarkably but not absolutely constant.In the other types of movement the limbs move irregularly, and hence their vibra-tions cannot be analysed stroboscopically. It is the steady forward motion whichresults in the feeding current and that is the type analysed here.

The maxilla shows no rhythmical movement. For the greater part of the timeit remains still, but if the mouth becomes congested with food particles, the maxillaemay be flexed ventrally throwing the accumulated food away from the body intothe swimming swirl.

The maxillipeds, maxillules, mandibular palps and antennae vibrate regularly,in the case of Diaptomus at the rate of about 1000 times a minute and in the caseof Calanus about 600 times a minute. Their movements are synchronous but notin the same phase, as apparently Storch and Pfisterer assumed. They exhibit amarked metachromial rhythm of the type shown in other Crustacea such as Chiro-cephalus or Nebalia (Cannon 1927, 1928), that is, each limb commences its backstroke just before the limb immediately anterior to it. The phase differences aresuch that the maxilliped commences its back stroke (outward movement) just afterthe antenna commences its forward stroke, so that these two limbs move almostin opposite phase. The phase difference between mandibular palp and maxilluleis very small. The back strokes of antennae, mandibular palps and maxillules arefaster than their fore strokes. The tip of the maxilliped exhibits a rotary movement inan obliquely frontal plane (see Text-fig. 5) such that the outward part of the rotation

140 H. G R A H A M C A N N O N

(backward stroke) is faster than the inward part. The movements of these limbsare represented graphically in Text-fig. 6.

Cutrenis. Lin-.l; Mouemen+s.Text-fig. 5. Diagram of anterior region of Calanus finmarchicus. The endopodite of the antenna,

the mandibular palps and the distal parts of the maxillules have been removed. The position ofthe swimming trunk limbs is indicated by the shaded area inside the dotted line. On the rightside of the figure the limb movements are indicated, on the left, the water currents, ant. 1 =an-tennule; ant. 2 =antenna; ant. 2 ex.r. =rotation path of tip of exopodite of antenna ;f.ch. =filterchamber; Ibr. =labrum; mdb. =mandible; mx. 1 =maxillule; mx.x ex.r. =rotation path of tipsof setae of maxillulary exite; mx.z =maxilla; mxpd. =maxiliiped; mxpd.r. =rotation path of tipof maxilliped; s.ch. =suction chamber.

The antennae move through by far the greatest arc. The innermost setae ofthe endopodite, at the end of the back stroke, reach the tips of the njaxillules. Therange of movement decreases in the more dorso-lateral setae. This results from thefact that, as the endopodite moves backwards the exopodite swings forward. The

Feeding Mechanism of Copepods 141

former moves practically in a straight line sloping posteriorly towards the sagittalplane. The tip of the exopodite rotates in an ellipse passing nearer to the sagittalplane on its forward stroke (Text-fig. 5).

It is difficult to say what is the meaning of the recurved exopodite. It mayfunction in assisting the feeding swirl but it also may serve to diminish the centri-fugal drag on the base of the limb. The tissues of a copepod must be relativelyheavy compared with water, especially in such forms as Diaptomus and Calanuswhich reduce their total specific gravity by the production of a drop of light oil.Consequently the mass of the moving limb must be considerable. The positionof the exopodite does not reduce the inertia of the limb but it shifts the centre ofgravity almost to the axis of rotation, and this must diminish the centrifugal dragon the body.

anienna.

mandible.

maxiflule.

OuiLiord StrakP OT

= Stichon from filler chamber

maxiiii.'ined.

outward siroke = suction info filter chamber.

Text-fig. 6. Graphic representation of the movements of the head swimming limbs ofCalanus finmarchicus.

The mandibular palps and maxillules move backwards and forwards in a para-sagittal plane over a comparatively small arc.

The swimming swirl is produced by these three limbs, and, of these, the an-tennae are the most effective. The feeding swirl results primarily from the swimmingswirl. This can be best understood by considering the path of a jet of water whensquirted into a volume of still water. If, from a cylindrical jet, a small mass ofwater is suddenly expelled it does not move forwards as a moving cylinder of fluidbut spreads out into the form of a vortex, and this is the more marked the greaterthe velocity of the moving jet (Text-fig. 7, a). Considering two points, one just insidethe cylinder of fluid immediately it has left the nozzle, and another point just out-side in the still water, there will be marked discontinuity between the velocitiesat these two points and discontinuity in velocity in a fluid leads to vorticity. Or

142 H. GRAHAM CANNON

again, the viscous drag of the moving cylinder immediately it has left the nozzle,will drag in still water from behind while the viscous resistance at the front endof the cylinder will flatten it out and these two effects together will produce thevortex.

If now, instead of a simple jet, an annular cylindrical jet is considered, the samereasoning applies. The moving water will tend to spread out. It will move outwardsfrom the axis of the jet but, at the same time, it will spread inwards and so producean inner vortex (Text-fig. 7, b).

This latter type of jet is produced by the steadily swimming copepod. Thehead swimming limbs move at greatest speed at their tips while their bases, beingattached to the body, are stationary. Their setae spread out in fans extendingalmost half way round the body. The result is that they produce a moving layerof water in the form of half an annular jet (Text-fig. 7, c). This spreads outwardsas the swimming swirl but at the same time spreads inwards and produces thefeeding swirl.

a

Text-fig. 7 (a) Diagram of swirl produced by jet from simple tube.(h) „ „ „ annular tube.(c) „ „ „ antennae of copepod.

