12 Ridgway — American Families of Oliyomyodian Passer es. 



(3) The Elaininae restricted by elimination of the genera 

 Rhi/nrhorych(S, Legatiis, Myiozetetes, Conopias, Pitangtis, Sirystes, 

 and Myiodynastes, which, except the first (referred to the Eus- 

 carthminae), sliould constitute a subfamily Pitanginse, morpho- 

 logically intermediate ])etween the Elaininae and Tyranninae, 

 and doubtfully separable from the latter. 



In my opinion these suggested modifications in the arrange- 

 ment of the Tyrannidae are, in the main, entirely justifiable ; but 

 I am convinced that they are but a step toward much more 

 radical changes which will be necessary before a satisfactory 

 exposition of the phylogeny of the group can be made; and, 

 while not prepared to forecast these with any degree of certainty 

 will express my belief that (1) the genera Agriornis and Musci- 

 saxicola are each quite sui generis and are not by any means as 

 closely related to Tfeiiio2}t era as the latter is to Tyrannus ; and 

 (2) that a considerable number of genera do not belong to the 

 Tyrannidae at all but must be transferred to other groups, since 

 in none of them is the tarsal envelope exaspidean, the only ex- 

 ternal character except that of slight syndactylism (very variable 

 within the group) which can be said to l)e really diagnostic of 

 the family. These genera are (1) Lawrencia which has a typic- 

 ally Oscine tarsus and is without doubt a member of the Vire- 

 onidae. (2) Sh'yv7atura, (3) Hapalorerciis, and (4) Hfdirura, 

 which have taxaspidean tarsi; (5) Miisrigralla, (6) Sirystes, 

 (7) '' Myiarrlms^'' validus and (8) Kdnvphotrigon, which have 

 essentially holaspidean tarsi; (1)) '' Pogonotriccus^^ zeledoni, 

 (10) '' Myiojxigis" gaimardi, (11) Tyrannidns elatvs, (12) 

 " Tyranmdns^^ (i.e. Microtricciis) semifiavus and hrunneicapilhs, 

 and (13) Ornithion inerme, which have essentially pycnaspidean 

 tarsi, and (14) Cvlirirora, which has non-exaspidean tarsi and 

 only ten rectrices. 



These genera, which I conclude do 'not belong to the Tyran- 

 nidae, unless some new definition of the family be made, may 

 be again referred to in order to show more clearly why they 

 should be excluded from the group under consideration, as at 

 present susceptible of definition. 



1. Lawrencia Ridj^way. 

 (Type Empidonax luniuH Lawrence.) 



This genus has a typical Oscine acutiplantar tarsal envelope. It lias ten 

 obvious primaries, of which the tenth (outermost) is about half as long as 



