INTRODUCTION. 



Taxonomic Considerations. 



While this memoir pertains only to the fishes of New England, and the present part 

 particularly to a very hmited number of species of one genus comprised within a re- 

 stricted area, a proper understanding of them involves more comprehensive considera- 

 tion than the subject at first would seem to imply. In the previous part (Kendall 

 1914, p. 5) it was stated that no one of the external or internal structures, usually 

 enumerated as the chief characteristics of the salmon, will alone distinguish it, but that 

 combinations of characters have to be considered. Opinions vary regarding the value 

 of the distinguishing combinations of characters, and a satisfactory classification 

 depends upon a thorough knowledge of comparative anatomy of salmonoid, and related 

 fishes. According to Regan (1913, p. 288-289) the Salmonidse, in which he includes 

 Thymallidse of Gill, differ from the salmonoid families Argentinidse, Microstomidse, 

 Osmeridse,' Retropinnatidse, Salangidse, Galaxiidae, and Haplochitonidse, in having an 

 opisthotic, and in having upturned vertebrae at the base of the caudal fin. Again Regan 

 (1914, p. 405-406) follows Gill and recognizes the family Salmonidse to comprise two 

 subfamilies: viz., Salmoninse and Coregoninse. In the former he includes Salmo, Sal- 

 velinus, Hucho, and Brachymystax, and in the latter Stenodus, Coregonus, Phylogephyra, 

 and Thymallus, in which he differs from Gill who established separate famihes for 

 Thymallus and Stenodus. The meeting of the parietals on the middle line of the cranium 

 in front of the supraoccipital is the stated differential character of the Coregoninse, 

 the parietals not meeting in the Salmoninse. 



It was this distinction upon which Cope (1871, p. 333) based the family Coregonidse. 

 Gill (1895, p. 117) later reduced this family to subfamily rank through some error of 

 interpretation, thinking Cope was mistaken concerning the relation of the parietals and 

 supraoccipital. But chiefly upon the same character he erected the family ThymalUdse. 



Later Boulenger (1895, p. 299-302) indicated Gill's error and showed that the white-- 

 fishes and grayling were alike in respect to this cranial character. Therefore owing to 

 certain supposed intergradations of cranial and other characters, he recognized only one 

 family, Salmonidse, and that apparently without subfamily groups. This family com- 

 prised all of the previously mentioned genera included in the two subfamilies Salmoninse 

 and Coregoninse. 



There are undoubtedly two distinct groups distinguished by this one character of the 

 parietals. These are the salmons and trouts on the one side and the whitefishes and 

 graylings on the other. Boulenger's supposed intergradations are exhibited by two other 

 genera, Brachymystax and Stenodus. In one species of Stenodus the parietals are separate, 



•Regan establishes this family for the smelts, etc., which had been previously successively placed in Salmonidae, Micros- 

 tomidse, and Argentinidse. 



7 



JAN 1 3 1939 



