— 17 — 



The caudal rays are present even in very small larvae of A. vulgaris. Larvae 

 of less than 2'/2 cm in length showed as yet no incipient ray formation; this was, 

 however, found to be present in larvae 3 cm long, while the full complement was 

 not apparent until a length of Sy^— 4 cm was reached. 



6. Summary. 



If we now consider as a whole the results of all the measurements of per- 

 manent chai-acters published in Report I and in the present work, we arrive at the 

 following general view: 



It has proved possible, by means of the permanent characters, to distinguish 

 between the three species, A. vulgaris, A. rostrata, and A. japonica. All the per- 

 manent characters investigated have proved to be of importance to the classifica- 

 tion, this being, however, only to a lesser degree the case with the caudal rays, the 

 variation of which has been found to be so slight as not to permit of any specific 

 distinction on the basis of the material available. 



It has not been found possible to demonstrate, by means of the permanent 

 characters, the existence of »smaller species« or »races« within the separate species; 

 the question has, however, owing to lack of adequate material, only been fully 

 investigated in the case of the European eel. 



The number of vertebrae is, in general, the most convenient character for 

 separation of the species, in any case, when it is desired to distinguish between the 

 American eel and the other two species'. Of 532 specimens of American and 

 European eels, equally mixed, all save only 0.56 % of the total number could be 

 distinguished, with absolute certainty, while in a similar mixed sample of American 

 and Japanese eels, all the specimens could probably have been classified with 

 certainty. Such cases are, however, purely exceptional. As a rule, we find the 

 same variants for the diflerent species, only the distribulion of these variants 

 exhibiting any difference. It is therefore necessary, in all other cases, to examine 

 a certain more or less considerable number of specimens before anything can safely 

 be said as to the species to which they belong. 



A rostrata is generally remarkable as having the lowest, A. japonica the highest 

 values, while A. vulgaris occupies an intermediate position, this being, however, 

 much nearer to A. japonica than to A. rnstrala. A comparison betv,'een A. rostrata 

 and A. vulgaris gives the following: 



Vertebrae. Anal rays. Pectoral rays. Branchiostegal rays. 



A. rostrata low low low high 



A. vulgaris high high high low 



It is here very remarkable to see, from the above, that A. rostrata, which other- 

 wise has everywhere lower figures than A. vulgaris, suddenly takes up the reverse 

 position with regard to the branchiostegal rays. 



^ If it is desired to separate samples of A. vulgaris and A. japonicu however, the number of 

 branchiostegal rays, which Is far less than that of the vertebrae, is easier to work witli. 



3 



