The Later Larval Development of Amphioxus. · 2006-05-19 · LATER LARVAL DEVELOPMENT OP AMPHIOXUS....

LATEB LARVAL DEVELOPMENT OP AMPH1OXTJS. 183

The Later Larval Development of Amphioxus.

Arthur Wllley, B.Sc.(Of University College, London).

With Plates XIII—XV.

IN the account of the development of the atrial chamber ofAmphioxus which appeared in this journal last August, underthe joint names of Professor Lankester and myself, it wasstated that we had not been able to trace the progressive stepsin the history of the gill-slits, during their passage from theunilateral position which they hold in the larva, to the sym-metrical condition which we find in the adult; nor the trans-lation of the lateral mouth of the larva to the anterior medianposition of the mouth in the adult.

We added that we had taken steps to obtain the criticalstages in the living condition during the summer of last year(1890). I accordingly went down to Messina, for the secondtime, in July, to undertake the study and to make drawings ofthe living larvae—being enabled to do so by the continued kind-ness of Professor Lankester, to whom for this, as also for invalu-able assistance and advice since received, I am deeply indebted.

According to Professor Lankester's instructions I spent themonths of July and August at Faro, daily fishing in thePantano, and concentrating my attention on larvae of theparticular stages which were required. To obtain a largenumber of individuals in the desired condition of transition,from the extreme asymmetry of larval life to the nearly sym-metrical condition which is reached with scarcely appreciable in-crease of size, was no easy matter. By examining daily a greatnumber of larvae I succeeded in obtaining an ample series of the

184 ARTHUR W1LLEY.

transitional forms, which were carefully drawn and described asobserved in the living condition. These observations form thesubstance of the present memoir. I may possibly find it desir-able to add to this hereafter an account of some of the struc-tures involved, as determined by means of sections. For thisdetermination I have an ample supply of preserved material.

HABITS OF THE

The conditions under which the material was obtained lastyear differed in a curious way from those of the preceding year.In 1889 the larvae of Amphioxus were present in the lake at Faroin great numbers, especially during the months of July andAugust; while last year, during the same months, they werecomparatively rare, and their place seemed to have been takenby an incredible host of Doliola of the first or larval generation.

Exactly in what way the presence of vast numbers ofDoliola affected the larvae of Amphioxus I was not able tomake out, because the spawning of Amphioxus was prolific inthe extreme. Possibly the breeding of Amphioxus as awhole commenced rather earlier last year, and the larvse mayhave taken to the sand mainly before July. I found numerousDoliola with ova and gastrulse of Sagitta in their pharynx, some-times as many as nine ova in one Doliolum. This might leadto the death of the embryos in question, as it appears to causethe death of the Doliolum. I did not, however, find the ova ofAmphioxus in this position, or possibly only in one or two doubt-ful cases. At all events, the scarcity of the larvse of Amphioxusat this particular time of the year was evidently in correlationwith the great predominance of Doliolum.

Professor Kleinenberg informed me of an analogous fact,namely, that Salpse are sometimes very numerous in the harbourof Messina, to the exclusion of other small pelagic organisms.

When placed in glasses containing fairly clean water thehealthy larvse of Amphioxus are seen to be suspended, appa-rently motionless in the water, in a highly characteristicver t ica l position. The suspension is no doubt effected by themovement of the long cilia with which the epidermis is provided.

LATER LARVAL DEVELOPMENT OP AMPHIOXUS. 185

It will be remembered as a curious circumstance that the adulthabitually assumes a similar vertical position in the sand.1

The time of life at which the larva seeks its home in the sandvaries greatly. While, on the one hand, I have dredged all thestages described below pelagically, I have, on the other hand,found larvae which had only reached the third stage of the laterlarval development (see below; also PI. XI I I , fig. 4) in the sand.

RESUME OF THE ENTIRE DEVELOPMENT OF AMPHIOXUS.

The various phases in the development of Amphioxus maybe conveniently arranged in the following way :

I. THE PERIOD OF EMBRYONIC DEVELOPMENT comprising the

first thirty-two hours. It commences with the segmentationof the ovum, and ends with the formation of the mouth andfirst gill-cleft. According to the habit of the embryo, thisperiod may be subdivided into—

(a) The time—namely, the first eight hours—during whichthe rapidly developing embryo is confined within the vitellinemembrane, the successive stages being marked by the progressof the segmentation and gastrulation, the commencement ofthe formation of the myocoelomic pouches, the differentiation ofthe medullary plate, and the formation of the neurenteric canal.

(6) The time which elapses between the emergence of theciliated embryo from the vitelline membrane, and the appearanceof the mouth, first gill-cleft, and anus—the stages being markedby the successive formation of myoccelomic or archentericpouches to the number of fourteen pairs. The myotomes

1 The accounts of Amphioxus given by Rathke (Konigsberg, 1841) and J.Miiller (Berlin, 1844) have given rise to the impression that the usual statusquo of Amphioxus is, to be lying on its side on the sand. As stated above,this is not the case. If it were so, it might be supposed to offer a simpleexplanation of the one-sided character of the larva—not that I think it would,however. J. Miiller, indeed, says that Amphioxus is fond of burying the caudalhalf of its trunk in the sand, as an occasional divertissement. The lyingon one side, however, is what is occasionally done; and that is the result,not so much of a subtle inherited tendency to assume that position of rest, asof a gross incapacity on the part of Amphioxus to maintain its equilibrium inany other way when out of the sand.

186 ARTHUR WILLET.

which are added after this period never communicate with theintestine (Hatschek).

II . THE PERIOD OF EARLY LARVAL DEVELOPMENT during

which fresh gill-slits appear one after the other—i. e. meta-merically—slightly to the right side of the median line (sub-sequently passing well up to the right side), to the number oftwelve to fifteen. Towards the close of this period the longi-tudinal metapleural folds appear, and the closure of the atriumcommences behind by the fusion of the small subatrial ridgeswhich are developed on the inner faces of the metapleura.

III . THE PERIOD OF LATER LARVAL DEVELOPMENT during

which the second row of gill-slits is formed on the right side;the first or primary row of slits passes across to the left side,the mouth assumes an anterior median and vertical position,the prseoral cirri appear, and the endostyle is developed fromits pre-existing rudiment.

IV. THE ADOLESCENT PERIOD when the young Amphioxus,

having now attained most of the essential features of the adultstructure, has definitely ceased to lead a pelagic life, and hastaken up its abode in the sand, where its further growth in sizeand maturity is accomplished.

Of the above four main periods in the development ofAmphioxus, the first has been studied by Kowalewsky (2) andmore thoroughly by Hatschek (3); the second has been treatedby Professor Lankester and myself (8), and the third forms thesubject of the present paper. The fourth entails the considera-tion of the adult anatomy, for which Professor Lankester'spaper (5) may be consulted.

The only observations recorded as to the course of events inthe third period were made by Kowalewsky (1), whose all toobrief account was considered so extraordinary that Balfour1 was" tempted to suppose that his observations were made on patho-logical specimens."

Kowalewsky confined himself wholly to the question of theorigin of the second row of gill-slits, and with reference tothat he said that six disc-like thickenings appear together on

1 'Comparative Embryology,' vol. ii, 1885.

LATEB LARVAL DEVELOPMENT OF AMPHIOXUS. 187

the right side above the row of primary slits. The secondarythickenings then become perforated and increase in size, andas they do so the primary slits pass under the pharynx, and soto the left side.

In his figures he places the first secondary slit immediatelyover the second primary. This is not correct. He did notdescribe the formation of additional secondary slits, and he didnot observe the closure of any of the primary slits.

He was right as to the number of secondary slits which firstappear together, and as to their formation on the right sideabove the primary slits, with the consequent moving of thelatter round to the left side.

My observations on the gill-slits, therefore, in the first placeconfirm, in the second place correct, and in the third place addto Kowalewsky's description. Over and above this, the forma-tion of the velum, the origin of the buccal or praeoral cirri, thefate of the club-shaped gland, and the development of theendostyle will be found fully described below.

In the following account the later larval development will,for convenience, be divided into eight stages. The descriptionof the separate stages will be followed by a brief summary ofthe stages, and this by a further summary of the facts relatingto the various organs; the paper will be concluded with a fewgeneral considerations.

S tage I.—Pigs. 1, 2, 20, and 21.

This is the stage immediately succeeding the older of thetwo stages figured in the paper referred to above (8), and theprimary gill-slits and metapleurs present very much the cha-racters there described. Thus there are fourteen primaryslits (fig. 1), most of which open directly to the exterior; butthe last three or four discharge into the atrial tube, which,while closed behind, is widely open in front, where the rightmetapleur is seen to overhang the anterior slits.

The new feature with which we have now to deal is theappearance of a continuous ridge in the right wall of thepharynx above the row of primary slits.

188 AETHTJU WILLBT.

On this ridge are developed typically six oval thickenings orenlargements, alternating with the primary slits and appearingsimultaneously. They consist of a fusion of the pharyngealwall with the body-wall at six different points, the first pointlying above and between the third and fourth primary slits.They are the forecast of the second row of slits. Between theridge with its nodal thickenings and the row of primary slitsa longitudinal blood-vessel is seen to hold its coursej thisbecomes eventually the ventral or subintestinal vessel whichlies beneath the eudostyle (see Lankester, 5).

The first and last thickenings, as Kowalewsky also noticed,are usually rather smaller than the intervening ones.

Slight deviations from the above mode of procedure occur.One of these is shown in fig. 2, where only five secondarythickenings have appeared at once, the first of them occurringbetween the fourth and fifth primary slits, while the one whichusually has the first place at this stage has been somewhatretarded in its development.

In other cases, in which there are also only five thickenings,it may be the sixth that is late in appearing.

In the larva represented in fig. 2 there were only twelveprimary slits. As mentioned above, the number of primaryunpaired slits which are formed in the larvae varies from twelveto fifteen. The most usual number is perhaps fourteen, whilefifteen is exceptional.

In front of the gill-slits is seen the club-shaped gland,closely apposed to the remarkable patch of modified hypo-blastic epithelium which, in the paper by Professor Lankesterand myself (8) describing the larva, was termed the glandulartract. The nature of this organ has been enigmatical up tothe present time ; but I may as well at once call it the" endostyle," since it certainly becomes that organ.

The opening of the club-shaped gland into the cavity of themouth—i. e. the intra-buccal orifice—lies at this stage slightlydorsal, and at a later stage quite dorsal, to the endostyle.This position of the internal aperture of the gland is important,as will appear in the sequel.

LATER LARVAL DEVELOPMENT OF AMPHIOXUS. 189

On the left side (figs. 20, 21) the large lateral mouth is seenwith its margin ciliated. A ciliated groove leads from theprseoral pit to the antero-dorsal margin of the mouth. Ante-riorly the mouth ends in a sharp point; but in fig. 21, on closeinspection, a very slight protuberance can be seen in the edgeof the mouth projecting into the ciliated groove. This minuteprotuberance is the first indication of the combined forwardand transverse growth by which the mouth moves from alateral to an anterior median position.

Just below the mouth in front is the external orifice of theclub-shaped gland, and some way behind this (fig. 20) is seena small circular piece of homogeneous tissue, which is reallya differentiation in the mesoblast of the lower lip of the mouth,showing through the outer integument, and is, in fact, thefirst element of those peculiar cartilaginoid structures whichform the skeleton of the prseoral cirri. In fig. 21 there aretwo such elements, and they go on increasing in number byadditions at both ends.

Running below the mouth, and then bending up to assumea dorsal position, is a ciliated band composed of columnarhypoblastic cells, forming a slight ridge in the pharyngealwall. It is continuous in front with the base of the lower armof the endostyle. On the other side there is a similar bandrunning from the upper arm of the endostyle, but it cannot beseen in surface views at this stage.

