25] THE SKULL OF AM lURUS— KINDRED 25 



family is b-c-4, thus having one more element in the occipital region than is 

 found in Amiurus and adding another vertebra to the cranium. The myo- 

 tomes which disappeared in this case were those which took part in the forma- 

 tion of the protometameric neocranium, most traces of which have entirely 

 disappeared in the teleosts (Fiirbringer). But here, in the larval condition, 

 there are more somites related to the occipital region than in Amiurus and yet 

 the ultimate development is the occipital bones. If we assume with Fiirbringer 

 that there are more verebrae in one case than in the other, our homologies 

 are no longer such, but are analogies of structure without any natural relation- 

 ship. Even within one family it would be possible to have an exoccipital 

 vertebral articulation v/liich was not constant, allowing for the attachment and 

 detachment of vertebrae. If we assume that the first vertebra, however 

 modified by specialization, remains constant throughout the series, and that the 

 changes in the occipital region are brought about by intercalation of parts, 

 then our homologies and our basis for natural relationship are m^aintained 

 throughout all the groups. 



Jordan (1893) in his work on the relation of the number of vertebrae and 

 the distribution of fishes has shown conclusively that homologies cannot be 

 based upon numerical sequence. He counted the number of vertebrae in 

 closely related species of teleosts from northern and tropical waters and found 

 that the tropical forms usually had the smaller number. The cranial nerves 

 and bones were constant and yet there were cases of intercalation and excala- 

 tion of the vertebrae. 



Schauinsland (1906) has also shown that vertebrae can be intercalated, and 

 that myotomes, nerves and blood vessels of the body cannot be serially homolo- 

 gized. There are various degrees of intercalation, from the presence of both 

 arches, nerves and myotomes, to the absence of one or two of these elements. 

 These facts give conclusive evidence that the nerves leaving the cranium pos- 

 terior to the vagus cannot be serially homologized with free spinal nerves. 



The Maxillary region. The premaxillary ossification, one of the few centers 

 of ossification appearing in the cranium at this stage, is a thin horizontal osseous 

 plate lying beneath the anterior end of the ethmoid cornu and extending pos- 

 teriorly below the nasal organ forming the anlage of the nasal fossa floor (Fig. 

 18). It is connected with the ethmoid cartilage by a few strands of fibrous 

 connective tissue, and with its fellow of the opposite side, but otherwise lies 

 free in the mass of embryonic connective tissue in this region. Five or six 

 developing teeth are attached to the ventral surface of the ossification. 



Schleip (1903) says that the bone and the teeth arise separately in Salmo and 

 fuse later. In the youngest Amiurus I have studied, the earlier stages have 

 already been passed through, but some tooth germs lie below, unconnected 

 with the posterior part of the ossification, and lend support to the view that 

 Amiurus resembles Salmo in this respect. The ascending part of the ossifica- 



