ABERRANT FORMS OF LAMELLIBRANCHIAT GILLS. 59E …jcs.biologists.org/content/joces/s2-21/84/595.full.pdf · ABERRANT FORMS OF LAMELLIBRANCHIAT GILLS. 59E 5 ... is rather blunt, ...

ABERRANT FORMS OF LAMELLIBRANCHIATE GILLS. 5 9 5

On the STRUCTURE and SIGNIFICANCE of some ABERRANTFORMS of LAMELLIBRANCHIATE GILLS. By K. MITSU-KUEIJ Ph. B,, of Tokio, Japan, late Fellow of the JohnsHopkins University, Baltimore. (With Plate XXXIV.)

T H E following contribution to the morphology of theMolluscan branchiae is part of an investigation on whichI have for some time past been engaged, under the directionof Dr. W. K. Brooks, in Professor Martin's laboratory atthe Johns Hopkins University. The gills, of which thedescription is here given, are those of Nucula prozima andYoldia limatula. They are extremely interesting becauseof their simple structure, and this account of their minutestructure is published with the hope that it may throwsome additional light on the nature of Lamellibranchiategills. I wish to express here my sincere thanks to Dr.Brooks for his constant advice and assistance. I am alsodeeply indebted for specimens used in the investigation toProfessors A. E. Verrill and S. J. Smith, of Yale College,and to Mr. Richard Rathbone, of the United States FishCommission.

Nucula prozima, SAY.This Lamellibranch shows many departures from the

structure which is generally regarded as characteristic ofthe class. Fig. 1 gives a fair idea of what is seen when theleft valve of the shell has been taken away, and the mantleof the same side removed along the lower border of thevisceral mass near the line xy. a. a. is the anterior abduc-tor muscle made up of several fasciculi; p. a. is the posteriorabductor. It will be noticed that Nucula possesses one ofthe few shells in which the umbo is turned toward theposterior end. In the specimen figured, the visceral mass(v.m.) shows convolutions on the surface, which, under themicroscope, proved to be the male reproductive organ, pro-bably enormously developed for the breeding season, andthis character enables one to distinguish the sex of a speci-men without difficulty. All the males have these convolu-tions, and, when preserved in alcohol, are of a greyish colour.The females show hardly any convolutions, and are much moredarkly coloured. The foot (f) is folded longitudinally atits end, and can accordingly be spread out into a flat circu-lar disc. The labial palpi (I) are unusually developed, andmight at first sight be taken for gills. The inside of the

5 9 6 K. MITSUK.URI.

outer and the outside of the inner palpus are raised intonumerous parallel ridges, which, as shown in the figure,can be seen from the outside, and do not extend to thelower margin. At. their posterior end there are two remark-able structures. One of them is a hood-like structure(/. b., figs. 1 and £), which is the posterior prolongation ofthe united upper edges of the inner and outer palpi. Theother (/. a., figs. 1 and 2), lying immediately below the first,is a long tentacular appendage. It is a hollow tube, open,however, along a line on its posterior aspect, and having itscavity continuous with the space between the two palpi.As it has been seen protruded, with the foot outside of theshell (Woodward's 'Manual of Mollusca,' p. 426), andsince, in alcoholic specimens, a great deal of dirt and sandis found along its length and between the palpi from itsbase to the mouth, it is no doubt a food-procuring organ,probably sending a constant stream of nutritive matters tothe mouth by means of its cilia. It is interesting to noticein connection with this appendage that in Nucula, the gills,unlike those of ordinary Lamellibranchs, must be practicallyuseless for obtaining food, as will be evident from the fol-lowing descripion of them.

The gill (g., figs. 1 and 3) is comparatively small. It issituated quite posteriorly, and is suspended by a membrane(wt., figs. 1 and 3), which is attached to the body along thebroken line zyzw. It is united to the visceral mass (v.m.)from x to y, and to the upper part of the foot (_/, fig. S) fromy to z (see figs. 1 and 3). At the last point, having cometo the median line of the body, it joins with its fellow of theopposite side, and they continue in this way as far as w]Here they separate again, each proceeding to the posteriortip (p) of the gill of its own side. It should be remarkedthat, as the point x is further from the median line of thebody than the point y (fig. 3), there is a considerable freespace beneath the suspending membrane of the gill.

