84 E. T. NEWTON ON FISHES' TAILS. 



trum, but beyond this it is not quite certain whether the segmenta- 

 tion includes the bony parts as well as the notochord. The neural 

 arch of the fifth vertebra is broad, but not nearly so long as the one 

 preceding it (rca*). The haemal arch of the fifth vertebra is peculiarly 

 modified, and is reckoned as the first hypural plate (1). The basal por- 

 tion of this arch on each side is enlarged into a rounded head, which 

 fits into a corresponding depression of the underside of the vertebra 

 and forms what seems to be a definite joint. The main stem of this 

 bone on each side is flattened towards its outer extremity, where it 

 joins its fellow of the opposite side, and has much the form of the 

 preceding hasmal arch ; but towards its proximal part each limb of 

 the arch sends a broad plate forwards, and a process for muscular 

 attachment is seen upon its outer side. The main caudal 

 artery passes through this arch, and then seems to pass backwards 

 along the upper surface of the second hypural bone (2). From each 

 side of the fifth vertebra another bony process arises (sp. 1), which 

 jmsses backwards and upwards along the sides of the notochord as a 

 protective splint, and giving off a long plate-like process from its 

 upper and inner side, forms, with the corresponding bone of the 

 opposite side, a roof-like covering for the notochord. Two other 

 splint-like bones (sp. 2) partly overlap these processes, and being 

 lower down on the sides and somewhat more behind, still further pro- 

 tect the notochord. Beyond these a small double-curved bone is 

 seen (sp. 3) lying at the sides of the notochord, the latter, however, 

 being free for some distance beyond it. Only one of these bones is 

 seen in the preparation, but doubtless it is one of a pair, similar to 

 those seen in the carp's tail at the same place. Besides the modified 

 hasmal arch above mentioned (fig. 1, No. 1), it will be seen that there 

 are six other broad plates placed below the notochord. Taking No. 1 

 as the first hypural, it will be noticed that the second (2) is much 

 broader at its outer part, and forms a triangular plate ; but its 

 proximal part could not be traced into connection with the vertebral 

 column, although in the adult this connection does take place. The 

 third plate is narrower, and abuts upon the sixth vertebra of the 

 figure. The fourth plate is again triangular, and its proximal end 

 abuts upon the notochord. The outer and lower corner is not 

 ossified, but remains imperfect. At first I thought this was acci- 

 dental, but it seems to be constant, and to remain imperfect even in 

 the adult. The fifth, sixth, and seventh plates are gradually re- 

 duced in size, and become less triangular. The proximal part of 



