﻿78 40 



riorly as an oblique, hooked and compressed point. The scales are large in pro- 

 portion to the size of the lish and give it a spiny appearance. 



The components of the dorsal and ventral armour are quite of similar type; 

 apart from their larger size they only differ from the other scales in that a comb 

 rises from each half transverse diagonal and is connected with the vertical comb; 

 this transverse comb can also be detected on some of the other scales, e. g. near 

 to the ventral plates; it is also found on many of the small scales in the adult. 

 It is thus quite clear that the large plates in the dorsal and ventral armour are 

 simply greatly enlarged scales. The plates in the dorsal armour are not yet con- 

 nected by sutures and those of the ventral armour do not yet overlap. As devel- 

 opment proceeds, the basal plates gradually become very large in proportion to 

 their comb, and in the adult the latter merely appears as part of the sculpture. 

 The sculpture is indeed somewhat different on the elements of the dorsal and 

 ventral armour, but in both it is easy to trace the common ground-plan found in 

 the young. 



The case is quite different with the other scales; in most of these the comb 

 on the scale of the young fish develops the scale plate, described above, with its 

 ridges on the upper side, its marginal teeth etc. The basal plate and the scale 

 plate may grow equally, or the one or the other may develop more strongly, hi 

 order to follow the different stages in the transformation from the original common 

 type in the young to the different forms in the adult, it is not necessary to examine 

 a number of young stages of different age and size; in the adult itself a compar- 

 ison of a series of the smallest and the small scales with the larger and more 

 complicated gives a correct picture of the process of development. 



It appears from the above, that Hertwig was wrong in his view, that the 

 large bony structures of the skin in Centriscus arose from fusion of the smaller'". 



The endoskeleton. 

 Vertebral column (PI. II, fig- 2 and text-fig. 24). This consists of 24 verte- 

 brae, of which I refer 9 to the abdominal and 15 to the caudal region. The first 

 5 vertebræ are elongated and much stouter than the others, especially the first 4; 

 further, they are immovable owing to the manner in which their arches are con- 

 nected and because their spinous processes are bound to the greatly enlarged, 

 anterior interspinous bones, as also from their connection with the dorsal armour; 

 it is only between the head and the 1st vertebra that there can be some movement, 

 in the direction up and down. Two fairly large lateral processes (a) from the 

 anterior end of the 1st vertebra are placed in deep, transverse grooves in the exoc- 

 cipital (cf. fig. 24); they seem at first glance to be. transverse processes, serially 

 homologous with those on the following vertebræ; on closer examination however 

 it seems to me that they must correspond rather to the anterior articular processes 

 on these; their position on the anterior end of the vertebra and their connection 

 with the skull is in favour of this; to the groove on the exoccipital corresponds 