In these, and all drawings of the left side, the anterioraperture of the nerve-tube described by Hatschek (3 and 4) isshown very clearly.

S tage II.—Figs. 3, 22, 23, and 24.

The distinction of this stage lies in the perforation of thesecondary thickenings. The second to the fifth inclusive ofthe thickenings become perforated, as a rule, before the firstand sixth; and, again, the first is usually perforated rathersooner than the sixth. This order, however, is subject tofrequent exceptions. In fig. 3, for instance, the sixth secondaryslit is open, but the first is not.

190 ARTHUR WILLEY.

The apertures of the secondary slits are at first extremelysmall, and appear as dark spots, with transmitted light, inwhich long cilia are to be seen working.

In fig. 3 it will be noticed that a seventh secondarythickening has been added behind, but it frequently does notappear before the next stage.

In the same figure there are fourteen primary slits, but thefourteenth is only indicated in side view by a median de-pression in the floor of the pharynx. This means t h a t t hesl i t is in process of c losure , as I found by occasional ven-tral views, which were obtained through the struggling of theanimal when placed between slide and cover-glass. In thisway the larva frequently got on its back, and became fixed inthat position by the slight pressure of the cover-glass.

The endostyle and club-shaped gland present the samefeatures as in the preceding stage.

The metapleura are also in much the same condition; but itmay be noticed that while in fig. 1 the right metapleur con-curs with the left about the region of the ninth and tenthprimary slits, in fig. 3 it does so in the region of the seventhprimary slit, thus indicating that the closure of the atriumhas advanced forwards.

A view of the left or oral aspect of the larva (figs. 22—24)shows that in this stage the anterior extremity of the mouthbecomes no longer pointed. This is due to a continuation ofthat hunching up of the antero-dorsal margin of the mouthinto the ciliated groove, of which we saw the first indication inStage I. I t is ultimately carried to such an extent as toentirely change the original shape of the mouth. Meanwhilethe anterior portion of the mouth sinks inwards deeper anddeeper towards the other side of the body, so that it has to beexamined with a deeper focus than in Stage I.

The upper part of the oral hood commences to form simply bythe continued growth downwards of the upper margin of the prse-oralpit and ciliated groove; while the lower part, in which alonethe buccal cirri have their origin, arises independently beneaththe under lip of the mouth. Figs. 22—24, besides showing the

Thickerperf

Nos. 1„ 1

No. 6Nos. 1,

ungs notorated.

and 6and 2

6. and 7

LAT.ER LARVAL DEVELOPMENT OP AMPHIOXDS. 191

gradual change in the shape and level of the anterior half of themouth, show also the progressive increase in the number of theelements of the buccal cirri—which, as they grow in number,grow also in size, and soon becoming irregular in their contour,show signs of growing out into tentacle-like processes.

I append a list of half a dozen observations on different larvaeat this stage, to show the nature and extent of the variations inthe condition of the secondary slits. In all cases the first slitor thickening was above and between the third and fourth,primary slits.

Number of Becondary Number of

slits or thickenings. open slits.

(i) . . 6 . . Nos. 2 to 5 .(ii) . . 6 . . „ 3 „ 6 .

(iii) . . 6 . . „ 1 „ 5 .(iv) . . 7 . . „ 2 „ 5 .(v) . . 7 . . „ 2 „ 6 . . „ 1 and 7

(vi) . . 5 . . „ 2 „ 5 . . No. 1

In the second row of the above table the numbers areinclusive.

It is worth noting here that the first primary slit is distinctlysmaller than the second and following slits. If reference bemade to the figure of the larva with three gill-slits given in thepaper previously quoted (8) it will be found that the first slitis larger if anything than those which follow it. At a certainstage its growth is arrested, and later still, in a most remarkableway, it not only becomes relatively but actually smaller in size.

Stage III.—Figs. 4 and 25.In a larva of this stage the closure of the atrium has extended

forwards so as to leave only a small portion unclosed in front, sothat in this and all subsequent stages the gill-slits are entirelyseen through the transparent wall of the atrium.

In fig. 4 there are twelve primary slits; the usual numberat this stage is, however, thirteen. The secondary slits areseven in number; the first is circular in outline, and has justopened; while the seventh is only present as a thickening notyet perforated, and is also circular. The other secondary slits

VOI. XXXII, PART II .—NEW SEE. N

192 ARTHUR WILLET.

are more or less elliptical in shape ; but it will be noticed thatthe third and fourth, which are the largest, are shaped like aplano-convex lens, their tops being flattened.

The increase in size of the secondary slits is accompanied by,and in fact is evidence of, a transverse growth by which theprimary slits are gradually taken round to the left side.

The hindermost primary slits are always bent under theventral wall of the pharynx; but in fig. 4 this bending underextends to the more anterior gill-slits, namely, up to andincluding the fourth primary slit. This should be comparedwith figs. 1, 2, and 3.

The first primary slit is now smaller than we have hithertoseen it. The twelfth is very small, but when seen in ventralview does not yet show the definite signs of closing which willbe described later.

The anterior wall of the mouth (fig. 25) has now sunk in orbent round so far that it can be easily seen through the rightbody-wall at this stage (fig. 4). This part of the mouth becomesthe right half of the oral sphincter (velum of Huxley andHatschek). As the mouth sinks towards the right side in thisway, so also does the prseoral pit; and the latter graduallybecomes flattened out as the development of the oral hood pro-ceeds, and eventually becomes simply a ciliated tract on theunder side of the oral hood, known in the adult as the " Itader-organ," which has been identified by Hatschek with the prjfioralpit of the larva.

The club-shaped gland and endostyle are approximately inthe same condition as in the preceding stages, but in somelarvae of this stage I detected signs of the backward growth ofthe endostyle, in that the club-shaped gland was frequentlyseen to overlap the posterior edge of the endostyle.

Of the six secondary slits which have been described aboveas appearing at the same time, the first, which nearly alwayslies between the third and fourth primary slits, does noteventually become the first secondary slit, but it becomes thesecond, a new one being formed in front of it. The latter,however, occasionally appears as soon as the others. Thus

LATER LARVAL DEVELOPMENT OF AMPHIOXTJS. 193

when all the secondary slits1 are established, the one that standsfirst in position is not the first formed, but arises rather laterthan the six slits which follow it. The usually late appearanceof the first secondary slit is possibly correlated in some waywith the complicated growths which are taking place in theanterior region, and is not of any special significance ; while, onthe other hand, the simultaneous appearance of the greaternumber of the secondary slits is of considerable importance, aswill appear later.

To bring the complicated nature of the transformation whichis being effected vividly before the mind, it may be noted that,while the mouth is moving forwards, the endostyle is growingbackwards, and the primary gill-slits are crossing bodily over tothe left side.

The oral aspect of a larva at this stage (fig. 25) presents noparticularly new features, but merely an extension of the pro-cesses described in the preceding stage as being in operation ;thus the anterior portion of the mouth is seen at a stilldeeper focus, the apparent length of the aperture of the mouthwhen seen from the side has decreased, and the buccalskeleton has further advanced in development.

The following table comprises a few observations chosen outfrom a number to show some of the variations that are metwith in this stage :—

ii ]

inIV

V

VI

Vll

Number ofsecondary slitsor tbickenings.

6656795

Nu

openof

slits.

Nos. 1 to 6, 1, 1 ,, 2 ,, » ,, 2 ,, 1 .

, 5. 4, 6, 6

7 and 9. 5

Thickeningsnot

perforated.

_

No. 6., 5„ 1

Nos. 1 and 7„ 1 „ 8

—

Position of first secondary slitor thickening.

Between 3rd and 4th primary.

„ 2nd and 3rd primary.

„ 4th and 5th primary.

1 The primary slits form, as has been said, to the maximum number offifteen. The secondary slits never normally exceed the number of nine. Byfar the greater number of the gill-slits of an adult Amphioxus must, there-fore, be distinguished by the name of tertiary slits. The reason of this willbecome clear as we proceed.

194 ARTHUR "WILLEY.

This table will be intelligible if one compares what has beensaid above with reference to the first secondary slit with theposition of the latter as given in the fourth column. InNo. 7 it will be noticed that the first two secondary slitswere late in appearing. No. 6 in the above table wasobviously abei-rant, but is interesting as exhibiting a hasteningof the development of the full number of secondary slits. I twas also slightly abnormal, since, although the ninth slit waswell open, the eighth was only present as a thickening.

In all cases the eventual first secondary slit arises above andbetween the second and third primary slits.

The trifling confusion which may attend this description isunavoidable. The straightforward course of events, unham-pered with variations, will be given in the summary.

S tage IV.—Pigs. 5, 6, and 26.

This is a well-marked stage, characterised by a generalincrease in size of the secondary slits, accompanied by thedipping downwards or bending inwards of the dorsal wall inthe largest of them—namely, Nos. 3 to 5 inclusive. Thefusion of the down-growth from the dorsal wall with a smallup-growth from the ventral wall of the slit (which, however,does not occur in this stage) results, as is well known fromthe figures of Kowalewsky and others, in the formation of theso-called tongue-bar of the slit.

The bending under of the primary slits from the right tothe left side has now proceeded much farther; and the firstprimary slit is very much reduced in size. In the next stagewe shall see that it closes up and ultimately disappears withoutleaving a trace.

In fig. 5 the thirteenth primary slit is seen in course ofclosure—that is to say, in a ventral view it would present theappearance shown in the case of the twelfth slit in figs. 9 and10. There are seven secondary slits, the first being betweenthe second and third primary slits.

The atrium is now completely closed anteriorly.The most striking and unforeseen characteristic of this


stage is that the patch of modified epithelium which we havealready spoken of as the endostyle has definitely commencedto push its way past the club-shaped gland, so that the latternow lies upon instead of behind it. This can also be expressedby saying that the endostyle has begun to fall away from itspreviously oblique position, so as ultimately to assume a longi-tudinal horizontal and ventral position. The incipient changeof position of the endostyle is not accompanied in this stageby any growth in its length. The movement in bulk of a patchof modified epithelium from one position to another is suffi-ciently remai'kable. It is probably, however, effected by thesame transverse growth which affects the primary gill-slits.A consideration of the figures will show that it would be quitepossible for a growth of this kind to have the effect ofdragging down the at first very oblique endostyle.

It is not so much the primary obliquity of the endostyle, asits primary anterior position in the region of the first myo-tome, which should be especially noted.

A view from the left side of a larva of this stage is given infig. 6, the object being chiefly to show to what extent theprimary slits have come round to this side. The first primaryslit is hardly recognisable from this side, but can just be seenwith a median focus. In this larva there were fourteenprimary slits; but the more usual number for this stage isperhaps thirteen, and, moreover, the thirteenth is usually incourse of closure.

The oral hood, both as to its dorsal and ventral halves, isnow well marked, and the cartilaginoid elements of the buccalcirri, of which there are six, are growing out into distinct ten-tacles, which give a crenate margin to the lower half of theoral hood. The external orifice of the club-shaped gland isundoubtedly present, but is covered over by the buccal cirri.

The apparent length of the mouth is rather greater in fig. 6than is usual at this stage, and in this respect the conditionof the mouth represented in fig. 26 is much more typical; aglance at this drawing will make it possible to understandthat, while the contour of the mouth undergoes a complete

196 AUTHOR WILLEY.

alteration in form, the actual size of the mouth does notmaterially change; though in the adult, of course, the relativesize of the mouth or velum is much less than in the larva.

