1909.] 



The Caudal Fin of Fishes. 



143 



reason for supposing that the caudal fin was supported primarily in a 

 manner similar to that of other median fins, and that the present hypaxial 

 supports are only a modification of this primary form. Modification was 

 inevitable owing to the adoption of a propulsive function by the tail, and 

 fusion of parts was adopted to ensure the requisite strength. In a large 

 number of caudal fins, the line of fusion can still be traced in the adult. 

 The same thing applies to the formation of epurals ; they consist of neural 

 arches fused with radials or somactids. 



It should, however, be remarked that it does not necessarily follow that 

 every hypural and epural is a compound structure, since, if necessary, there 

 is no reason why a hiiemal arch or a neural arcli should not itself support fin- 

 rays ; in fact, this frequently appears to be the case, but only in connection 

 with the small rays at the anterior end of the fin. 



However, radials do not always fuse with neighbouring parts of the 

 skeleton of the caudal fin, but frequently remain free ; such is more usually 

 the case dorsally, though radials commonly remain as such ventrally, too. 

 This point needs to be emphasised, for it is sometimes assumed that radials 

 or somactids do not share the support of the caudal fin-rays — at any rate 

 ventrally. 



Unfortunately, the species chosen for tliis paper do not illustrate this 

 liypaxially, but Clupea and Trigla possess radials dorsally. The persist- 

 ence of dorsal radials is interesting as illustrating that, under adverse 

 conditions, such as overcrowding, due to excessive upturning of the end of 

 the axis — the radials have persisted while the corresponding neural arches 

 have disappeared or have been greatly reduced. 



In conclusion, a detailed study of the piscine caudal fin has forced me to 

 support the theory that tlie present homocercal caudal is, in reality, a 

 posterior anal, which owes its present position to the great abbreviation of 

 the axis, coupled with the excessive upturning of the end of the chorda. 

 I will even go so far as to suggest that the heterocercal Elasmobranchs may 

 come to be considered as affording evidence for this view, and I hope before 

 long to present all my reasons for consideration. 