When we turn to the gill itself, we find an altogetherunusual structure. Fig. 4 shows it dissected out and seenfrom below and slightly from one side. In general appear-ance it resembles a boat which is suspended by its keel.xcp, fig. 4 (seen in cross section at ij, fig. 5), is the lineof attachment and corresponds to the keel; x dp, fig. 4(seen in cross-section at d, fig. 5), represents the bottomline of the hollow of the boat. The latter is bounded bythe two surfaces xapd and xbpd (fig. 4 ; seen in cross-section at b d and a d, fig. 5). The anterior end (x, fig. 4)is rather blunt, while the posterior end (p, figs. 1, 3, and 4)

AfeERRANT FORMS OF LAMELLIBRANCHIATE GILLS. 5 9 7

is quite pointed. The resemblance of the gill to a boat is,however, only very superficial, as the gill is not one solidmass, but is made up of a series of paired plates of a peculiarshape, placed one after another from the anterior to the pos-terior end. A little dissection under a lens will show thatthe part above the line x dp (fig. 4) and below the line ofsuspension (x cp), is continuous along the entire length ofthe gill, and that, with this part for the stem, the plates aregiven off, one after another, in pairs to the two sides (seefig. 5). The plates constitute the proper respiratory parts ofthe organ. They are largest in the middle, and diminishin size toward the two extremities.

It is evident from this description that the gill in Nuculais of quite an exceptional nature. It does not, as in mostLamellibranchs, extend along the whole length of the sideof the body, constituting the most conspicuous object of themantle cavity, but is comparatively insignificant, beingpushed back and freely suspended in the mantle cavity. Itcannot, therefore, divide the latter into the suprabranchialand infrabranchial chambers, and is, of course, utterly de-void of any structure like the ciliated water-passages in theordinary gill, for driving water from the lower to the upper.It cannot, also, as has been said, serve as an effective food-procuring organ. The gill in Nucula must for these reasonsbe of vastly less functional importance to the animal than itis in common Lamellibranchs, and, so far as I am able tosee, serves only as the organ of respiration. It seems to me,however, that the division of the mantle cavity into theupper and lower chambers is begun in the posterior part.It has been seen that ventral to the membrane suspendingthe gill (m, figs. 1 and 3) there is a large space continuouswith the general branchial cavity, and there certainly isa space dorsal to this membrane. These spaces seem to bethe rudiments of the supra- and infrabranchial chambers.Moreover, the arrangement of the different parts at theposterior end, as seen in fig. 3, recalls that of the correspond-ing parts in many of those genera in which the mantle cavityis divided into two parts. It is not difficult to conceive howthe same division might be brought about in the case ofNucula, by proper development of the gill and the mem-brane.

Fig. 5 shows a pair of opposed plates considerably en-larged. The solid part (i dj) which I have called the stem,and which is continuous throughout the whole length of thegill, together with the suspending membrane (k ijl) is seenin cross section in the middle, and from this middle portion

VOL. XXI. NEW SER. B, R

5 9 8 K. M1TSUKURI.

