506 Correspondence. [o" t k . 



thurberi, or any other name, if it shows the characters of the form, no 

 matter where it is taken." 



Do not my readers immediately see, with me, the extreme danger into 

 which the spread of this conviction will inevitably lead our science? What 

 will be the value of subspecific determinations by Dwight, Bishop and the 

 others of like mind, in accurate studies of migration and of distribution in 

 general? Can they be used at all, without incurring the risk of making 

 wholly incorrect inductions? If such practice becomes universal, wherein 

 could there be any further use at all for recognizing subspecies and slightly 

 differentiated species? Would we not have to restrict ourselves to dealing 

 with simply black-headed juncos, slate-colored juncos, and gray-headed 

 juncos, or, safer yet, with just juncos? 



The rational employment of the subspecific concept as different from 

 the specific one requires the exercise of judgment based on experience — 

 just such as is needed in any other advanced field of knowledge. Further- 

 more, the essential factor involved in the use of trinomials (as designating 

 subspecies as distinguished from species) is variation. After years of 

 study on the part of scores of systematists in ornithology and mammalogy, 

 there are admitted by all, I believe, but two criteria for use of the trinomial: 

 (1) relatively small degree of difference, and (2) the fact of intergradation 

 either through individual variation (as in insular races) or through geo- 

 graphical blending, where the ranges are continuous. Intergradation has 

 always been, among the greatest number of vertebrate systematists, the 

 basis for the use of the subspecies concept, and it should continue so to be. 

 Now, the existence of normal fluctuational variation in two forms means 

 that there has to be overlapping where the means are sufficiently close 

 together; in other words, intergradation occurs, and the convergent 

 extremes will be alike. In any case, if we take a considerable number of 

 representatives of an animal which is subject to geographic differentiation, 

 from one locality, and another lot from another locality, in a separate 

 area of differentiation, and plot graphically their different characters 

 separately, which is essentially what Dwight has done with color in the 

 Genus Junco, we find that some of the specimens fall together, as demon- 

 strated by him in this particular case; but who, until now, would think 

 of calling such individuals as fall in the small area of coincidence of the 

 polygons by other than the name of the race to which they geographically 

 and genetically belong! 



I insist, Dwight's repeated assertions to the contrary notwithstanding, 

 that we simply must consider locality inhabited as one of the most import- 

 ant characters possessed by a species or subspecies. Otherwise, our 

 efforts to classify specimens as to species and subspecies are liable to be 

 worthless. From time immemorial "habitat" has been included as one 

 of the first and most important diagnostic characters of a species. Why 

 begin to disregard it now! 



The main object of classification, from top to bottom, is to express 

 genetic relationship, irrespective of superficial resemblances or such as may 



