V ° 1 "if^ XVI ] Correspondence. 317 



must be allowed for the genius of skilled intuition but a sharp dividing 

 line must always be drawn between ascertained demonstrable facts and 

 hypotheses. 



The truth is, we cannot with absolute certainty identify every specimen 

 we study. Why then deceive ourselves and mislead others by making a 

 bluff at doing the impossible? Why not own up honestly and admit that 

 we cannot name such material? We may state that we think it is so and 

 so and where necessary give reasons for the conclusion, but to pass as fact 

 what is only opinion is not the spirit of modern science. The logical 

 solution of the problem is to name subspecifically only such specimens as 

 are humanly demonstrable and use the binomial for the rest. In other 

 words reverse usual practice and instead of using the trinomial regularly 

 and the binomial on occasion use the binomial generally and the trinomial 

 only where necessity or the facts justify its use. 



P. A. Taverner. 

 Museum Geological Survey, 



Ottawa, Ont., Dec. 27, 1918. 



[While there are some points in favor of Mr. Taverner's plan, which by 

 the way he has put into practice in his article on 'The Birds of the Red 

 Deer River' in this and the preceding numbers of 'The Auk,' there are 

 others which count against it. 



First of all we must realize that the practice of duplicating the specific 

 name when referring to the earliest subspecies of a group — i. e. Melospiza 

 melodia melodia — is by no means universally adopted, and in very many 

 recent papers and all of those of earlier date the binomial Melospiza melodia 

 is used for the first described race and trinomials for the others. Now Mr. 

 Taverner would use this binomial for some one race (seen but not posi- 

 tively determined) of M. melodia. In the A. O. U. 'Check- List' the same 

 binomial is used to indicate the whole group of subspecies of Song Sparrows 

 collectively. Hence we have three different concepts which we try to 

 denote by one expression. In an index these are hopelessly confused and 

 we are likely to miss valuable information about some form that we are 

 investigating because it is masquerading under some specific name where 

 we would never think of looking for it. 



Now as we have in current use a form of name to indicate just what Mr. 

 Taverner has in mind, why not stick to it — i. e. Melospiza melodia subsp.? 

 This would avoid all ambiguity. As his practice stands I find it is quite 

 misunderstood, as all of those of whom I inquired, and who had not read 

 Mr. Taverner's published views on the subject, thought that he was simply 

 following Mr. Leverett M. Loomis in abandoning subspecies entirely. 



Another difficulty presents itself when we try to follow out Mr. Taverner's 

 plan in the matter of closely related species. There are many species that 

 so closely resemble one another that differentiation would be impossible 

 in the field should they happen to occur together. Now Mr. Taverner in 