the paired plates (e. e.) are seen to proceeds *The colouredpart at the bottom represents the complex chitinous frame-work. The membrane (k i j I) is made up of fibroustissue, the bundles of which this is composed crossing eachother in many directions. Its free surfaces are coveredwith columnar epithelium. The stem consists mostly of asolid mass of large irregular cells with rather large nuclei.There are, I am almost certain, two blood-channels exca-vated through it; a lower larger (n), and an upper smaller(o). The latter seems to be in connection with a free space(q.) found often in sections of the suspending membrane.The large channel («) sends a branch (r) into each plate.The fibrous tissue found in the upper membrane dips downinto this part at regular intervals, viz. between every branch(r) of the lower blood channel (n). How these fibres endbelow, when they reach the chitinous framework, I havenot been able to make out. A few fibres (w) are sent downinto the plate a little above the blood-channel (r), andgradually approach and finally touch the latter near itslower end. A few more fibres (t) are seen along the upperedge of the plate. Exactly what this fibrous tissue is I amunable to make out, but it seems to be some sort of toughconnective tissue, with perhaps muscular fibres more or lessintermixed. That it is very tough and serves as a supportto the whole structure is seen by the fact that the fibresoften stick out beyond the broken edge of the soft tissues.The trough of the chitinous framework is seen at s, incross-section. It extends along the whole length of the gilland sends but two branches into each plate. I have ob-tained the appearances, in some sections, of a bundle offibrous tissue running in it and filling it. The frameworkwill be described more fully further on. The plates (e), theproper respiratory organs, are comparatively speaking verybroad and quite thin, and hang down from the solid part ofthe gill. The epithelium of the plates which is repre-sented in the figure as ending abruptly at the edges i d and

j d, turns at a right angle at these lines to cover the stem,and is soon reflected outwards again to form the epitheliumof the next plate in the series. This is evident from aninspection of fig. 8. Each plate may be said to be simplyan enormously widened blood-channel (fig. 6), and as theblood is necessarily spread out in a thin layer over a largearea, the purposes of aeration must be admirably served.The columnar epithelial cells seen at ad, fig. 5, are verycharacteristic of the plates under a microscope, and are thecells [da, fig. 6) around the chitinous bars (h, figs. 5 and 6)

ABERRANT FORMS OF LAMELLI BRANCHIATE GILLS. 5 9 9

seen in optical section. The surface of the irregularly rect-angular cells placed just inside these columnar cells in fig.5, ought therefore to he continuous with the outer edge ofthe columnar cells, but in order to avoid confusion is not sorepresented in the figure. This is also the case with thecubical cells along the upper edge. The chitinous support(h, figs. 5 and 6) of the plate runs near the lower edge (fig.5) to its tip (a or b, fig. 5), and is made up of two entirelyseparate parts (seen in cross-section in fig. 6) applied closelytogether. Owing to the shape of these parts there is, how-ever, a narrow oval space between them. This space, aswill be shown further on, is continuous with the space in thetrough (s, fig. 5) of the stem. The cells along the loweredge of the plate are columnar, and surround the chitinoussupport in a characteristic manner shown in fig. 6. Theirsurface outlines are irregularly rectangular, contrastingwith the irregularly polygonal cells covering the rest of theplate. The branch (r, figs. 5 and 6) of the lower blood-channel (n) in the stem, is seen to be circular in cross-sectionand to bulge out the surface of the plate. These points arenot, however, constant, as the vessel is sometimes constrictedinto more or less separate channels, while the amount ofbulging seems to depend on the quantity of blood present.The remaining part of the plate (e, figs. 5 and 6) is flat andquite thin, enclosing a broad blood-channel between its twoepithelial surfaces. It is here no doubt that the aeration ofblood is accomplished. The cells of this part are cubical, asseen in fig. 6. Some of them send processes inward to joinothers from the opposite side. This gives a labyrinthineappearance to this part of the plate. The course of theblood is evidently from one blood-channel in the stem to theother, through the space in the plate. For instance, theblood may start from the upper channel (o) in the stem, goto the broad flat part (e, fig. 5) of the plate where it getsaerated, then enter the branch (r), along its upper edge, andrun up this to reach the lower blood-channel (v e) in the stem.This is, however, a purely hypothetical course. I have hadno means of determining whether the blood goes from theupper to the lower channel or vice versa.