The variations in the condition of the secondary slits at thisstage are not sufficiently striking to make it necessary totabulate them. Their number, as a rule, varies between sevenand eight; and almost invariably the first secondary slitappears at this stage between the second and third primaryslits (fig. 5).

Stage V.—Figs. 7—10 and 18.There is a considerable difference between the condition of

a larva at the beginning of this stage and at the end of it, aswill be seen by comparing fig. 7 with fig. 10. The constantpeculiarity for the stage is, that although the primary slitshave not quite attained their final position on the left side,yet the tongue-bars have commenced to grow down from thedorsal borders of the slits, but do not meet the ventral bordersduring this stage. There is also a feature which is moreobvious in the preceding stages, and which we see for the lasttime in the present stage. I refer to the fact that the longaxis of the secondary slits is at this and the foregoing stagesparallel to the long axis of the body, and at right angles tothe long axis of the several primary slits. This is the caseeven in fig. 10, where, however, one would probably fail to seeit at a glance. It is important in that it is characteristic ofthe earlier stages in this period of the development.

In fig. 7 a view is given of the right side of a larva whichhad just entered on this stage. There are eight secondarygill-slits, the first, as one would expect, being between thesecond and third primary slits. The tongue-bars, as we havealready seen, are in course of formation, and that of the thirdsecondary slit has actually fused with the ventral border ofthe slit. Unfortunately in the larva here figured (fig. 7) theendostyle is not typical for this stage, being in the conditiondescribed in Stage IV. A characteristic endostyle is, however,shown in fig. 18. It has become much more horizontal in


position, and has grown a long way past the club-shapedgland. The latter is still present with its large intra-buccalorifice, but when seen in the living animal it seems to presentsigns of disintegration, the constituent cells assuming aloosely aggregated and turgid appearance preparatory to dis-solution. The intra-buccal orifice is apparently the mostpersistent part of the gland. In fact, before the end of thisstage the club-shaped gland atrophies altogether. Fromobservations which I made on numerous larvae belonging tothis stage, I am inclined to believe that the cells of the glandbreak away from each other and pass into the alimentarycanal, where they are possibly absorbed.

The ciliated hyperpharyngeal band of the right side, whichhas been referred to above, can now be seen proceeding fromthe upper arm of the endostyle; while the corresponding bandof the left side proceeds from the lower arm of the endostyle.These upper and lower portions become, at a later stage,respectively the right and left halves of the endostyle; and,indeed, the latter—namely, the lower half—when looked atfrom the right side, is found to lie at a deeper focus than theright or upper half.

One of the most curious events of this stage is the closureof the first primary slit, which occurs very shortly after, if notat the same time as, the atrophy of the club-shaped gland.In fig. 7, which is a drawing of a younger larva than thoserepresented in figs. 8, 9, and 10, the first primary slit can justbe seen in side view ; on the other hand, in the larva of whicha three-quarter ventral view is given in fig. 9 it could nothave been distinguished at all in side view; but a ventral viewshowed it in the condition of a minute aperture with ciliaworking in it, surrounded by cells in such a state of aggrega-tion as to present a coarsely granular and dark appearance,which, combined with the loss of a sharp outline to its wall, isthe distinctive feature of a closing slit, and enables it to beplaced in marked contrast to a newly formed slit, of which thewall is clear and refringent.

In fig. 10 there is the merest trace of the first primary slit

198 ARTHUR WXLLEY.

just in front of the one marked " second primary slit." Infig. 8 there are indications of fourteen primary slits; thetwelfth is, however, very small, and the thirteenth and four-teenth are on the verge of closure.

In both fig. 9 and fig. 10 the twelfth primary slit is shownin course of closure.

Fig. 10 is quite an exceptional view of a larva from theventral or ventro-lateral aspect, obtained by the larva gettingfixed on its back between slide and cover-glass, as a result ofits struggles to get free. Such a complete view as is hererepresented can only be got very rarely, and when it doeshappen it lasts but a very few minutes, as the larva eitherspeedily rights itself or dies.

The shape of the primary and secondary slits has beenalready described. The tongue-bars of most of the latter havefused with the ventral borders of the slits, but not one hasdone so in the case of the primary slits. I t should be noticed,also, how the row of primary slits curves round to the middleventral line behind; the twelfth is nearly closed, the eleventhwould close later on, while the tenth might or might not closelater: judging from the shape and size of it, one may saythat the balance of probability is in favour of its not closing.In some of the primary slits there is a small up-growth fromthe ventral border of the slit, pointing towards and subse-quently destined to fuse with the down-growing tongue-bar.A.U the slits are of course seen through the transparentatrial wall. The endostyle is seen anteriorly between theprimary and secondary slits, and the buccal cirri are also wellshown.

The mouth of this stage is shown in fig. 8. The lower por-tion of the oral hood, carrying the buccal cirri, is now moreor less continuous with the upper portion into which the cirrihave not yet extended; but, at the point of junction, a ridgeis formed under which the cartilaginoid tissue eventually forcesits way (cf. figs. 12, 13). The upper and lower portions of theoral hood become respectively left and right.

Thus, while the right and left halves of the oral hood are


independent of one another in their origin, the elements of thebuccal skeleton arise entirely unilaterally from a single differen-tiation of mesoblast, which grows at both ends, and is situatedprimarily in the right or lower half alone of the oral hood, andsecondarily continues itself into the left or upper half.

The lower edge of the oral hood is prolonged anteriorly as aridge, into which the buccal skeleton is subsequently con-tinued.

With regard to the variations met with in this stage, it isonly necessary to say that the number of secondary slits variesfrom seven to nine; and as to the primary slits, it is usual tofind the twelfth, sometimes also the thirteenth, on the point ofclosing; but I have found an instance in which, although therewere nine secondary slits and the club-shaped gland wasbeginning to atrophy, yet there were fifteen primary slits, allin a healthy condition, with the exception of the first, whichwas closing.

S tage VI.—Figs. 11,13, and 19.

The constant characteristic for this stage is that the primaryand secondary gill-slits are of equal size and similar shape forthe most part j and, excluding the first two and the last two oneach side, they are approximately as broad as they are long.Further, the tongue-bars in most of them have fused with theventral borders of their respective slits.

The first primary slit has entirely gone, and it is usual tofind the eleventh primary slit, together with the twelfth andsometimes also the thirteenth (figs. 11 and 12), in a statebordering on closure. All the slits of the left side are now, asa rule, entirely confined to that Bide throughout their wholeextent (cf. fig. 8), with the exception of the ninth, whichbends under ventrally. The first slit of the left side in thisand all succeeding stages is the original second primary slit;and similarly the ninth slit of the left side is the original tenthprimary, and so on for the other primary slits.

Owing to the atrophy of the first primary slit, the first slitof the right side—i. e. the first secondary slit—comes to lie

200 ARTHUB WILLET.

opposite to the boundary between the first and second slits ofthe left side—and this is its final position.

Except when the above-mentioned niuth slit is found in anundoubtedly rudimentary condition, it is impossible to predictwhether it would eventually close or not. My observationsshow that in the majority of cases it does close, but that, onthe other hand, it frequently does not. I have never found aless number of primary than of secondary slits, so that whennine of the latter are formed (of which instances have beengiven above) it is certain that nine of the primary slits willpersist.

On comparing fig. 12 with fig. 11 it will be found that thegill-slits of the former are not quite so far advanced as those ofthe latter, while the buccal cirri in the former have reached ahigher stage of development than those of the latter. This isa slight variation.

The junction of the originally independent upper and lowerportions of the oral hood is very distinct.

The anterior or right wall of the oral sphincter or velum hasnow passed so thoroughly round to the right side that itcannot be inserted in a drawing of the left side. Four velartentacles make their appearance during this stage j they aremuch clearer, however, in the next stage.

The left ciliated (hyperpharyngeal) band is seen to join theleft or lower half of the endostyle; the right or upper half ofthe endostyle is not seen from this side naturally.

In all the drawings of larvae belonging to previous stages,from the left side, there is shown a peculiar structure, de-scribed as a nephridium by Hatschek (4), and placed justbelow and to the left side of the notochord in the regionbetween the prseoral pit and the mouth.

I t presented the appearance represented in the figures, butI could not certainly detect cilia in it, and in fact was unableto understand its import. It seems to possess a superficialresemblance to the head-kidney of Annelid larvae (trocho-spheres), but I can form no opinion as to the reality of anysuch resemblance. Hatschek discovered that it opened at one


end into the cavity of the mouth. (See also 8, pi. xxx,figs. 2—4.)

It requires, and would probably repay, further study. Fromthe present stage onwards I could not recognise this so-callednephridium in the larvae. The structure marked " x " in thefigures is probably part of the ciliated tract of the prseoral pit,seen through the oral hood.

In fig. 19 a view of the endostyle is given. It does notdiffer markedly from its condition in Stage V, extending aboutthe same distance backwards between the anterior secondaryand primary slits, but the absence of the club-shaped gland (ofwhich feeble remnants are still shown) has to be noticed.

The number of secondary slits iu this stage varies againfrom seven to nine.

S tage VII.—Pigs. 13, 13a, and 14.

The slits—primary and secondary—have now begun to elon-gate in a direction at right angles to the long axis of the body.

A view of the left side of a larva is given in fig. 13. Thefirst slit on this side (i. e. the original second primary slit) issimple and remains simple in the adult, while all the othersbecome doubled by the formation of the tongue-bars.

The ninth slit of this side is very small and lies at a ratherdeeper focus than the other slits, so that in this case it wouldin all probability close. Behind the ninth (i. e. the tenthprimary) is seen a rudimentary indication of the eleventhprimary slit.

A separate view of the velum with the four velar tentaclesis given in fig. 13a. Fresh tentacles arise between the fourprimary ones at a much later stage, when the gill-slits areabout three times as long as they are now, and the endostyletwice as long. The final number of velar tentacles is twelve.

The right wall of the velum is not quite opposite to the leftwall even yet: it is slightly in front of the left wall, and theextent to which it is so corresponds to what has been hithertospoken of as the apparent length of the aperture of the mouth.

On the right side (fig. 14) it is usual to find eight slits. The

202 ARTHUR WILLEY.

ciliated tract of the prseoral pit is now beginning to alter itsshape—becoming constricted in the middle. This constrictionis carried still further in the next stage (fig. 15), and, as isalready known, the ciliated tract subsequently grows out intoseveral lobes, which give its characteristic appearance to theRader-organ.

The most important feature in fig. 14 is the endostyle, whichhas increased considerably in length, reaching to the fourthslit. Its double nature is very apparent; the left (lower) halflying, as before mentioned, at a much deeper focus than theright (upper) half, so that the former only is seen from theleft side (fig. 13).

In the larva represented in fig. 14 there were eight slits onthe left side, and the ninth was seen to be in course of closurein the mid-ventral line.

We are now approaching the period in the larval develop-ment in which there are an equal number of slits on both sidesof the body, all those on the left side being primary and allthose on the right side secondary.

After a long interval, during which the gill-slits increase invertical height and the endostyle extends further backwards,fresh t e r t i a r y gill-slits form behind on each side in a mannercharacteristic of later growth, and continue to do so through-out life.

The stage, then, with an equal number—seven to nine pairs—of primary and secondary slits, and before the formation ofany tertiary slits, may well be called " t h e C r i t i c a l S tage . " 1

As a general rule the critical stage is characterised by thepresence of eight pairs of slits, namely, eight primary slits onthe left side and eight secondary slits on the right.

