[71] THE EVOLUTION OF THE FINS OF FISHES. 1051 



pressed, bat enough of it still remains to compel us to regard this type 

 of tail as still truly heterocercal. In Perca (Fig. 2, PI. Vll) we have a 

 stage intermediate between ISalmo and Cottus. 



The next jjhase in the evolution of gephyrocercy is that shown in 

 Fig. 4, PI. Vlll, in Fierasfer acus* where the process has been a 

 direct one without a previous tendency to the development of hetero- 

 cercy. The straight, probably partly archicercal, caudal filament is 

 absorbed or lost in some way and the blunt end of the chorda abuts 

 abruptly against the integument at the tip of the tail. The rays are 

 not developed all the way round over the end of the (caudal axis, so 

 that there is an actual hiatus, h, between the last epaxial and hypaxial 

 fin-rays. In Echiodon, gephyrocercy is more pronounced, as shown in 

 Fig. 3, PI. VII, because the last epaxial and the last hypaxial rays have 

 been swung round so as to be nearly in contact and parallel, the ab- 

 ruptly terminated vertebral column being included so as not to come 

 into contact posteriorly with the integument. This mode of the devel- 

 opment of gephyrocercy was not preceded by the evolution of hetero- 

 cercy, but it is nevertbeless easy to see that in those heterocercal forms 

 where the urostyle tends to be aborted or where it is greatly bent up- 

 wards it would need little more than the degeneration of the latter to 

 produce the same ora similar result. 



The Eel, while it has a structurally heterocercal tail in the adult, in 

 the young (Fig. 4, PI. IV) the caudal skeleton presents more nearly its 

 embryonic condition, and it is noteworthy that the urostyle is so short- 

 ened and reduced as to be practically included by the surrounding 

 parts, so that the last epural interspinous piece and the last hypural 

 process become approximated much more perfectly than in any of the 

 heterocercal forms hitherto discussed, and a near approach is thus made 

 to a truly gephyrocercal caudal fin. Moreover, the last two interspinous 

 bones if slightly prolonged forward would rest with their proximal ends 

 upon the posterior face of the neural spine of the antepenultimate 

 vertebra, thus simulating the arrangement found in Mola. 



While Amiurus exhibits certain primitive traits, such as the posses- 

 sion of an adipose dorsal and a pneumatic duct, the tail is highly dif 

 ferentiated and strongly heterocercal. The hypural cartilages, how- 

 ever, show evidence of proximal concreiscence, while at least sixe])ural 

 arches or spines have been lost; nevertheless, the tendency towards 

 gephyrocercy is undisguised, and the opisthural element op in Fig. 1, 

 PI. IV, shows that the ancestry of Amiurus during a remotely bygone 

 period possessed more hypaxial bones than the existing species. This 



* Dr. Gill kindly called my attention to this type of tail, which resembles ihat ob- 

 served by us jointly in Lahichthys, a new genus of Nemichtiyoid eels. Something of 

 the same sort is said to occur occasionally in Macrurus, but Dr. Beau thinks that its 

 development in this instance is due to injury or accidental loss of the lash-like end 

 of the tail, because this mode of development of the caudal extremity is not coustant 

 ^n the same species of this genusj. 