The framework which supports the gill can be separatedout by heating it in dilute caustic potash, as it is insolublein weak acids and alkalies. It is stained by carmine andother colouring reagents. Whether it is really formed ofchitin I do not know, but as previous writers have describedthe substance as of that nature it will be convenient to usethe term " chitinous support" for the present. The frame-

600

work consists of a trough (seen in cross-section at s, fig. 5;longitudinally from helow in fig. 8; diagramatically repre-sented in fig. 7), which runs along the whole length of thegill, and from which a pair of closely-applied parallelbranches (h, figs. 5, 6, 7, and 8) is given off into each plate.The trough is divided into two unequal parts: an upperlarger and a lower smaller, by a cross piece (c. p., figs. 5and 7), which stretches from one side of it to the other, alittle below the middle. This cross piece is not, however,continuous, but is pierced through by oral openings (o v, figs.7 and 8) whenever branches are given off laterally to theplates. The space enclosed between each pair of closely-applied branches (see h, figs. 6, 7, and 8) is connected withthe lower compartment of the trough by means of somewhatcircular openings (op, and o'p', figs. 7 and 8) found nearthe bottom. In fig. 8 the letters a, a, a, are placed oppositeeach pair of the branches that go into a plate. It will beseen how one half of the chitinous support of one plate, afterforming an arch at the trough, turns round to enter thenext plate in succession, and to constitute there one half ofthe support of that plate. The framework treated with pot-ash, and sometimes without any treatment, shows markedlongitudinal striation (fig. 8), and some of its fibres stick-ing out at the broken edge beyond the others resemble inappearance the fibres found in the suspending membrane,at t and u, fig. 5, and give reasons for thinking that the•whole chitinous framework is nothing but the fibrous tissuefound in other parts cemented closely together and formingone cohering mass.

Although, owing to the state of the specimens, I haveobtained only here and there evidences of cilia, it seemsreasonable to suppose that the whole gill is covered withcilia. On two rows of cells (/. / . , fig. 6 ; d. a., fig. 5) on thelower edge of the plate I believe there are larger cilia thanon the rest, as I have now and then seen their remains, andas, without any question, cells in the corresponding posi-tions in Yoldia have long and conspicuous cilia.

Yoldia limatula, SAY.

Yoldia resembles Nucula in several structural peculiari-ties—in its well-developed labial palpi, with their peculiarfood-procuring appendage, in its feather-like gills, in the pos-terior position and comparatively small size of the gills, andthe consequent absence of the division of the mantle cavityinto the supra and infrabranchial chambers. It differs from

ABERRANT FORMS OF LAMELLIBRANCHIATE GILLS. 6 0 1

Nucula in having a siphon, and further shows its departurefrom the ordinary lamellibranchiate structure in having ahighly specialised tactile organ in the siphon.1

The gill, although different in details from that of Nucula,is essentially of the same structure as the latter. It is sus-pended by a membrane, as in Nucula. Fig. 9 shows it dis-sected out by itself. The line of suspension is xcp; x dpis the ventral median line, and corresponds to x dp in fig.4. As in Nucula, the gill is made up of a series of pairedplates, placed one after another, and attached to the centralsolid stem continuous throughout the whole length of thegill. The plates do not, however, project downward, as wehave seen in the case of Nucula, but here turn upward (seefig. 11). The plates are largest in the middle, and graduallybecome smaller toward the extremities. At the front end(je, fig. 9) there is a rather interesting arrangement. Fig.10 shows diagrammatically the relations of the various partsat the anterior termination of the gill. It will be seen thatthe plates of the gill gradually become smaller and finallydie out toward the front, and the gill is continued simply asa flat membranous structure {x, fig. 10), which goes intothe visceral mass (p.m., figs. 9 and 10). A cross-section ofthis part shows that at its lower portion, at least, there is ablood-channel, probably continuous with one of the channelsin the stem of the gill. In some specimens this membrane-like portion of the branchia is longer than in others, andgoes some distance around the visceral mass.

