60 Proceedings of the Biological Society of Washington. 



the general principle that a name should not be changed 

 except upon absolutely convincing evidence. However, even 

 if we disregard this and attempt to settle the case by reference 

 to a specific code rule we are again forced to choose canadensis. 

 Article 28 of the International Code of Zoological Nomenclature 

 is the only one bearing upon it. This reads as follows: 



" Art 28. A genus funned by the union of two or more genera or 

 subgenera takes the oldest valid generic or subgeneric name of its com- 

 ponents. If the names are of the same date, that selected by the first 

 reviser shall stand. The same rule obtains when two or more species or 

 subspecies are united to form a single species or subspecies. 



"Recommendation. — In absence of any previous revision, the estab- 

 lishment of precedence by the following method is recommended: 



" (a) A generic name * * * 



'"(/>i A specific name accompanied by both description and figure 

 stands in preference to one accompanied only by a diagnosis or only by 

 a figure. 



" (c) Other tilings." * * * 



It is evident that the first reviser principle can not apply to 

 eases of this kind, for all the early authors were unaware of the 

 facts and never had in mind the idea of revision in the sense 

 of selection, being wholly concerned with priority. Thus in 

 1880 Alston (1. c.) adopted cervina over montana, hut on the 

 erroneous suppositions that one dated from 1818 and that the 

 other was preoccupied. Again, Biddulf (1. c.) in 1885 gave 

 preference to canadensis after finding its date to he 1804, but 

 be too had no date for cervina earlier than 1818. Even Allen's 

 paper of 1912 can not be called full revision, because he con- 

 sidered only the names cervina and canadensis, believing 

 montana to date from 1816 and therefore to lie negligible. 

 Upon the basis of mere adoption, it would be necessary to 

 select montana, its use by Tiedemann in 1808 being the first 

 sul (sequent to 1804. It is obvious, however, that a reviser can 

 not qualify as such in a matter of names of even date unless 

 he is aware that the dates really are even. Moreover, it is well 

 understood that the main provisions of Article 28 are aimed at 

 a class of cases very different from the present one; but Recom- 

 mendation (/>) of this article is clear and definite and evidently 

 intended for cases of any kind not previously covered. This 

 brings us definitely to the selection of Ovis canadensis, for this 



