NUMBER OF PRIMARIES IN OSCINE BIRDS 183 



establishing (Orn. Notizeti, ii. 327) the family upon these tarsal characters, 

 its position is "still doubtful" (nock zweifelhaft). In G. R. Gray's system, 

 it immediately follows his Emberizidce ; and, indeed, some of the Larks bear 

 no slight superficial resemblance to some of the Buntings. But in the 

 sequence of oscine families adopted in the present and other of my works 

 (which is substantially the same as that employed by the best late authori- 

 ties in this country), I find no more fitting place for the family than where 

 I ventured to put it in the "Key" next to the Motacillidce, and between 

 the foregoing set of 10-primaried families, and such 9-primaried groups as 

 the MotacillidciB, Sylvicolidce, and Fringillidce, which are to follow. These con- 

 siderations lead up to another remarkable characteristic of the Alaudidce ; 

 namely, the apparently variable number of the primaries. 



The number of primaries among oscine birds, whether " nine" or "ten", 

 has been rightly considered an important item in classification, ranking in 

 value with the modifications of the tarsal envelope just discussed. Oscine 

 families, and even groups of families, are conveniently distinguished by 

 this character, and as naturally as by the " booting", or scutellation, of the 

 tarsus. In certain families, however, the distinction fails to hold. In the 

 Vireonidoe, for instance, species of the same genus have indifferently "nine" 

 or " ten" primaries. Thus, Vireo philadelphicus and V. gilvus are two species 

 so much alike that presence or absence of a spurious " first " primary be- 

 comes the readiest means of distinguishing them. Noting this remarkable 

 circumstance in 1885, Professor Biird was led to look rn >re closely into the 

 matter. His results are summed on page 325 of the " Review of American 

 Birds" (see also p. 160) ; from which it appears that in those Vireos which 

 seem to have only nine primaries, two little feathers, distinct in size, shape, 

 and to some extent in position from the general series of primary coverts, 

 are found at the base of the supposed first primary ; while in those Vireos 

 with an obvious spurious first primary, making ten in all, only one such 

 feather is found. "In all the families of Passeres where the existence 

 of nine primaries is supposed to b3 characteristic," he continues, " I have 

 invariably found, as far as my examinations have extended, that there were 

 two of the small feathers referred to, while in those of ten primaries but 

 one could be detected." He does not specify how far his examinations 

 extended. 



Believing this to be an important matter, which would bear further investi- 

 gation, I have been led to look into the question, with the most satisfactory 

 results, confirming Professor Baird's observations and extending them to 

 include every one of the North American families of Oscines, excepting, per- 

 haps, Laniidce (in Collurio) and Ampelidce (in Ampelis). With the possible 

 exception of the two genera specified, I find, on examining numerous genera 

 of all the North American families, that those ratedas 10-primaried have 

 but one of these little feathers, while all the rest have two. 



In clamatorial Passeres, perhaps without exception, there are ten fully 

 developed primaries, the first of which may equal or exceed the next in 

 length. In the single North American clamatorial family Tyrannidce, I find, 

 as before, only one of these little feathers. In a Woodpecker, remarkable 

 among picariau birds in possessing only nine fully developed primaries, the 

 first being short or spurious, there is also but one. 



It seems to be conclusively proven that among the supposed 9-primaried 