Owing to the rather poor state of preservation of thealcoholic specimens, I have not been able to make out thehistology of the Yoldia gill as fully as I should like, but thefollowing description I believe to be correct in essentialpoints :—Fig. 11 represent an opposed pair of plates, and cor-responds to fig. 5 of the Nucula branchia. The suspendingmembrane (k ijl) consists of fibres crossing each other inseveral directions, and is covered on its two surfaces bycolumnar epithelium. The solid stem {idj) of the gill hastwo blood-channels, an upper (w) and a lower (o). Thelatter seems to be in communication with a comparativelyfree space (q) in the middle of the suspending membrane.Directly below the upper blood-channel (o) there is a bundleof tissue, which appears to be fibrous, running the lengthof the gill (seen in cross-section a t / , fig. 11). It serves nodoubt for support. The floor of the lower blood-channel (f)is covered by a V-8naPed bundle of longitudinal fibres (s).

1 W. K. Brooks, 'Proo. Amer. Ass. Adv, Sci.,' 1874 (end of note).

602 K. MITSUKTJRI.

This would seem to be homologous with the trough-shapedchitinous structure in Nucula, but seems to be formed of thesame fibres already referred to several times, which arefound in the suspending membrane and other parts of theNucula and Yoldia gills, and I cannot establish any con-nection between this bundle and the chitinous bars (h, fig.11) in each plate. The latter, when they reach the longi-tudinal bundle (s), make a bend and turn out again to enterthe next plate in the series. In some sections I have ob-tained indications of a very thin layer of chitin beneaththe fibrous bundle [s), which may, therefore, correspond tothe fibres found in the trough of the framework in theNucula gill (see above). If, however, this V-snaPed struc-ture is really homologous with the trough of the Nuculagill, it goes far in support of the view advanced above, thatthe chitinous framework is really made up of the fibroustissue which is found in other parts, here cemented intoone compact mass. In such a case fusion has gone furtherin Nucula than in Yoldia, and we see in the first genus thetrough well united with the branches (h) in each plate.The plates (e, fig. 11) in Yoldia spread themselves upwardinstead of downward, as in Nucula. The chitinous bars(h), of which there are two in each plate, follow the curveof the plate and end rather bluntly about half way up, atthe point a. That the part from d to a corresponds to thelower inner edge of the Nucula plate (da, fig. 5) is shownby the characteristic rows of columnar cells having longercilia than those found in other parts of the gill. There isanother system of chitinous strnctures (ch, fig. 11). Manyfine chitinous filaments come down together in a bundle oneach side from the suspending membrane, and as soon aseach bundle reaches the plate of its own side filamentsspread themselves out like the frame of a fan over the wholeplate. Several fibres sometimes proceed together, and thenseparating give the appearance of branching. They arefound directly beneath the epithelial cells that cover theplate. The .effect of this framework must be to keep theplate well spread out for the purpose of aeration. I havenot succeeded if} obtaining any single section which showsthe structure of the plata well, but from the comparison of agood many sections which I have made, I feel tolerably surethat the whole space between the epithslkl surfaces is per-vaded by what Peck1 calls " lacunar tissue " (fig. 12). It is

1 R. Holman Peck, " The Minute Structure of the Gills of Lamelli-branch, Mollusca," ' Quart. Journ. Micros. Soi.,' 1877.

ABERRANT! FORMS OP LAMELLIBRANCHIATE GILLS. 6 0 3

a loose trabecular tissue with many nuclei and within whosenetwork blood can flow. The space between the chitinousbars (h, fig. 11), which is quite large in Yoldia, seems to betolerably free from this lacunar tissue. Fig. 11 a gives theoutline of the plate seen from one side.

Theoretical Considerations.The gills, here described, of Nucula and Yoldia are, 1

think, the most rudimentary of any that have been studiedso far. In fact, at first sight, the resemblance to the ordi-nary Lamellibranch gill is not apparent, and they suggestmore the Cephalopod gill. But I believe, the homology oftheir various parts with those of more complex gills inUnio, Mytilus, Area, &c, is not difficult to make out.After consulting the articles by Peck (loc. cit.), Posner,1

