ON VARIATION. 59 



posit, and have supposed that this species has also the 

 same power. 1 The difference between such larvae and 

 the adult which has passed the metamorphosis is great. 

 It extends not merely to the branchial processes, but 

 to the splenial teeth, which are shed, and to the palato- 

 pterygoid arch, which is absorbed, and to the pos- 

 terior ceratobranchial and epibranchial cartilages, which 

 are absorbed. In the larva of the C. tenebrosus the 

 palatopterygoid arches and epibranchials are ossified, 

 so that the probability of its being able to maintain an 

 independent existence as a larva is greater than in the 

 case of the A. tigrinum. In this type, then, each spe- 

 cies displays variations concomitant with reproductive 

 maturity, which are not only of generic, but of family 

 significance. In a third species, the Siredon mexica- 

 num, no metamorphosis has yet been shown to take 

 place, so that it is probable that it reproduces ordina- 

 rily while in the branchiferous stage. Yet it is only 

 specifically different from the larva of the Amblystoma 

 tigrinum. 



Excellent illustrations of the serial appearance of 

 generic characters may be seen in the family of the 

 dogs (Canidae). In the true genus Canis, the dental 

 formula is, I. |; C. \; P. m. f ; M. \. The inferior sec- 

 torial (m. i) has a metaconid, and the second inferior 

 true molar has two roots. It not unfrequently hap- 

 pens, however, that the last inferior molar (m. 3) is 

 wanting ; and in some cases the infferior m. 2 has but 

 one root. When in addition to this, as in some of the 

 black-and-tans, in the Mexican naked dog, and in the 

 pug, the inferior m. i loses its metaconid, we have the 

 genus Synagodus. Occasionally the pug dog, and fre- 

 quently the Mexican dog, loses one of its premolars 



^Batrachia 0^ North America, 1888, p. 113, PI. xxii, xxiii. 



