MORPHOLOGY OF TELEOSTEAN HEAD SKELETON. 5] 7 



back growth of the ethmoid region, but by inward extension 

 of the supra-orbital and epiphysial cartilages over a pre- 

 existing fontanelle. 



In Engraulis and Atherina, Pouchet (78, pi. x, figs. 44, 49) 

 found both fontanelles. The anterior one is in precisely the 

 same condition as in Gasterosteus ; but the posterior one of 

 Engraulis is divided medianly by a strip of cartilage into 

 two lateral fontanelles. A transverse epiphysial bar is present. 



I have found this bar in Siphonostoma also (fig. 48, ej). c). 



These facts make it evident, therefore, if we take develop- 

 mental stages into account, that the epiphysial cartilage, so 

 far from being confined to tlie Ostariophysi, as Sagemehl (91, 

 p. 576) thought, is an element widely distributed among 

 Teleosts, and probably derived by them from an ancestor 

 with no other cranial roof. 



In the stickleback it never undergoes any further extension 

 forwards than is shown at this stage. If it is safe to assume 

 that, because primitive Teleosts have a well-developed tegmen 

 cranii, this specialised little fish has lost it phylogenetically, 

 wo have, in the epiphysial cartilage, an interesting example 

 of an old character which has survived all the vicissitudes of 

 a shortened ontogenetic record. 



In the oldest individuals of this stage all the membrane 

 bones, except the mesethmoid, nasals, and parietal s, have 

 appeared; but there is no marked ossification of the chondro- 

 cranium itself. 



Stage IV. — The notochord has undergone only a slight 

 increase in length, as a comparison of figs. 25 and 26 (PI. 30) 

 will show ; but it has now begun to assume characters which 

 anticipate the adult condition. In front it has the same 

 proportions as at previous stages, whilst behind it has become 

 expanded and funnel-shaped (fig. 26, ch. s.) in preparation 

 for the articulation with the first vertebra. Internally it is 

 no longer filled with a few highly distended vacuolated cells, 

 but, especially at its anterior end, it consists of a great number 

 of collapsed cells of a similar type crowded together. Where 

 the extra cells arise is uncertain, for in^iercraniall^ the chorda! 