Lacaze-Duthiers,3 Bonnet,3 and others, and also after ex-amining the sections I myself have obtained of Unio,Modiola, Scapharca, &c, I have no doubt whatever that theplates in Nucula and Yoldia represent the descending orattached limb of the filaments in the outer and inner gill-plates in forms like Mytilus, Modiola, and Area, and accord-ingly are homologous with the folds on the inner lamella ofthe outer gill-plate, and on the outer lamella of the innergill-plate in Unio, Anodon, and Dreissena. If a comparisouis made of my fig. 6 with any of the cross-sections of gill-filaments given by Peck, it will be seen at once howsimilarly the paired chitinous bars are placed, how almostidentically the epithelial cells are arranged around them,how two rows of those cells {l.f., fig. 6)—called by Pecklatero-frontal epithelial—have longer cilia than the rest.In fact, Peck's fig. 12 (a transverse section of a filament ofthe Anodon gill) agrees with my fig. 6 in all essential points.The left hand-figure in his fig. 5 (the superficial view of theedge of a gill-filament of Mytilus showing the latero-frontaland other epithelial cells) and the upper part of his fig. 20(the same view of a gill-filament of Anodon) would pass verywell for the corresponding part in Nucula. So far as I canmake out from rather poor specimens, the latero-frontal cellsin Nucula are strikingly like those represented in Peck's

1 Carl Posner, " Ueber den Bau der Najadenkieme," ' Archiv. fiirmikros. Anat.,' 1875.2 Henri de Lacaze-Duthiers, " M6moire sur le Developpement desBranehies des Mollusques Accphales Lamellibranches," ' Ann. d. Sci. Nat./Ser. iv, tome v, 1856.

3 Robert Bonnet, " Der Bau u. die Circulations-verhaltniBse der Ace.phalenkieme," ' Morphologisches,' Jabrbuch iii, 1879.

60-1 K. MITSUKtmi.

fig. 20. If, then, the plates in the gills of Nucula andYoldia represent the gill-filaments in other genera, itfollows from the embryological observations of Lacaze-Duthiers' (loc. cit.), and from the position of the chitinousbars in the plates, that they are homologous with the de-scending limb of the gill-filaments in ordinary Lamelli-branchs. Professor Huxley seems to have no doubt what-ever of the homology stated here, as will appear from thequotation given further on. Admitting, then, that this sup-position is correct, and that the gills in Nucula and Yoldiaare in an unusually rudimentary condition, what light, ifany, do they throw on the organogeny of the Lamellibran-chiate gill ? But, before proceeding to the discussion of thispoint, let us review briefly what theories have been ad-vanced as to what is the most primitive type of the bran-chiae of this group. Setting aside older authors likeWilliams and Hancock, I consider the articles, alreadyalluded to, by Peck, Posner, and Lacaze-Duthiers as havingthe most important bearing on the subject. Posner, aftera careful histological examination of the gills of Anodon,Unio, Cardium, Mya, Mytilus, Ostrea, Pecten, Pholas,Pinna, Scrobicularia, Solen, Solecurtus, and Venus, putsforward, although with hesitation, the theory that the pouch-like gills of the Unionidse are the most primitive type of theLamellibranchiate gill. Stepanoif,1 so far as I can gather,inclines to this view. Peck, on the other hand, after aninvestigation of Area, Mytilus, Anodon, and Dreissena,comes to the conclusion that " the gill-plates of the Unio-nidse are a highly modified form derived from a simple con-dition in which the gills consist not of plates but of a seriesof juxtaposed independent^/amewfe, such as we see in a lessmodified state in Area and Mytilus." This view is themore generally accepted of the two. The only completehistory of the development of the Lamellibranchiate gill byLacaze-Duthiers (loc. cit.) and all the fragmentary embryo-logical observations on the organ show that the gills are atfirst of a tentacular or filamentary character. Those whoread carefully Mr. Peck's paper, will, I think, feel con-vinced by the arguments he brings forward. So high anauthority as Professor Huxley is entirely of this view. Hesays:—"In its simplest form, the branchia of a Lamelli-branch consists of a stem fringed by a double series of fila-ments {e.g. Nucula). The next degree of complicationarises from these filaments becoming, as it were, doubled