;e VIII .—Figs . 15—17.During this stage the critical stage, as defined above, is

firmly established, and in fact may be said to commence as soon1 The remarkable fact that the number of pairs of gill-slits in the critical

stage agrees approximately with the full typical number of the craniate gill-clefts should not be lost sight of.


as the fate of the tenth primary slit has been decided. Infig. 16 the ninth slit has not perfectly assumed its positionon the left side, and so its permanence cannot be consideredas being beyond question, although its size and appear-ance would lead one to think that it would persist and notclose.

The length or vertical height of the gill-slits has furtherincreased, and has been taken as the distinguishing feature ofthis stage.

Fig. 17 represents the last (seventh) pair of slits of a larva,in which not only had the tenth primary slit closed, but theninth was also on the point of closing. The seventh slit of theright side is seen with a surface focus, the ninth rudimentaryprimary slit is seen with a median focus, and the seventh slitof the left side, which is doubling in, is seen with a deep focus.

The mouth or velum (fig. 16) is now anterior and median inposition; the right and left walk of it are quite opposite to oneanother, and thus the apparent length of the mouth in sideview is now reduced to zero.

The oral hood and buccal cirri have not yet, however, attainedtheir final condition.

The endostyle (fig. 16) has greatly increased in length,extending to the fifth gill-slit on this side.

The right side of another larva of this stage is shown infig. 15, chiefly for the sake of the endostyle. This is the laststage at which its double origin can be recognised. The righthalf of the endostyle is now seen to lie at the side of, and notabove, the left half.

The ciliated hyperpharyngeal band appears to die out behindthe pharynx. On this side there are eight (secondary) slits:on the left side of this larva there were also eight (primary)slits, and no indication of any more, so that it is a typicalexample of a larva which has entered upon the criticalperiod.

It is important to emphasise the fact that the critical period,in which the larva has usually eight pairs of slits, has a con-siderable duration, in the course of which the only changes

204 ABTHOU WILLEY.

that take place are the completion of the oral hood and theincrease in the size of the gill-slits and endostyle, but no freshslits are formed.

The close of the critical period may, therefore, be taken as thecloae of the entire larval phase of development. Up to thispoint all the gill-slits of the left side are primary—that is tosay, they appeared at first as a single series on the actual rightside of the larva, and were pushed across to the left side by thedeveloping secondary slits. We shall find reason to supposethat the actual right side of the larva is not precisely identical•with the morphological right side.

Summary of t he Stages .

In the following summary it will be found desirable to givea few measurements, which, however, have only an averagevalue, the unit employed being a hundredth of a millimetre.It is also intended, for the most part, to ignore the variationswhich have been detailed above, and to state only whatappears to be the usual condition of things, as inferred fromobservations on numerous larvae.

Stage T. R i g h t Side.—Fourteen primary slits, noneclosing; six secondary thickenings above the primary slits ;endostyle in front of club-shaped gland, and therefore in frontof all the gill-slits ; atrium widely open anteriorly.

Left Side.—Liarge lateral mouth; apparent length (in thiscase the actual length), 45 units; 1 to 2 elements of buccalskeleton.

S tage I I . R igh t Side.—Fourteen primary slits, four-teenth closing ; secondary thickenings perforated; endostyleunaltered ; atrium somewhat less widely open anteriorly.

Left Side.—Lateral mouth beginning to bend round infront; apparent length, 36; 3 to 5 elements of buccal skeleton.

Stage I I I . R igh t Side.—Thirteen primary slits; se-condary slits simple ; endostyle unaltered ; atrium slightly openanteriorly.

Left Side.—Mouth still further bent round in front;apparent length ranges from 28 to 12, average about 25 ; 5 to 6

LATER LARVAL DEVELOPMENT OT AMPHIOXUS. 205

elements of buccal skeleton commencing to grow out astentacles or cirri.

Stage IV. Right Side.—Thirteen primary slits, all bend-ing under the pharynx, first very small and thirteenth closing;seven secondary slits, the new one appearing in front; thelarger secondary slits commencing to double in, i.e. to formtongue-bars; endostyle extending a short way beyond theclub-shaped gland; length of endostyle, 24; atrium closedanteriorly.

Left Side.—Primary slits are now partly on this side;average apparent length of mouth, 17 ; oral hood containing inits lower half incipient cirri.

Stage V. Right Side.—Twelve primary slits just visibleat base of pharynx, twelfth closing; the first primary slitundergoes atrophy ; eight secondary slits with complete tongue-bars in the larger ones; endostyle extends a long way past theclub-shaped gland; length of endostyle, 28; club-shaped glandundergoes atrophy.

Left Side.—Primary slits not fully arrived on this sideyet; tongue-bars commencing in the primary slits; averageapparent length of mouth, I I ; junction of upper and lower por-tions of oral hood.

Stage VI. Right Side—Eight secondary slits, lengthequal to the breadth, measuring about 11 units, both in a direc-tion parallel to long axis of body and in a direction at rightangles to it; length of endostyle, about 34.

Left Side.—Ten primary slits (i.e. Nos. 2 to 11 inclusive),the eleventh closing ; tongue-bars complete in the larger slits,which are equal in size to the corresponding secondary slits;mouth same as in Stage V; buccal skeleton commencing topenetrate the upper portion of the oral hood.

Stage VII.—Slits on both sides commencing to elongatein a vertical direction; average length of the central slits, 14.

Right Side.—Eight slits; length of endostyle, 39.Left Side.—Nine slits, ninth (i.e. tenth primary) usually

closing; apparent length of mouth or velum, 4; four velartentacles.

206 ARTHUR WILLBT.

Stage VIII.—Average length of central slits, 21.Right Side.—Eight slits; length of endostyle, 45. The

two halves of the endostyle are now definitely right and left.Left Side.—Eight slits ; apparent length of velum, 0.Of the above stages, the fifth is certainly the most eventful.

With regard to the last stage, it is well to be reminded that itinaugurates the critical period; that at the commencement ofit there is often a rudimentary or doubtful primary slit behind ;and, finally, that the presence of eight pairs of slits is not auniversal rule, as there are sometimes only seven and some-times nine pairs, but never normally more than nine.

The central feature, perhaps, of the foregoing observationsis the development of the endostyle; next in importance comesthe atrophy of the club-shaped gland and the first primarygill-slit during Stage V, and of almost equal importance andnovelty is the atrophy of from four to six of the hinder slits ofthe primary series.

It goes without saying that there is no hard and fast linebetween any two consecutive stages. The division into stagesis entirely for convenience, all the processes of developmentand transformation being gradual.

SUMMARY OF THE HISTORY OF THE INDIVIDUAL STRUCTURES.

1. The Gill-slits.The primitive or ancestral left row of gill-slits has, by a

process which will be discussed later, been made to assume aposition on the right side of the larva, where slits form to thenumber usually of fourteen, but occasionally fifteen.

The actual perforation in the case of each slit occurs at firstat the base of the pharynx, near the mid-ventral line, and thenextends npwards.on the right side; the hindermost slits, how-ever, retain their ventral position all through.

The slits which thus appear at first in a single unpairedseries are called the primary slits.

The right row of gill-slits, which originally corresponded tothe primary slits before the latter had forsaken their primitive


situation on the left side, has suffered retardation of develop-ment. When these retarded slits do appear, they are called,arbitrarily, the secondary slits.

The secondary slits form to the number usually of eight(sometimes seven and sometimes nine), above and alternatingwith the primary slits on the right side, appearing long beforethe completion of the atrium in front. The first secondaryslit always lies between the second and third primary slits, andusually arises somewhat later than the succeeding six secon-dary slits, which develop simultaneously. I t will be shownlater that the simultaneous appearance of the secondary slits isonly due to a masking of the metameric mode of formation.

As the secondary slits grow in size the primary slits gradu-ally pass round to the left side, and meanwhile certain of themcommence to atrophy. It is especially noteworthy that afterStage I the first primary slit becomes progressively smaller andsmaller, until by the end of Stage V it has disappeared alto-gether. The view will be supported below that the closing ofthe first primary slit has a profounder significance than theclosing of the hinder slits, or at any rate a different meaning.

By the closing of some of the primary slits their number isreduced to the same figure as that of the secondary slits,whether it be seven, eight, or nine. The primary slits whichatrophy are, therefore, the following :

a. When 7 primary slits persist—Nos. 1 and 9, 10, 11,12, 13, 14 close./3. „ 8 „ „ „ 1 „ 10, 11, 12, 13, 14 „y. „ 9 „ ., .. 1 ,. 11, 12, 13,14 „

It is during the passage of the primary slits from the rightto the left side that the completion of the atrium is effected.

The long axis of the secondary slits is at first parallel to thelong axis of the body, while at the same time that of theprimary slits is at right angles to the long axis of the body.The axes are equalised as soon as the two series of slits areproperly adjusted on their respective sides, and then the slitson each side commence to elongate in a direction at rightangles to the long axis of the body.

VOL. XXXII, PART II.—NEW SEE. O

208 ARTHUR WILLET.

The formation of the tongue-bars commences very early inthe secondary slits, but not in the primary slits until they arewell round to the left side. The first slit on each side remainssimple, and does not form a tongue-bar.

As soon as all the primary slits that are going to closehave done so, the larva enters upon the so-called criticalperiod. At the end of this period fresh slits begin to formnormally on both sides, and continue to do so throughout life;these are called the tertiary slits in order to distinguish them.

Thus the critical period may be defined as the period inwhich the slits are arranged regularly—an equal number ofpairs—in two rows, right and left j but all those on the leftside are primary (in the sense in which the term is hereemployed), and all those on the right side are secondary.

2. TheEndos ty le .

The forecast of the endostyle appears in the embryo at avery early stage, namely, shortly after the formation of theclub-shaped gland in front of which it lies. It consists of apatch of modified columnar hypoblastic epithelium, bendingobliquely backwards and folded forwards upon itself, andlying on the right side of the mouth-cavity in the region ofthe first myotome.

The upper arm of the endostyle is at first much shorter thanthe lower arm; the former becomes the right half of the adultendostyle, and the latter becomes the left half.

The endostyle retains its original shape and its anterior posi-tion until some time after the first appearance of the secondaryslits.

At Stage IV it begins to fall away, as it were, from itsoblique position—the evidence of this being found in the factthat in this stage it extends a short distance beyond the club-shaped gland, so that the latter, instead of lying behind it, liesupon it (fig. 5). Although the endostyle has thus grownbackwards, it ha3 done so in bulk—that is to say, it has bodilyshifted its position without increasing in length.

In Stage V, in addition to a further change of position, it


has commenced to grow in length (fig. 18), and now extendsbetween two or three of the anterior secondary and primaryslits.

In Stage VII (fig. 14) it has greatly increased in length, andis quite horizontal and ventral in position; but the upper andlower (i. e. right and left) halves are still perfectly distin-guishable.

In Stage VIII the two halves of the endostyle, which havehitherto been in the relation to one another of upper and lower—though the lower has been in the later stages at a deeperfocus than the upper—come to lie alongside of one another,and so become definitely right and left respectively.

In fig. 16 the two halves can just be recognised in frontfor the last time.

From the anterior extremity of each half of the endostyle aciliated ridge composed of the pharyngeal epithelium passesdorsalwards, and becomes on each side the hyperpharyngealband.

The anterior portion of the ciliated bands, where they curvedorsalwards from the endostyle, is undoubtedly to be comparedwith the pericoronal ridges of Ascidians which pass from thefront end of the endostyle round the mouth to the dorsal lamina.

The following table shows the proportionate average increasein the length of the endostyle, together with the concomitantgrowth in length or vertical height of the gill-slits during thecritical period. The unit of measurement, as before is -j-jj-g-millimetre.

Length of central slits. Length of endostyle.