In the copepod there is an additional effect which increases this vortex produc-tion. In the annular jet there is a continuous supply of water being forced throughthe nozzle. In the copepod this supply comes from the spaces between the swim-ming limbs. As the antenna moves backwards it obliterates the space between itand the mandibular palp and forces out the water into the swirls. On extendingforwards this space is opened out and water must pass in again. It will not passin from the tip of the limb to any great extent as water in this region is movingrapidly backwards. There is, however, a slight tendency foi water to pass in, forif particles are watched passing over the tips of the swimming limbs they are seento pass slightly inwards towards the base of the limbs but are immediatelythrown out again on the backstroke. The main mass of water will naturally besucked in from the bases of the limbs where the water is relatively still. That is, aregion of low pressure must exist at the bases of the swimming limbs. This willsuck in water partly from in front (Text-fig. 5) and partly from behind. The lattersuction will serve to increase the feeding swirl,

Feeding Mechanism of Copepods 143

The action of the maxiliulary exite and maxilliped is to force part of the feedingswirl through the comb of filter setae on the maxillae. The two limbs co-operate.The maxilliped sucks water into the filter chamber while the maxiliulary exitesucks it out of this chamber through the maxillary setae.

The movement of the maxillipeds has already been described (p. 139). On theiroutward stroke the setae on the distal joints spread out into a fan, so that a suctionis produced in a ventro-lateral direction. The maxillipeds lie just underneath thesplits between the tips of the trunk swimming limbs and the maxillae (Text-fig. 2).Hence the suction must extend into the filter chamber and draw in water frombehind.

Just after the maxillipeds have finished their outward stroke the maxillulescommence to move forwards (Text-fig. 6). The maxiliulary exites, at the end oftheir back stroke lie flat against the outer faces of the maxillae, thus diminishing,

\

Diagram showing four consecutive phases in the movement of the maxiliulary exite.

and at the same time closing the suction chamber between them and the maxillae.On their forward stroke they simply tend to enlarge this space and so produce init a region of low pressure. The extent of this suction can be seen from the curvatureof the setae as the exite moves forwards (Text-fig. 8). The suction effect of the setaeis increased by their armature of setules. These project laterally from the outerfaces of the setae so that, on the outward movement of the exite they spread outand fill up the inter-setal gaps, while on the inward movement they collapse andallow the escape of water through the setae (Text-fig. 8). The maxillipeds andmaxillules thus work together. The former sucks water into the filter chamberand this is immediately followed by a suction of the maxillule which draws waterthrough the maxillary setae. In the back stroke of the exite the setae spread outand sweep the water backwards into the swimming swirl.

In the backwards and forwards motion of the maxillule the setae on its exite,

144 EL G R A H A M C A N N O N

which project at right angles to the axis of the limb, move through the same angleas the limb itself. The tips of the setae thus move in a dorso-ventral direction.This, combined with their in and out motion resulting from the suction activityof the exite, results in the tips of the setae moving in a flattened ellipse lying closeagainst the ventro-lateral body wall (Text-fig. 5). The tips of the exites thus beattowards the posterior opening of the filter chamber. Thus, while the anterior partof the maxillulary exite is producing suction in the suction and filter chambers,the posterior whip-like ends are actively sweeping particles into the latter.

Small particles sucked into the filter chamber are deposited on the setae ofthe maxillae. These are heavily armed on their inner faces with laterally projectingsetules so that the whole limb forms an efficient filter (Text-fig. 2).

Particles so filtered, if they happen to be deposited near the anterior end of thechamber are scraped off, as Storch and Pfisterer (p. 354) point out, by the enditesof the maxillule and passed directly forwards on to the mandibles. In addition,however, there are several long setae which arise on the basal joints of the maxilli-peds and project forwards on to the inner faces of the maxillae. These scrapeparticles off the hinder parts of the filter plates and push it forwards on to themaxillules. The maxilla is thus brushed clean on both its inner and outer faces.

SUMMARY.

1. Calanus finmarchicus and Diaptomus gracilis both feed automatically whenswimming slowly and steadily through the water.

2. A feeding current is produced which is filtered by the stationary maxillae.Food so obtained is passed on to the mandibles by the maxillulary endites andsetae on the bases of the maxillipeds.

3. The feeding current is a vortex passing through the mouth parts whichresults automatically from the swimming activities of the antennae, mandibularpalps and maxillules.

4. The feeding vortex is caused to pass through the maxillae by the combinedactivities of the maxillipeds and the maxillulary exites. The former suck waterinto the filter chamber between the maxillae while the latter suck it out throughthe maxillary setae.

5. The views of Storch and Pfisterer on the feeding mechanism of Diaptomusgracilis are criticised. There is no powerful antero-posterior swimming currentas described by these authors. The swimming current is in the form of a vortexencircling the body and most marked at the sides in the angle between the bodyand the antennules.

LITERATURE.CANNON, H. G. (1927). "On the Feeding Mechanism of Nebalia bipes." Trans. Roy. Soc. Edin.

55-3SS-7O.(1938). "On the Feeding Mechanism of the Fairy Shrimp, Chirocepkalus diapkanus." Trans.

Roy. Soc. Edin. 55. 807-23.STORCH, O. and PFISTERER (1935). "Der Fangapparat von Diaptomus." Zs. vergl. Physiol., Berlin,

Bd. 3. 330-76.