\ Paul Stepanoff, " Ueber die Geschlectsorgane nnd die Entwioklungvon Cyclas," ' Archiv f, Naturgesch,,' 1865,

ABERRANT FORMS OF LAMELLIBRANCHIATE GILLS. 6 0 5

upon themselves at the free ends, the reflected portionslying on the outer side of the outer, and on the inner sideof the inner, series of filaments . . . (Mytilus Pecten).In most Lamellibranchs, the gills are four elongated plates,each of which is in fact a long narrow pouch, with its openend turned toward the haemal face of the body " (r Inverte-brates,' pp. 408-9, Am. Ed.). My own observations lead meto the same conclusion. In fact, it is difficult to see howthe pouch-like gills of XJnio can give rise to such forms ofbranchiae as are found in Nucula and Yoldia. By a verycircuitous route they may have degenerated into theirpresent rudimentary state, it is true, but all recent observa-tions tend to show that while other organs in the Lamelli-branchiata have been steadily degenerating, the gills, on thecontrary, have become highly developed and perform func-tions which the probable change of the animal from themotile to the sedentary habits of life has forced on thesegills. If, then, there has been no considerable degeneration,and if the homologies of different parts of these branchiseare, as I have stated above, the filamentary character of theprimitive Lamellibranchiate gill is placed beyond doubt.

I believe further light is thrown on the subject by thegills of Nucula and Yoldia. Peck shows that the gillsprimarily consisted of a series of filaments, but does notattempt to account for the fact that these filaments havecome out in long rows on the side of the body. I ventureto suggest an explanation. If we reflect for a moment, Ithink we shall see that the gills of Nucula and Yoldia maybe considered as a stem which, being folded on either sideto increase the surface of contact with the water, gives riseto the flat plates which I have homologized with the de-scending limb of the gill-filament of Mytilus and other likeforms. The plates are, strictly speaking, nothing but theepithelial covering of the stem raised into folds and enclos-ing between the two sides of the folds a blood-channel. Inthe case of Yoldia mesoblastic lacunar tissue is carried outinto the folds. According to this theory, the gill of theLamellibranchiata was originally a longitudinal ridge on theside of the body. Probably in this a blood-vessel ran, andmust have served as the organ of respiration. In course oftime, however, this ridge became folded for the increase ofthe surface of contact with the water and thus producedpapilla on its two sides—rudiments of the future gill fila-ments. The gills of Nucula and Yoldia have gone butlittle beyond this stage. I think there is much to supportthis view. Stepanoff (loc. cit.) observed in Cyclas that the

606 K. MITSCKURI.

gills arise first as a ridge on each side of the body. Leydig1

makes the same statement. M. LoveuV observations havea still more important bearing on the point. He says :—" Nous avons, si je ne me trompe, vu la premiere formationdes branchies; nous en savons assez pour &tre su.r qu'ellesse montrent sous la forme d'un cordon fin, renfle & certainsintervalles ; que ces renflements se contournent plus tard enanses, qui s'allongent de plus en plus, et sur lesquelles sedeVeloppent les cils vibratiles reguli&rement disposes et d'unforme particuliere,"3 "Un cordon fin renfle el certains in-tervalles" is, it seems to me, nothing but a ridge with slightswellings or papillae. Lov^n's figures are not exactly clearto me, but what he designates as the gills are certainly infavour of my view. In all the fragmentary embryologicalobservations, the gills are generally seen as papillae, ornothing but the folds of a blood-channel. I have alreadycalled attention to the anterior part of the Yoldia gill wherethe plates die out and the gill is continued simply as aridge containing a blood-channel. Whether this is a rem-nant of the primitive ridge or not it is difficult to determine,but the fact that there can be on the side of the body a thin-walled ridge which, containing a blood-channel, must servemore or less for respiration, goes far in support of the viewhere advanced.

