GIBSON— CEPHALOCHOEDA : « AMPHIOXIDES." 



237 



more left slits, making 21 in all. The position of these last three slits is instructive ; the 

 last of them lies exactly in the (topographical) mid-ventral line, the other two only very 

 slightly to the left of it. Thus they are proved to be persistent larval slits, and not 

 " tertiaries," since we know the latter to be formed simultaneously on both sides, in their 

 definitive symmetrical position. It follows that the larva from which their possessor 

 was derived had at least 21 median slits. Further, we can see behind these slits an 

 indication that some five more have previously existed, in the presence, on the left side 

 only, of coilomic canals uniting hypobranchial and epibranchial coelom ; between these 

 canals the two cavities are separated by concrescences of atrial and branchial epithelia, 

 and on the right side a like concrescence is continuous. If these canals have the 

 supposed significance, and if we may assume the most anterior larval slit to have closed, 

 as it does in the Branchiostoma larva, 27 slits— the number almost always found in 

 A. pelagicm—m&j be assigned also to the larva from which our metamorphosed animal 

 was derived. 



Pig. 3. — Diagrammatic horizontal section through anterior gill- 

 slits of a young metamorphosed animal. Hypobranchial coelom 

 overlaid by the endostyle, shown by the heavy dotted line. 

 The light dotted lines indicate the inner limits of the pterygo- 

 coeles, lying below the epibranchial coeloms, and opening into 

 the stomoooeles in front. The shaded areas represent the gill- 

 bars of the second and succeeding slits. 



The retention by the postirior gill-slits of their larval, median position involves a 

 further retention of the larval asymmetry in the position of the sub-endostylar blood- 

 vessel, or branchial artery. This, together with the hypobranchial coelom, in the roof of 

 which it lies, takes its course well to the right-hand side in the region of the last three 

 slits. The endostyle itself does not yet extend so far back, but appears a little further 

 forward as a band of clear cells : at its hinder end it, too, is asymmetrically affected, and 

 follows an undulating course, being pushed to the right, so to speak, by each left "-ill- 

 slit, but lying in its definitive median position in the intervals between them. Erom the 

 region of the 13th to the 1st gill-slit, the endostyle is perfectly symmetrical. As already 

 mentioned, at its extreme anterior end, immediately behind the velum, it lies slio-htly to 

 the left. The appearance of a section, through a primary giU-bar is seen in PI. 15. fio-. IS. 

 In most of the primary bars, but not in the tongue-bars, chitinous supporting-rods are 

 developed. The atrial epithelium covering the primary bars of the left, not those of the 

 right, side is peculiarly modified, its cells being large, projecting, and of irregular 

 outline. Tongue-bars are not present in the last three left slits, in the last right, or in 

 the first on either side. 



The histology of the pharynx could not be satisfactorily made out, its whole ventral 

 portion appearing as little more than a mass of deeply staining nuclei, in which the 

 openings of the gill-slits were seen as clear areas. One thing is evident, that the 

 branchial epithelium is extremely thick The larval separation of dorsal and ventral 

 portions by means of lateral folds is still seen anteriorly, though it is not so marked as 

 in AmpJiioxides ; towards the middle of the gill -region, the dorsal portion becomes more 



SECOND SERIES. — ZOOLOGY, VOL. XIII. 32 



