58 GENEEAL EVOLUTION". 



to the young of Uromastix, etc. The same thing occurs among 

 the Scaroid and Labroid fishes. In this most natural family we 

 find the majority of generic forms provided with a normal com- 

 plete dentition ; in others (Chaerops, Xiphochilus, Pseudodax, 

 etc.) the lateral teeth are gradually and normally replaced by a 

 more or less cutting edge of the mandible ; and finally, in the 

 Scarina and Odacina, the entire mass of teeth and jaws are coa- 

 lesced, forming a beak with sharp cutting edges, the single teeth 

 being still visible in the true Scarus, while they have entii'ely dis- 

 appeared in adult Pseudoscarus and Odax.* Thus, in dentition, 

 the adult Scarus is identical with not fully developed Odax ; 

 Chaerops, with the teeth less confluent, equals a still younger stage 

 of Odax, while those with distinct teeth are the same in this point 

 as the embryos of the highest — Odax, etc. I venture to predict 

 that here will be found a long series of exact parallelism, in which 

 the different genera, resting exclusively on these dental characters, 

 will be found to be identical generically with the various stages 

 of the successively most advanced. 



11. Prof. Agassiz states that the absence of ventral fins is char- 

 acteristic of an embryonic condition of the Cyprinodont fishes. 

 The genus Orestias does not progress beyond this stage in this one 

 point. Probably the genus will be found which will only differ 

 from Orestias in the presence of ventral fins. If so, Orestias will 

 be identical with an imperfect stage of that genus, if, as will ]5rob- 

 ably be the case, the fins appear in the latter, after other struct- 

 ures are fully completed. 



yy. Parallelism in Higher Groups. 



It is not to be anticipated that the series of genera exhibiting 

 exact parallelism can embrace many such terms, since compara- 

 tively few stages in the develoj^mental condition of the same part, 

 in the highest, would bring us back to a larval condition, which, 

 as far as we yet know, has no exact pa7'allel among existing genera. 

 But it is to be believed that the lowest terms of a number of the 

 most nearly allied of such series do of themselves form another 

 series of exact parallelisms. 



Thus exact parallelism between existing genera of mammals 

 ceases, with all characters which are larval or foetal, only prior to 



* Giinther on Ilatteria, "Philosophical Transactions," 1867, ii. I had already 

 noticed the peculiar development in Uromastix, but not published it. 