To review the whole matter, the Lamellibranch gill wasperhaps originally a simple ridge on the side of the body,but to increase the surface of contact with the water foldsmay have arisen on two sides of this ridge. If such wasthe case, Nucula and Yoldia are still in a stage only verylittle advanced from this primitive condition. In course oftime, however, as some of the Lamellibranchiata, eitherowing to degeneration or some other cause, become incapableof extensive locomotion, these buds or folds were perhapsprolonged to form tentacular filaments, which, going on intheir development, finally produced such complex gill struc-ture as we see in Mytilus, Unio, Ostrea, and other forms,taking on at the same time functions totally foreign to their

1 Franz Leydig, "TJeher Cyelas cornea" 'Miiller's Archiv,1 1855. Hesajs:—" Die letzte Hauptmandering im ausseren Habitus esfahrt derEmbryo durch die Bildung der Kiemen. Aucli sie wachsen als Leisten vonhinten nach vorne und zwar geben sie urspriinglich von Mantel aus"(p. 62).

* " Bidrag till Kamedornen om utvecklingen af mollusca aceplialaLamellibranchiata," ' Memoirs of the Academy of Stockholm/ 1848,lately reprinted in an abridged form in German.

3 Translated by M. Young, and quoted by Lacaze-Duthiers iu the articlealready referred to,

ABERRANT FORMS OF LAMELUBRANCHIATE GILLS. 6 0 7

original one. Between the simple gills of Nucula and mostcomplex ones known, there are a great many intermediatestages, some going more in one direction, others in another.For instance, Lucina and Corbis are said to have .only onegill-plate on each side (' Owen's Inverteb.'). According toSars, Pecchiola is in the same condition ('RemarkableForms of Animal Life,' G. 0 . Sars). Chamosirea and Myo-chama are described by Hancock (' Ann. and Mag. of Nat.Hist.,' 1852-3) as having the inner gill-plate complete, butthe outer plate lacking the outer lamella. In these tentacularfilaments seem to be fused with each other. On the otherhand, although Area, Mytilus, Modiola, have all the lamellaepresent, the filaments composing them have not fused withone another. It is interesting to notice that Nucula andYoldia, in which the gills have remained rudimentary,have, as Dr. Brooks first pointed out to me, an unusualpower of locomotion, while forms wholly or almost whollyunable to move, as Ostrea, Pholas, &c, possess highly-developed gills.

For some reason the inner gill-plate seems to developfurther than the outer. For instance, in many genera, theinner is much larger than the outer. In Chamostrea andMyochama, already referred to, it is the inner gill-plate thatis complete, and the outer gill-plate that lacks a lamella. Itwill also be seen a little further on that in Anodon the innergill-plate has gone further than the outer in its development.In the embryological study of the branchiae of MytilusLacaze-Duthiers observed that the filaments of the innergill budded out first.

It is very instructive to see the process of secondary fold-ing going on in higher varieties of the gill. The twolamellae of a gill-plate are, in such a case, no longer parallel,but wavy, and the surface of a lamella is thus considerablyincreased. In Anodon this process is perhaps going on, forPeck shows that in that genus the cross-section of the outergill-plate has parallel and straight edges, but that the outerlamella of the inner has a wavy margin. Posner showssuccessive stages of secondary folding in the gills of Pholasdactylus, Venus (sp), Mya arenaria, Ostrea edulis, Solenvagina, Cardium edule, Pinna nobilis.

Diametrically opposite, as the views advocated by Posnerand Peck may seem, it is not difficult to reconcile the two.

If we look over the list of the genera examined by Posner,we shall find all of them, except Mytilus and perhaps Pccten,to possess more complex gills than Unio, and starting, as hedid, from the last genus, it is no wonder that he considered

608 K. MIT6UKURI.

it to possess the primitive gill. On the other hand, Peckinvestigated forms simpler than Unio, and arrived at the-probably true conclusion. Posner simply began wherePeck ended. The two investigators, therefore, supplementeach other, and now, with the addition of the extremelysimple gills of Nucula and Yoldia, the series is fairly com-plete, and it seems to me that the filamentary character ofthe primitive Lamellibranch gill is made tolerably certain.