764 MR. E. S. GOODRICH OX [Nov. 12,. 



reason to think that it can ever develop separately. Xor is there 

 any evidence that denticles do ever really conti-ibute to form 

 dermal bones. Even in the case of the palatal bones of fish and 

 amphibians, the teeth do not actually combine to build up the 

 supporting bone, but become fused sooner or later to bony 

 sulDstance independently developed at their base. 



The history of the palseoniscoid variety of ganoid scale would 

 seem be this (text-fig. 200, p. 761). It first arose as a dermal plate- 

 of bone to which became attached a number of denticles, eventually 

 forming an outer layer of cosmine — the cosmoid stage. It then 

 sank deeper into the dermis, which grew over the oviter face, and' 

 so enclosed it in a mesoblastic pocket secreting complete concentric 

 layers. The outer dermis continued to give rise to denticles, 

 which thus came to be situated on the top of the sunken scale, 

 and subsequently became attached to its surface. Whether the 

 cosmoid scale ever was provided with such denticles is unknown.. 

 None have been described in extinct forms ; and I have looked 

 for them so far in vain. Possibly the cosmoid scale always 

 remained close under the epidermis, and so there would have been 

 no room for their development. At all events such denticles, 

 are only known to occur in Lepidosteus, Polypterus, and some 

 Siluroids, and the extinct Coelacanths (which have no true cosmoid 

 or ganoid scales). 



This hypothetical history of the cosmoid and palseoniscoid 

 scales is illustrated in the diagrams (text-fig. 200, A to g). 



Whether the lepidosteoid variety can be considered as a modi- 

 fication of the palfeoniscoid type seems to me extremely doubtful. 

 It is true that the scale of " Dcqndius gramdosu,s" as described by- 

 Williamson, might be taken as intermediate, but my own obser- 

 vations on this species do not confirm his description of cosmine 

 tubules. Indeed this scale appears hardly to difFer from the 

 typical lepidosteoid form. Future research on fossil forms may 

 perhaps enable us to determine what has been the history of the 

 lepidosteoid scale — whether it has ever passed through a cosmoid 

 stage. 



So far we have merely developed, in a more modern form, 

 Williamson's theory of the origin of the scales of the Teleostomi. 

 It is obvious, however, that another view might be held. The 

 alternative theory which can be put forward is simply this : that 

 the outer layer of the cosmoid scale is merely a special region of 

 the bony plate, which has come to acquire a dentine-like struc- 

 ture, and not a product of superficial denticles. This theory 

 would difFer from the first only in its account of the origin of the 

 cosmine ; the palajoniscoid scale would still be derived from the 

 cosmoid as explained above. One of the chief difiiculties to be 

 met by this view, is that the more primitive and ancient fish 

 (early Dipnoi, and Osteolepidoti (Crossopterygii)) have the most 

 perfectly developed cosmine. On the whole Williamson's theory 

 seems the best, though it cannot be considered as fully established 



