340 BULLETIN r)0, UNITED STATES NATIONAL MUSEUM. 



which have taxaspidean tarsi; (5) Muscigralla," which lias holaspi- 

 dean tarsi; (6) Culicivora ,^ which has non-exaspidean tarsi and only 

 ten rectrices ; (7) Sirystes, which also has holaspidean tarsi and belongs 

 very close to Lipangus, among the Cotingidse, where I have placed it; 

 (8) " Pogonotriccus" zeledoni,'^ which, -possessing ultra-pycnaspidean 

 tarsi, must be transferred to the Cotingidaj; (9) Elainopsis (type 

 ElaineaelegansVelzelnJiilO) Tyrannulus, and (11) Microtriccus {type 

 Tyrannulus seinijlavus Sclater and Salvin), which have essentially 

 pycnaspidean tarsi, and therefore can not belong to the Tyrannidse 

 as here defined, and (12) Myiarclius validus'^ Cabanis, which has 

 holaspidean tarsi and is apparently not distantly related to the 

 Cotinginc genus Attila. Possibly additional removals should be 

 made of genera which I have not been able to examine or from species 

 not available, it being a very obvious fact that comparatively few 

 of the genera as currently understood are really homogeneous or 

 natural groups, having been segregated as alleged genera more by 

 superficial examination than close attention to structural details. 



KEY TO THE GENERA OP TYRANNID^. 



a. Tail not forficate (if forked, the lateral rectrices less than twice as long as middle 

 pair). 

 h. Middle pair of rectrices elongated, much reduced in width along middle portion. 



Copurus (p. 349) 

 hh. Middle pair of rectrices not elongated. 

 c. Exposed culmen more than one-fourth as long as wing; a conspicuous trans- 

 versely fan-shaped crest of yellow, orange, or red, tipped with lilack and steel 



blue or metallic violet Onychorhynchus (p. 352) 



cc. Exposed culmen less than one-fourth as long as wing; no fan-shaped crest (if 

 crested the crest not colored as above). 



cMuscigralla is a very peculiar looking form, the appearance of which does not in the 

 least suggest any relationship to the Tyrannidse, while its holaspidean tarsi certainly 

 exclude it from that family. It probably belongs to the Formicariidse. 



''The genus Culicivora Swainson also is almost certainly not a member of the Tyran- 

 nidse, since it has non-exaspidean tarsi and only ten rectrices. The tarsal envelope at 

 lirst sight appears to be exaspidean; but, while the acrotarsium entirely crosses the 

 outer side of the tarsus and occupies the greater part of the inner side, there is inter- 

 posed, between the two edges, a continuous series of very distinct lozenge-shaped 

 scutella. The style of coloration (conspicuously streaked above) is different from 

 that of any true Tyrannine form and recalls that of some of the SjTiallaxinpe 

 (Furnariidse). 



cTruo Pogonotriccus I have not seen, but F. plnmbeiceps Lawrence (which Von 

 Berlepsch places, I think properly, in Tyranniscus) has typically exaspidean tarsi. 

 P. zelcdoni I have been obliged to make the type of a new genus. Idiotrimis, Avhich, on 

 account of its pycanspidean tarsi, I refer to the Cotingidae. 



d" Myiarchus" ralidus I have placed under Cotingidaj in the monotypic genus 

 Hylonax. 