1 . . . 27 units . . . 52 units.2 . . . 32 56 „3 . . . 35 „ . . . 66 „4 . . . 41 „ . . . 76 „

The endostyle does not reach the posterior end of the pharynxbefore the close of the critical period.

3. The Club-shaped Gland.The development of this gland was made out by Hatschek (3).

To his description of it has been added the discovery of the

210 ARTHUR WILLET.

intra-buccal orifice (8), and in the present paper the accountof its atrophy. It arises in an embryo with nine to ten pairsof rayocoelomic pouches, as a ventral transverse fold of the ali-mentary canal at a point intermediate between the first andsecond myotomes.

The fold extends up to the right side on the one hand, andventralwards round to the left side on the other. It thenbecomes separated throughout its whole length from thealimentary canal, and develops an opening to the exterioron the left side, just below the lower margin of the mouth.Its intra-buccal orifice does not appear until a rather laterperiod.

In striking contrast to all the other organs which adjoin it,the club-shaped gland does not undergo any change of positionduring what has been elsewhere called the process of symmetri-sation (cf. figs. 4, 5, and 18).

During Stage V the club-shaped gland commences to atrophy,and towards the end of that stage it disappears altogether, theintra-buccal orifice being the last to go.

Considering their close proximity to one another, it wouldbe natural to suspect that intimate relations existed betweenthe club-shaped gland and the endostyle.

I made a few observations which seemed to show that theendostylar epithelium, at least before it has assumed its ventralposition, is extremely delicate and liable to disrupture.

When a larva had been confined for a very short time betweenslide and cover-glass, the epithelium of the endostyle wouldcommence to give off1 numerous small round homogeneousciliated bodies, which passed into the alimentary canal, and werethere apparently absorbed.

This happened sometimes so soon after the larva had beenplaced upon a slide that I did not recognise it at first as apathological phenomenon, as of course it was.

The fact that the endostyle is such a highly specialised anddelicate (while young) structure suggests the idea that possiblya collateral organ like the club-shaped gland (concerning themorphology of which see below) might function as a sort of

LATER LARVAL DEVELOPMENT OP AHPHIOXCJS. 211

special outlet for any nauseous substances entering the mouth,which might otherwise act injuriously on the endostyle. Butthe histological structure of the cells and the physiologicalproperties of the gland require further study before any safeconclusion can be arrived at as to its function. The cells ofthe gland are ciliated, and the external orifice is surrounded bycilia.

4. The Mou th or Velum.

The mouth arises in an embryo of about thirty-two hours asa minute circular aperture on the left side in the region of thefirst myotome (see Hatschek, loc. cit.). It then grows rapidlyin size, and becomes a large lateral lens-shaped orifice, moreor less pointed at both ends. Prom Stage I onwards it beginsto grow gradually round the anterior extremity of the alimentarycanal, the change in position being accompanied by great alte-ration in shape, consisting chiefly in the gradual equalisationof its diameters, which are at first remarkably unequal, thelonger diameter being four or five times the length of theshorter one. The bending round of the mouth commences bya curious hunching up of the antero-dorsal margin of the mouthinto the ciliated groove which runs from the prseoral pit to thedorsal edge of the mouth.

By this means the previously pointed anterior extremity ofthe mouth becomes lost, and this part of the mouth then pre-sents a truncated appearance. Meanwhile the anterior half ofthe mouth begins to bend inwards towards the right side, and,in fact, it eventually becomes the right half of the velum. Thegradual attainment by the mouth of an anterior median andvertical position is fully illustrated in the plates at the end ofthis paper. The mouth is relatively much smaller in the adultthan in the larva, but not actually so.

The edge of the larval mouth is provided with two kinds ofcilia, namely, (i) a set of evenly distributed small cilia, and (ii)at a rather deeper focus much longer cilia, which hang togetherin groups at equal intervals. At the anterior extremity of themouth in the earlier stages there is a group of very long cilia,

2 i 2 ARTHUR WILLEY.

seen with quite a surface focus; and the whole external surfaceof the lower wall of the mouth is beset with small cilia (figs.21 and 23).

Four velar tentacles are present in Stage VII, and thisnumber continues all through the critical period. Eventuallyfresh tentacles arise and make up the number which occurs inthe adult, viz. twelve.

5. Ora l Hood and Buccal Ci r r i .

The right and left (lower and upper) halves of the oral hooddevelop independently of one another, and are also entirelyindependent of the metapleural or atrial folds, as will be readilygathered from the figures. It will be seen that the oral hoodis formed mainly after the completion of the atrium. ProfessorLankester has suggested (5) that " the oral hood is the prseoralportion of the epipleural folds." This, of course, is now seennot to be the case.

The oral hood develops concurrently with the change ofposition of the mouth.

The upper portion of the oral hood arises between Stage I Iand Stage I I I as a fold of the integument overhanging theprseoral pit and ciliated groove. Beyond the latter it mergesinto the upper mai'gin of the mouth. This can be seen insection by referring to the paper which has been alreadyquoted several times (8, pi. xxxi, figs. 13 and 14a). Asthe fold becomes larger it extends to the posterior extremityof the mouth, where it is met by the lower fold. Anteriorlythe upper fold of the oral hood is continued somewhat beyondthe limit of the prseoral pit, and ends as a small ridge on theleft side of the head region, just below the level of thenotochord.

The formation of the buccal cirri is coincident with thedevelopment of the lower portion of the oral hood. In theadult the right and left halves of the oral hood are equally pro-vided with cirri. It is, therefore, a most curious fact that thebuccal cirri take their prime origin in the lower or right halfonly of the oral hood, and, after reaching a considerable stage

LATER LAEVAL DEVELOPMENT OF AMPHIOXUS. 213

of development while still confined to the lower portion, theythen grow into the upper portion also.

Thus the oral hood itself has a double origin ; it may besaid to arise from two independent e"bauches (to employ aFrench term which has not, when used in this connection, asatisfactory English equivalent, " forecast" being the nearestapproach), while the system of the buccal cirri arises from asingle ebauche .

The formation of the cartilaginoid elements which form theskeletal or supporting tissue of the buccal cirri, and which,taken together, may be referred to as the buccal skeleton,commences in Stage I. In this stage we find first one andthen two minute circular and homogeneous diflFerentiations inthe mesoblast of the lower lip of the mouth, which can be seenthrough the outer integument. They appear some distancebehind the external orifice of the club-shaped gland, just belowthe mouth. They commence to form before the mouth hasbegun to turn round in front.

In the succeeding stages fresh elements are added at bothends—anterior and posterior—so that the central elements arealways the oldest; and they soon begin to get irregular in out-line, indicating a tendency to send out processes. Eventuallyeach piece gives rise to one process.

As the differentiation of the elements of the future buccalskeleton proceeds the superjacent integument becomes raisedup into a fold, which is the lower or right fold of the oral hood.About Stage IV the above-mentioned processes, which growout from the separate elements, attain such a size that the in-tegumentary fold over them is thrown into a series of pleats(fig. 6, &c). By an extension of this pleating the buccal cirriof the adult are formed.

From Stage V onwards the lower fold of the oral hood meetsthe upper fold, and the point of junction is very plainly markedby a ridge, under which the buccal skeleton penetrates intothe upper fold (figs. 8, 12, 13, and 16).

The lower fold is continued anteriorly some distance beyondthe buccal skeleton as a ridge. Eventually the cirri extend

214 ARTHUR WILLEY.

along the whole length of this ridge, as they do also along theupper fold.

It has been stated above more than once that the lower foldof the oral hood in the larva becomes the right half of theoral hood in the adult; and this half is, as Professor Lankesterhas pointed out (5), continuous in front round the extremepoint of the body with the dorsal fin.

A little reflection will show that this is the condition thatmight be expected from the development of the lower fold, asshown in the drawings accompanying this paper.

It is important to note that the buccal skeleton grows ateach end only, and that fresh elements are not formed inter-stitially. In the adult the median cirri are smaller than theothers; and one would at first naturally suppose that thesewere the youngest, and that this was the point at which freshcirri would be formed; as a matter of fact, however, the smallsize of the median ventral cirri of the adult is deceptive, forthey are the oldest cirri, and new ones are only added at thefree extremities, right and left, of the buccal skeleton.

GENERAL CONSIDERATIONS.

It is my intention to confine myself mainly to the attemptto give some explanation (or at least suggest one) of the threemost prominent features in the larva of Amphioxus, namely :

1. The asymmetry.2. The endostyle.3. The club-shaped gland.

The Asymmetry of the Larva.

"With regard, then, first to the asymmetry of the larva, wemust begin by assuming—and the assumption will be sup-ported later on—that in the primitive type from which theCephalochorda diverged the notochord did not extend to theanterior end of the body, but, on the contrary, that the forwardextension of the notochord in Amphioxus is secondary andadaptive; and further, that the primitive position for the


mouth was dorsal, as it is in the Ascidian tadpole.1 The last as-sumption may appear to be the most arbitrary, but it is justifiedby the close affinities which exist between Amphioxus and theAscidians, and which will be referred to again. It may as wellbe stated at once that the view I am about to put forwarddepends essentially on this way of regarding the forward exten-sion of the notochord; and the immediate reasons for supposingit to be secondary are—

i. That it does not extend to the front end of the body at itsfirst origin (see Woodcut, fig. 2).

ii. That its forward extension is an obvious advantage toAmphioxus when burrowing in the sand, which is its constantoccupation after a certain period of development.

iii. The analogy of the Ascidian tadpole.It is clear, then, that as the notochord pushed itself forwards

beyond the anterior limit of the nerve-tube, the mouth wouldbe compelled to forsake its primitive dorsal position—indeed,it would be forced, as it were, to one side, just as the anterioropening of the nerve-tube, which is at first actually dorsal, ismade to assume alateral position through the development of thedorsal fin ; or, again, as the anus is similarly displaced by thecaudal fin. It is a singular coincidence that the anterior neuralpore, the mouth, and the anus are all deflected to the left side.

The lateral position of the mouth is thus supposed to be dueto—or to occur in correlation with—the forward extension ofthe notochord ; and it is evident that the same process whichcaused the mouth to rotate from a dorsal position to its pre-sent position on the left side in the larva would also byimplication cause structures which were originally on the leftside, or ventral, to pass across to the right side. There canhardly be any doubt that a semi-rotation of the ancestralpharynx has virtually taken place; the only question is,whether we are right in correlating it with the forward exten-sion of the notochord. Supposing the ancestral mouth to havebeen small and circular, as it is in the Ascidian tadpole and inthe very young larva of Amphioxus, it is possible that this

1 See Balfour, ' Comparative Embryology,' vol. ii.

216 ARTHUR WILLEY.

rotation (which, as above indicated, need only be a virtualphenomenon) may only have affected at first the most anteriorportion of the pharynx, namely, the portion containing theendostyle, club-shaped gland, and first gill-slit.

In the ventral view shown in fig. 10 it will be noticed howthe primary slits curve round posteriorly to the ventral middleline. It may be that formerly this tendency of the hinderslits to lie in the direction of the left side, though now onlyreaching the middle line, was more accentuated, and extendedto the more anterior slits. The position of the primary slitswhich succeed the first on the right side has thus to be accountedfor, and this is done by assigning the cause of it to the furtheradaptation which led to the huge size of the lateral mouth.

Having got on to the left side in the way above described,it was found advantageous to have as large a mouth as circum-stances would permit, and accordingly the capacious mouth ofthe larva was established by the action of natural selection.A glance at the figures should carry with it the conviction thatwhere the mouth is, there the gill-slits could not be.

Consequently the gill-slits which succeed the first develop,like it, on the right side, though, as we have seen, behind theregion of the mouth they continue to retain a ventral position.

