BIRDS OF MIDDLE AND NORTH AMERICA. 351 



in eveiy case coinciding' strictly with uniformit}' or continuit}" of 

 ph^'sical conditions. Thus the form having the widest distribution is 

 that inhal)iting' the Atlantic watershed, or the entire region from the 

 wooded valleys of the Great Plains eastward, while those of most 

 limited range belong to the Pacific slope, where the topographic and 

 resultant climatic features are so varied and complicated. In Califor- 

 nia, for example, practically each distinct drainage area has its own 

 peculiar form, one being strictlv limited to the salt marshes fringing 

 San Francisco Baj^. From the last-mentioned point, inhabited by 

 decidedly the smallest of all the subspecific forms, northward along 

 the coast there is a gradual change, the size steadily increasing, the 

 plumage becoming first more rust}', then more soot}', and linalh^ 

 more grayish, until the extreme limit of variation is reached in the 

 gigantic 3L c. cinerea of the Shuraagin and Aleutian Islands. 



In preparing the following '"key" great difficult}^ has been experi- 

 enced in the attempt to characterize satisfactorih^ the different sub- 

 species of M. cinerea^ since however distinct these ma}' appear from 

 one another when specimens are actually compared (and the differences 

 are perfectly obvious to an unprejudiced eye) the differences are in 

 every instance comparative and therefore most difficult to formulate. 

 When size is involved the measurements of contiguous forms inosculate 

 on account of a decided average sexual difference in size, some males 

 of a smaller form exceeding, in some of their measurements at least, 

 some females of a larger form. It therefore becomes necessary to 

 compare specimens of the same sex. The chief difficulty, however, is 

 to decide as to where the line ))etween recogniza])le forms should be 

 drawn, a matter requiring most careful study of the largest possible 

 amount of material and entire absence of personal bias as to whether 

 the forms recognized be few or many. 



While intermediates connecting some of the forms haA'e not actually 

 been seen hy me, there can not ))e the slightest dou])t as to their exist- 

 ence, their absence being due in every case to lack of specimens from 

 intermediate localities. Between the following forms intergradation 

 may be considered as thoroughly established: 3L c. montana and 3f. 

 c. faUax,' M. c. montan<( and J/] c. Jien'inanni; M. c. inontana and 

 M. c. rnorphna; M. e. lieerinanni and M. c. coopei^i^ M. c. heermcmni 

 and M. c. sainuelis; M. c. saniuelis and M. c. pimllnla; M. c. 

 samuelis and M. c. eleonensis; M. c. morjyhna and M. c. rujina; M. 

 c. rufna and JL c. caurina,' M. c. caurina and J/, c. Jienaiensis,' 

 M. c. l:enao:ihsis and M. c. Insignisj 2L e. in.slgnls and J/i c. cinerea. 

 Respecting the northern Pacific coast series (beginning with 3f. c. 

 rnorphna and ending with M. c. cinerea) it necessarily follows, from 

 the nature of the case, that all the forms between the two extremes 

 are themselves intermediates; M. c. rufina connecting J/! c. nwrphna 

 with M. c. caurina., the latter connecting M. c. rafina with M. c. 