The lateral position of the mouth, then, is correlated with theforward extension of the notochord, and the unilateral positionof the primary gill-slits with the lateral position of the mouth.

The rotation of the original left side gill-slits on to the rightside has led, as might have been expected, to the temporaryobliteration of the slits which properly belong to the rightside, but it is not so much an obliteration as a retardation indevelopment that has been effected. Thus, in a larva withapparently only a single series of gill-slits, there is potentiallya double series j but, owing to the force of conditions whichhave been secondarily induced, one of the two rows is some-what late in putting itself in evidence.

That we have to do here with a retardation in the develop-ment of a whole series of gill-slits is certainly proved by thefact that when the secondary slits do appear, they form mainly

LATER LARVAL DEVELOPMENT OF AMPHIOXTIS. 21*7

not one after the other, as is usual with metamerised organs,but simultaneously. The successive or metamerie mode offormation, in fact, obtains theoretically with the secondaryslits of the larva of Amphioxus, only in this case it is l a ten t ,•with the natural result that the slits appear simultaneously.The usually but not invariably late appearance of the firstsecondary slit is due to a further slight retardation, and isprobably of no great significance.

The morphological mid-ventral line of the larva is indicatedby the blood-vessel which lies dorsal to the primary gill-slits(fig. 1, &c).

While the lateral mouth is serviceable to the pelagic larva,it would plainly be awkward to the animal during its career asa dweller in the sand—hence the necessity for the transforma-tion which has been described above.

I have now given a possible explanation of the asymmetryof the larva of Amphioxus, namely, that it can be traced ulti-mately to the adaptive forward extension of the notochord. Itis thus a purely ontogenetic phenomenon, and is not anancestral character. It is only a temporarily superinducedcondition, and is destined to give place after a brief existence tothe inherent but latent bilateral symmetry of the organism.

AMPHIOXUS AND THE ASCIDIANS.

To pave the way for considering the significance of the endo-style and club-shaped gland of the larva of Amphioxus, wemust institute a comparison between the embryo and larva ofan Ascidian and of Amphioxus. The remarkable researchesof Professors Ed. van Beneden and Charles Julin on themorphology of the Tunicates (6) have rendered it possible todo this the more effectually.

The peculiarities of an Ascidian embryo as described by theabove-named authors are as follows (see Woodcut, fig. 1):

The enteric cavity at first of two portions, viz. a prse-chordal and a subchordal portion; the greater part of the lumenof the latter portion disappears at a very early period—it hasbeen identified with the post-anal gut of higher Vertebrates

218 ARTHUR WILLEY.

(see Balfour,' Comp. Emb./ vol. ii), into which the neurentericcanal opens; but there is every probability that there is nomore post-anal gut in the Ascidian embryo than there is inthe embryo of Amphioxus.

The mesoblast consists of two longitudinal bands differen-tiated from the primitive hypoblast.

These bands were formerly thought to be quite solid, but vanBeneden discovered (loc. cit.) that anteriorly they occur as a pairof archenteric pouches (see Woodcut, fig. 1). The mesoblasticbands in the region of the so-called tail are solidj and consist ofonly a single layer of cells continuous ventrally with the remainsof the hypoblast; but in front they consist of several layers ofcells, which at first surround a short lumen which communi-cates, as stated above, with the enteric cavity on each side.

This single anterior pair of openings into the alimentarycanal is the only indication in the Ascidians of archentericpouches, and it is regarded by the authors as equivalent to thefirst pair of somites of Amphioxus. The rest of the mesoblastin Ascidians is unsegmented, and the general absence of meso-blastic pouches, with the exception of the first pair, is con-nected with the early atrophy of the posterior or subchordalportion of the alimentary canal, and is therefore probablya product of degeneration, and not an ancestral character.

In the embryo of Amphioxus, at a roughly correspondingstage (Woodcut, fig. 2) the notochord does not reach theanterior end of the body, but even at this early period it extendsbeyond the anterior opening of the nerve-tube (cf. Woodcut,fig. 1). Partly in front of the notochord and partly below itis a portion of the alimentary canal which lies in front of allthe mesoblastic somites. This anterior chamber consists of amedian portion and two lateral horns. The latter become con-stricted off from the chamber, and form the anterior intestinaldiverticula, whose entire development has been fully describedby Hatschek (3); while the median portion of the chambersimply merges with the anterior extremity of the alimentarycanal, and ceases to be recognisable as a distinct chamber. Asis well known, the left anterior diverticulum opens to the exterior


as the praeoral pit, while the right one becomes the expandedhead-cavity of the proboscis-like end of the body.

The anterior chamber with its two lateral horns, together

FIG. 1. FIG. 2.

EXPLANATION OF THE WOODCUTS.

FIG. 1.—Diagrammatic dorsal view of an embryo of Clavelina Rissoana(modified after van Beneden and Julin). The mesoblast is representedin an extremely diagrammatic way. It consists really of large polyhedralcells.

a. p. Anterior neural pore. n. t. Nerve-tube, lying on the bypoblastin front and on the notochord behind, p. v. Prrechordal vesicle. /. d.Lateral diverticulum (paired) of prrechordal vesicle, mes. p. Mesoblastioor archenteric pouch : only one pair, equivalent to the first pair of pouchesin Amphioxus. mes. Mesoblast. nch. Notochord.

FIG. 2.—Dorsal view of an embryo of Amphioxus of a correspondingage (after Hatschek).

Letters as above. The nerve-tube lies entirely on the notochord.There are numerous pairs of mesoblastic pouches.

220 ARTHUR WILLBT.

with the region of the first myotome, is held by van Benedenand Julin to be homologous with the prsechordal portion of thealimentary canal in the Ascidian tadpole. In the latter case,also, two horn-like outgrowths are developed from this part ofthe enteric cavity. They are subsequently met by two inva-ginations—right and left—of the epidermis, with which theyfuse to form the so-called primary branchial canals. Theybecome the right and left peribranchial cavities, and consistpartly of epiblast and partly of hypoblast, and are consideredto be homologous with the right and left anterior intestinaldiverticula of Amphioxus, in which, however, the epiblasticelement is wanting.

The peribranchial cavities then expand, and, growing rounddorsally, fuse together to form the atrium. Meanwhile thecharacteristic gill-slits or stigmata have commenced to form,leading from the branchial to the peribranchial cavity, andfinally the left portion of the latter cavity alone communicatesto the exterior by the atriopore, the orifice of invaginatiou ofthe right portion of the cavity having been lost. It is statedthat probably the original communication of the primarybranchial canals with the enteric cavity is lost, and that all thestigmata proper are secondary.

Thusthe visceral wall of the Ascidian atriumis derived mainlyfrom the hypoblast, while the peripheral wall is formed from theepiblast, and therefore the part of the atrium which leads tothe exterior on the left side by means of the atriopore consistsof epiblast; so that the praeoral pit of Amphioxus, which iscomposed entirely of hypoblast, does not exactly correspond tothe atrium of Ascidians, but is represented in the lattergroup by the visceral wall of the atrium, while the actualopening to the exterior of the prseoral pit corresponds to thejunction of the hypoblastic and epiblastic elements in theAscidian atrium.

Neglecting their subsequent modifications, and supposingthe epiblastic involutions to have been suppressed or reducedto zero in Amphioxus, the anterior intestinal diverticula ofthe embryo of the latter do actually coincide in every essential

LATER LABVAL DEVELOPMENT OF AMPHIOXUS. 221

respect with the primary paired atrial cavities, or, as they havebeen called, branchial canals of the Ascidian embryo ; and thefinal opening to the exterior in both cases is connected withthe original left diverticulum.

The two embryos represented in the woodcuts are thusregarded as practically identical with one another in all mor-phological respects. There is, in fact, no positive differencebetween them except in the relative proportion of parts of thebody; the general absence of mesoblastic somites in theAscidian being, as before stated, secondary and correlatedwith the atrophy of the subchordal portion of the archen-teron.

It will possibly have been noticed that the forward exten-sion of the notochord in Amphioxus commences at a remark-ably early stage, months before the larva takes to the sand.This can be accounted for as a hastening of the development,a familiar phenomenon.

The divergence, then, between the two embryonic typesconsists in the great development of the prsechordal region inthe Ascidian embryo, and in the reduction of the same regionin the embryo of Amphioxus.

It will have been inferred from what has been said that thetail of the Ascidian tadpole is regarded as the equivalent of thetrunk of Amphioxus.

In the Ascidian the expanded portion of the prsechordalregion after giving rise to the peribranchial diverticula, becomesthe branchial sac or pharynx, the floor of which is specialisedas a glandular organ known as the endostyle.

The narrow portion of the prsechordal vesicle, which alsoextends a very short distance beneath the anterior extremityof the notochord, gives rise to the oesophagus and stomach,while the intestine arises as a lateral outgrowth from the latter;but the extraordinary feature of it is that the intestine growsout from the stomach to the r i g h t of the mid-ventral line,and then passes across to open into the cloacal vesicle on theleft side.

There can be little doubt that the ancestor of the Ascidians,

222 ARTHUR WILLEY.

before the atrophy of the subchordal portion of the intestine,had an anus at the posterior extremity, as is the case withAmphioxus. This primitive outlet has, therefore, been sup-planted by the development of the outgrowth from the stomachmentioned above, which van Beneden and Julin call a collateraldiverticulum, and at the same time make the suggestion,which my observations tend to support, that the intestine ofthe Ascidian, arising as above described, is homologous withthe club-shaped gland of Amphioxus.

THE ENDOSTYLE AND CLUB-SHAPED GLAND OF AMPHIOXUS.

We must now define the position of the endostyle in thelarva of Amphioxus in order to grasp the situation of the club-shaped gland.

It has been already implied that in all probability theancestor of Amphioxus was provided with a prsechordal regionand was symmetrical; and, in fact, resembled an Ascidian tad-pole before the atrophy of the subchordal portion of theintestine.

In the larva of Amphioxus the endostyle lies at the extremefront end of the alimentary canal, which, as we have saidabove, represents the median portion of the anterior or ' prse-chordal' chamber, the lateral portions of which formed theanterior intestinal diverticula. It is further placed in theregion of the first myotome; its peculiar oblique and lateralposition has been already dwelt upon and explained. It alsolies immediately in front of the club-shaped gland.

It results, first, from the analogy of the Ascidian tadpole,and secondly, from the position of the endostyle in the larva ofAmphioxus, that it (the endostyle) was originally differentiatedas a mid-ventral groove in the prsechordal region. This primi-tive anterior position for the endostyle had been previouslysurmised by van Beneden and Julin, so that my account ofits development in Amphioxus harmonises well with their con-clusions .

The transverse growth which affected the anterior region


of the alimentary canal of Amphioxus at the time of theforward extension of the notochord naturally rotated theoriginally median ventral endostyle on to the right side incompany with the collateral organ, now known as the club-shaped gland.

A comparison of the figures accompanying this paper, illus-trating the development of the endostyle, shows clearly andbeyond question that everything dorsal to the endostyle is onthe right side of the morphological middle line, while every-thing below it is on the morphological left side. Thus theintra-buccal orifice of the club-shaped gland is on the rightside, and the first and succeeding primary slits are on the leftside—morphologically. At this juncture I want especially toemphasise the relative positions of the intra-buccal orifice ofthe gland and of the first primary slit.

The oesophagus and stomach of the Ascidian, which arise,ontogenetically, from the praechordal vesicle, are practically un-represented in Amphioxus—the part of the vesicle from whichthey arise being suppressed. In fact, in their respectivecapacities as oesophagus and stomach they are not eventheoretically present. Bearing this in mind, we can now pro-ceed to compare the position of the club-shaped gland and itsapertures with the position of the intestine in the youngAscidian.

In the larva of Amphioxus the club-shaped gland is situatedbehind the endostyle, i.e. at the posterior extremity of theprimitive prsechordal vesicle—in this respect agreeing with theintestine of an Ascidian. In both cases the aperture of thecollateral diverticulum (whether gland or intestine) into thealimentary canal is immediately to the right of the mid-ventralline—in the one case actually, and in the other case morpho-logically ; and in both cases the external aperture is on theleft side. The point of origin of the club-shaped gland betweenthe first and second myotomes also corresponds more or lessclosely with that of the Ascidian intestine. There is, therefore,a fairly strong case in favour of the homology of the two struc-tures. The question still remains as to whether the club-

VOL. XXXII, PART II .—NEW SEE. P

224 ABTHUE W1LLBY.

shaped gland is a rudimentary structure which formerly servedas an intestine, or whether it is what van Beneden and Julinwould call a primitive collateral organ, which in the Ascidianshas come to function as an intestine. A great many con-siderations render it only reasonable to suppose that the intes-tine and anus of Amphioxus are primitive, and hence theclub-shaped gland must be regarded as an ancestral organoccurring in company with a posterior anus.

Assuming, then, that the ancestor of Amphioxus possessed acollateral organ of unknown function situated behind theprimitive median endostyle, with internal and external orificesas described, it is difficult to conceive how it can have becomeevolved, except by the modification of some pre-existing organ.The spectacle of an unpaired hypoblastic gland opening to theexterior on the one hand, and into the alimentary canal on theother, without having a morphological equivalent in the wholeanimal kingdom, is certainly too much.

There are thus two points to settle :1. What was the nature of the pre-existing organ referred

to above ?2. Was it paired or unpaired ?It may be remembered that the intra-buccal orifice of the

club-shaped gland, which was described and figured for thefirst time in the paper by Professor Lankester and myself quotedseveral times above (8), is not present in the very young larvae,but forms later. This is probably another instance of a slightand unimportant retardation in development.

Before becoming acquainted with the views of van Beneden andJulin I had come to the conclusion, on grounds which seemedsufficient, that the club-shaped gland was a modified gill-slit;and, as I am aware of no other attempts to account for the mor-phology of the gland, I will give reasons for so regarding it.

In the second place, there are reasons for supposing that itwas originally one of a pair of gill-slits, and further, that theoriginal pair actually exists in the larva of Amphioxus—themembers of the pair being the club-shaped gland and the firstprimary gill-slit.


If, then, the club-shaped gland is shown to be a modifiedgill-slit on the one hand, and is homologous with the intestineof the Ascidians on the other, one is led to the apparently pre-posterous conclusion that the intestine of Ascidians is themorphological equivalent of a gill-slit. The secondary intes-tine of Ascidians is certainly not homologous with the intestineof Amphioxus. No other homology was thought of beforevan Beneden and Julin suggested that of the club-shapedgland of Amphioxus. My observations do not run. counter tothat suggestion, except possibly in so far as they show thatthe club-shaped gland is a modified gill-slit. The question is,can it be both ?

Reasons for regarding the Club-shaped Gland as aModified Gill-slit belonging to the Right Side —the Corresponding Slit of the Left Side being re-presented by the First Primary Slit.

1. The club-shaped gland communicates with the exteriorat one end and with the alimentary canal at the other end.

2. Its internal aperture, being dorsal to the larval endostyle,is therefore to the right of the morphological middle line.What was origiually a simple branchial passage has thus beenvirtually drawn out into a long tube, and the external aperturehas been shifted from the right to the left side.

S, The first primary slit, being below the endostyle, is there-fore to the left of the morphological middle line, and theappearance which it presents of being distinctly behind thegland is superficial and secondary. It lies in the region of thesecond myotome.

4. The club-shaped gland arises simultaneously with, and ina similar manner to, the first primary slit, as a ventral pouch ofthe alimentary canal (see Hatschek, 3); and though it arisesclearly in front of the first slit, this may be due partly to theremarkable tubular form that it has attained as a result of itsevolution. I am not attempting to explain why it passesacross to open on the left side, but as it does so it must pass

226 AETHUE WILLBY.

either in front of or behind the unmodified slit with which itwas primitively paired. As a matter of fact, of course, itpasses in front of that slit, and so it has come—possibly in somemeasure owing to the curious torsion which has affected thisregion—to arise in front of the first gill-slit, with which itreally on this view corresponds as a pair. I t thus belongsmorphologically to the region of the second myotome.

5. The atrophy of the club-shaped gland occurs simulta-neously—or nearly so—with the atrophy of the first primaryslit. This fact is worthy of the utmost emphasis.

6. In the formation of the secondary slits the first of thempairs with the second primary slit, and no true secondary slitever appears to correspond with the first primary slit.

7. After the simultaneous formation of the club-shapedgland and the first primary slit (the mouth and anus being alsopresent) there is a prolonged interval, during which no furtherformation of gill-slits occurs.

This might indicate a distinction between the first primaryslit and the succeeding ones.

These are the reasons, drawn from facts, all of which, exceptNos. 4 and 7, are illustrated by figures accompanying thepresent paper.

Deduction.—The club-shaped gland and first primary gill-slit represent an ancestral pair of gill-slits, which atrophiesin the course of larval development.

How or for what special object a gill-slit could be modifiedinto such a tubular collateral organ is quite another matter,which I do not now propose to consider farther (vide p. 210).Suffice it if the reasons given above are enough to lead us tosuppose that such a thing has indeed happened in the presentcase.

The closure of the hinder primary slits presents an interestdifferent from that of the first, and is the decisive step in whatI have before called the " symmetrization " of the larva. Thetransverse growth which carries the more anterior slits fromthe right to the left side does not extend back so far as thehinder slits, so that they are left in the middle ventral line,

LATER LAHVAL DEVELOPMENT OF AMPHIOXUS. 227

where the endostyle will eventually come to lie; hence thenecessity for their closure.1

MORPHOLOGICAL IMPORTANCE or THE DEVELOPMENT OF THE

ENDOSTYLE OF AMPHIOXUS.

1. Its anterior position in the larva in the part of the alimen-tary canal which is equivalent to the median portion of the" prsechordal" vesicle, shows its entire homology with the endo-style of Ascidians, which is also situated in the median portionof the praechordal vesicle.

2. Its anterior position also supports the view that the endo-style was primitively differentiated in and confined to the mid-ventral line of the ancestral prsechordal vesicle.

3. Its oblique position on the right side is secondary, and isdue to the rotatory growth which affected the anterior regionof the primitive alimentary tract, in correlation with theadaptive forward extension of the notochord.

4. It enables the morphological median line of the larva tobe fixed with absolute precision, and shows that it, viz. themedian line, has come to occupy a secondary and temporaryposition, high up on the actual right side of the larva.

5. The important fact can thus be established that the intra-buccai orifice of the club-shaped gland is situated to the rightof the morphological middle line.

6. The endostyle of the larva of Amphioxus possesses twodistinct halves, fused together posteriorly, but with free anteriorextremities. At first the right or upper half is much smallerthan the lower portion, and this is only natural when one re-members that most of the other structures belonging to theprimitive right side are wholly retarded in their development.

7. In the larva it is confined to the region of the first myo-tome which is destitute of gill-slits.

8. The backward growth of the endostyle of Amphioxus hasbeen acquired concurrently with the multiplication of gill-slitain the segmented region of the trunk.

1 It must be confessed, however, that the meaning of the first appearanceand subsequent atrophy of these slits is not quite clear.

228 ARTHTTE WILLET.

9. Since, as will be shown below, the pharynx of Amphioxusis not homologous, in the true sense of the word, with thebranchial sac of Ascidians, the adult position of the endostyleof Amphioxus cannot be homologous with that of the Ascidianendostyle. The observations recorded in this paper, however,prove unquestionably that the position of the endostyle in theadult Amphioxus is secondary , and that in its origin it is, asstated above, perfectly homologous with the endostyle of theAscidians.

Fur the r Homologies.

It will have been inferred, from what has been stated above,that the pharynx of Amphioxus, which is part of the trunk, isnot truly homologous with the branchial sac of Ascidians, whichis part of the prsechordal vesicle; so that we have to face theparadox that while a highly specialised organ of (speakinggenerally) identical structure and relations, and extending thewhole length of the pharynx, is present in both cases, yet thepharynx taken as a whole of the one is not homologous withthat of the other. This, no doubt, sounds at first like areductio ad absurdum of a morphological problem, butthe figures in the plates accompanying this paper, showing theprimary and secondary positions assumed by the endostyle ofAmphioxus, should assist in the realisation of this at firststartling paradox.

The gill-slits or branchial stigmata of the Ascidians arehomologous with those of Amphioxus only in the sense thatthey are distinct vertebrate structures, but they are nothomologous in their origin, position, or relations. There isthus a sort of group homology existing between them.

In Amphioxus the slits form metamerically in the segmentedregion of the trunk. In Ascidians they form in front of the(however imperfectly) segmented region, and, moreover, vanBeneden and Julin point out that they do not arise metameri-cally one after the other, but irregularly; for instance, the first-formed slit will be the fourth of the series eventually, and so on.

If these conclusions are as sound as they appear to be, it is


interesting to note that there are two kiuds of possible horao-logies, which may be called—

1. Individual homology, the one more commonly met with.This is a morphological factor.

2. Group homology, a physiological factor, but entirelydistinct from homoplasy.

It also follows that the peribranchial cavity of Ascidians isin no sense homologous with the atrium of Amphioxus ; in fact,this is a pure instance of homoplasy.

Batesou (9) points out the probable homology of the probosciscavity and proboscis pore of Balanoglossus with the anterior" prsechordal " vesicle and prseoral pit respectively of Amphi-oxus; and also he tirges the homology between the gill-slits inboth cases. These identifications are likely enough to beaccurate, but I need not recapitulate the reasons for regardingAmphioxus as more closely related to the Ascidiaus than toBalanoglossus, though, since it forms a considerable feature inthe present paper, the absence of a true endostyle in Balano-glossus might be pointed out. The forward position of the rudi-ment of a notochord in Balanoglossus has probably no relationwhatever with the forward extension of the notochord inAmphioxus.

As to the affinities of Amphioxus with Ammoccetes, I willonly refer to the homology existing between the endostyle ofAmphioxus and the thyroid gland of Ammocoetes, which wasadvocated first by W.Muller and then by Dohrn(7),who workedout the development of the latter structure. It is well knownthat Dohrn regards the thyroid gland, and therefore the endo-style, as representing a pair of gill-slits. This is of course a sideissue, and cannot be fully gone into here; but it may be pointedout that while the thyroid gland gives no reliable indicationof a double origin, the endostyle of Amphioxus is composedof two distinct halves, which are at first upper and lower re-spectively, but eventually become right and left—morphologi-cally of course their relation to one another is always rightand left.

As for the gill-slits in the two cases, attention may be drawn

230 ARTHTJB WILLEY.

to the pair of rudimentary gill-slits, called branchial diverticula,in front of the first pair of true slits, of which Dohrn establishedthe existence in the young Ammocoetes, and which neveractually open to the exterior; and this circumstance shouldbe remembered in connection with the atrophy of the first pairof gill-slits (i. e. the club-shaped gland and first primary slit)which occurs at a certain stage in the larval development ofAmphioxus.

Further, the fact that the larva of Amphioxus passes througha prolonged critical period, during which it possesses only eightor at most nine pairs of gill-slits, is at least suggestive.

ADDENDUM.

The common ancestor of the Ascidians and Amphioxuscannot be properly imagined until further knowledge is forth-coming as to the significance of the primary pair of diverticulaof the prsechordal vesicle and the function of the club-shapedgland.

Gill-slits were probably present in the segmented region ofthe trunk, which have been lost by the existing Ascidians,but whether one pair or several pairs is a question.

If the club-shaped gland is admitted to be a modified gill-slit, as I hope to have shown that it is, there must have beenat least one pair of such slits present.

LlTERATUKE REFERRED TO.

1. KOWALEWSKY.—' Me"m. Acad. Imp. des Sciences de St. Petersburg,' 7thseries, vol. xi, 1867.

2. KOWALEWSKY.—• Arcbiv fur mikro. Anat.,' Bd. xiii, 1877.3. HATSCHEK.—' Claus's Arbeiten,' 1881.4. HATSCHEK.—' Zoologischer Anzeiger,' 1884.5. LANKESTER.—'Quart. Journ. Mior. Sci.,' vol. xxix, 1889.8. VAS BENEDEN AND JTJLIN.—' Archives de Biologie,1 vol. vi, 1887.7. DOHBN.—Studien No. XII, ' Mittheilungen aus Zool. Stat. zu Neapel,'

Bd. vii, 1887.8. LANKESTER AND WIU/EY.—' Quart. Journ. Micr. Sci.,' vol. xxx, 1890.9. BATESON.—'Quart. Journ. Micr. Sci.,' vol. xxv, 1885.

LATER LAltVAL DEVELOPMENT OF AMPHIOXUS. 231

DESCRIPTION OP PLATES XIII—XV,

Illustrating Mr. Arthur Willey's paper ou " The Later LarvalDevelopment of Amphioxus."

All the figures, without exception, were drawn from the living animal, andeach figure is from a separate larva. The drawings are made to scale. PromStage I I I onwards all the slits are seen through the transparent atrial orbody-wall—i. e. they open into the atrium, and not directly to the exterior.

N.B.—The long cilia on the surface of the body are not represented in thefigures, nor in those of the memoir by Professor Lankester and myself (8).Their character and position are well shown by Hatschek in his memoir inClaus's ' Arbeiten,' 1881.

FIG. 1.—Stage I. Right side. Fourteen primary slits—notice small size ofthe first. Six secondary thickenings occurring in a continuous ridge abovethe primary slits, the first of them being above and between the third andfourth primary slits. The praeoral pit of course opens on the left side, and isonly seen by transparency on this side.

FIG. 2.—Stage I. Eight side. Twelve primary slits. Five secondarythickenings, the last being much smaller than the others, and the first beingsituated between the fourth and fifth primary slits. The second as well asthe first secondary thickening has in this case been slightly retarded; thusshowing that the retardation of the first secondary slit, which has been statedin the text to be usual, is not specially important. Of course, the generalretardation of development which affects the whole series of secondary slitsis extremely important. Notice the endostyle and club-shaped gland.

FIG. 3.—Stage II. Right side. Fourteen primary slits, the fourteenthappearing in side view merely as a pit in the floor of the pharynx; this showsthat it is on the way to close up. Seven secondary slits, one having beenadded behind; the first still lies between the third and fourth primary slits ;a new one is subsequently formed in front, so that what appears to be the firstslit or thickening at this stage eventually becomes the second seoondary slit.In this larva the second to the sixth of the secondary thickenings have justbeen perforated. The appearance of a seventh thickening behind is ratherearly to happen at this stage.

FIG. 4.—Stage III. Right side. Twelve primary slits j the first of them isgetting smaller. The fin-chambers seem to form about this stage. There isa small anterior portion of the atrium still unclosed. The right wall of thefuture velum is now seen from this side, owing to the great extent to which

282 AKTHUR WILLEY.

the mouth has bent round the anterior extremity of the alimentary cauaLThere are seven secondary slits, the seventh not yet perforated ; the first isstill above and between the third and fourth primary slits. Notice theflattened tops of the larger secondary slits. All the primary slits behind andincluding the fourth now bend under the pharynx towards the left side. Noticethe endostyle and club-shaped gland.

FIG. 5.—Stage IV. Right side. Thirteen primary slits, the thirteenthclosing. Seven secondary slits, the first and last having only just opened.The true Brst secondary slit has now appeared, and lies above and betweenthe second and third primary slits. The larger secondary slits have begun todouble in to form the tongue-bars. Notice the direction of the long axis ofthe secondary slits. All the primary slits are now bending under the pharynx,and the first is extremely small. The endostyle has definitely commenced togrow beyond the club-shaped gland. This should be noticed carefully andcompared with the condition of the two structures in the preceding stages.The atrium is quite closed in front.

PIG. 6.—Stage IV. Left side. This is to show the exlent to which theprimary slits have come round to the left side at this stage. The number—fourteen—of primary slits is not typical for this stage (see Fig. 5). The rightor anterior half of the velum is seen with a deep focus. There are sixelements of the buccal cirri in the lower half of the oral hood, growing outinto tentacles and covering up the external orifice of the club-shaped gland.The upper half of the oral hood is hanging over the upper wall of the mouthor velum and over the prteoral pit.

PIG. 7.—Stage'V. Right side. This is the very commencement of Stage V;and the first primary slit, though very small, is still to be seen in side view.The secondary slits, of which there are eight, have become larger; and in thethird slit the tougue-bar has actually fused with the ventral wall of the slit.The endostyle and club-shaped gland have the same relations as in Pig. 5.The caecum is commencing to grow out; there is no fixed stage at which thisoccurs. The primary slits have nearly disappeared from this side, but are stillseen below the secondary slits.

PIG. 8.—Stage V. Left side. The primary slits are not fully round to theleft side yet, but the larger of them are commencing to form tongue-bars.The twelfth primary slit is very small, and the thirteenth and fourteenth areclosing. Notice the point of junction of the upper and lower portions of theoral hood; the latter is prolonged anteriorly as a ridge. The right wall of thevelum is deeply focussed.

PIG. 9.—Stage V. Three-quarter ventral view. This shows the firstprimary slit in a condition of approaching atrophy. The twelfth is also closing.The bases of the secondary slits can just be seen.

PIG. 10.—Stage V. Ventral view. This should attract particular attention.


Feeble traces of the first primary slit can just be detected in frout. Thetwelfth primary slit is again seen to be closing. Notice the series of primaryslits curving round to the mid-ventral line behind. The tongue-bars arecomplete in most of the secondary slits, but in none of the primary slits.

PIG. 11.—Stage VI. Left side. Nine (primary) slits, the first being ofcourse originally the second primary ; but it has come to be the first in positionin consequence of the atrophy of the true first. Behind there are traces of theeleventh and twelfth primary slits. Only the left half of the velum can nowbe seen from this side. I do not know what has become of the nephridiumof Hatschek; x is probably part of the ciliated tract of the preeoral pit.Notice the left half of the endostyle which extends under three slits; the righthalf is out of focus. In this larva there were eight secondary slits on theright side.

FIG. 12.—Stage VI. Left side. This shows the twelfth and thirteenthprimary slits in course of closure. The condition of the gill-slits is not quiteso far advanced as in the preceding figure, but the buccal cirri are fartheradvanced. There were seven (secondary) slits on the right side.

FIG. 13.—Stage VII. Left side. In this larva the ninth slit ( = tenthprimary) would probably have closed. The eleventh primary slit is representedby a rudiment. Csecum is commencing to bulge in the region of the twentiethmyotome. Notice the length of the endostyle. The buccal skeleton isbeginning to grow into the upper portion of the oral hood.

FIG. 13 a.—Separate view of the velum of the same larva, showing thefour primary velar tentacles. The right and left halves of the velum are seento be not quite opposite to one another.

FIG. 14.—Stage VII. Right side. The slits are commencing to elongate ina direction at right angles to long axis of body. There are eight (secondary)slits on this side. On the left side of this larva there weie eight (primary)slits, and there was also a median rudimentary slit on the point of closure,viz. the tenth primary, which sometimes persists, as explained above, as theninth slit of the left side. The view of the endostyle is very important,showing its double nature—upper and lower halves. It has grown consi-derably in length. The right metapleur is omitted. The floor of the atriumconsists of the expanded subatrial ridges.

FIG. 15.—Stage VIII. Right side. In this larva there were eight slits oneach side and no rudimentary one behind, so that this is typical of the criticalperiod. The view of the endostyle is again very important. The two halveshave now assumed a definite right and left position. Its double origin can bedetected here for the last time. Notice in this and the preceding figures thehyperpharyngeal band. Right metapleur is omitted.

FIG. 16.—Stage VIII. Left side. Nine gill-slits j the ninth has not fullycome round to this side, but bends under the pharynx.. There were eight slits

234 ARTHUR W1LLET.

on the right side. Notice the ccelom just in front of the c»cum. The blood-vessel in this region is contractile. The endostyle is farther advanced than inthe preceding figure. The right and left walls of the velum are quite opposit eto one another.

PIG. 17.—Stage VIII. Combination of three focussings, showing the last(seventh) pair of slits of a larva in which the ninth primary slit was in arudimentary condition in the mid-ventral line.

FIG. 18.—Stage V. Right side (anterior portion only), to show the endo-style, which has grown a long way past the club-shaped gland between theprimary and secondary slits.

FIG. 19.—Stage VI. Similar view of right side, to show that the club-shaped gland has now atrophied, its remains being still visible on theendostyle.

FIG. 20.—Stage I. Left or oral aspect of a larva. Note pointed anteriorextremity of mouth, ciliated groove, external orifice of club-shaped gland, andone element of buccal skeleton.

FIG. 21.—Stage I. Similar view. Antero-dorsal margin of mouth, com-mencing to hunch up into the ciliated groove. Two elements of buccalskeleton.

FIG. 22.—Stage II. Similar view. Hunching up and sinking inwards ofthe anterior portion of the mouth continued. Three elements of buccalskeleton.

FIG. 23.—Stage II. Similar view. Four elements of buccal skeleton,commencing to grow out as tentacles. When the very young cirri are seenend-on they present a circular appearance.

FIG. 24.—Stage II. Similar view. Five elements of buccal skeleton,FIG. 25.—Stage III. Similar view, to show the anterior half of the mouth

bent at a deep focus towards the right side. Note also the two portions—upper and lower—of the oral hood, the latter containing the buccal skeleton.

FIG. 26.—Stage IV. Similar view. This is an important view of the mouth,showing its change in shape but not diminution in size. The external orificeof the club-shaped gland is covered over by the buccal cirri.

floor ofa/si.

1Z "-prim

ary slit

junction, of upperslower

left wall of velum

,portions of oral hood,

•na

Lp

ap

iUa

,^sllta

fri^U

s'^-

~

/s

floor o

f atriu

m/.

rruuraipore.. ^^

TT

TS

gf

^S

^^

^ ^SE

ci. ^^'JT

HU

ZI

jundwn,ofupper<

£lower velum

portions of 'oralhoads /

Jf-shtaflefis.: of L

eft melapleiu:

'• 9$'slit of left side. //.^xi^dJs'*-

floor of 'atriums. {'W

^prim.sU

t, primary slits.

WiU

ey <

UL

l?s1iJ>ofle/i,sul*.

. ,

'orig

inal Z

t*

en

Oo

style

. ~

jmm

arr

Fur. JX.

exier/m-l orifice- ofdui-skavcA flami:

The Later Larval Development of Amphioxus. · 2006-05-19 · LATER LARVAL DEVELOPMENT OP AMPHIOXUS....

Documents

Transcript of The Later Larval Development of Amphioxus. · 2006-05-19 · LATER LARVAL DEVELOPMENT OP AMPHIOXUS....