336 BULLETIN 50, UNITED STATES NATIONAL MUSEUM. 



broad; more rarely it is stout-terete or almost subulate. Rictal 

 bristles usually well developed, sometimes excessively so, extending 

 beyond tip of bill when pressed forward. Nostril usually small and 

 roundish or oval, rarely with obvious superior operculum, usually 

 partly concealed by antrorse or semierect bristly feathers of the 

 latero-frontal antise. Tarsus extremely variable as to relative length, 

 in some genera of terrestrial habits (peculiar to the treeless districts 

 of southern South America) very long and slender (twice as long as 

 middle toe without claw) ; in other genera (strictly arboreal) decidedly 

 shorter than the middle toe with claw. Tarsal envelope ultra-exas- 

 pidean to semi-pycnaspidean, usually typically exaspidean ; the acro- 

 tarsium extending on outer side to or around the posterior side of the 

 tarsus, the two edges often meeting on the inner side, but usually sep- 

 arated by a narrow (or very rarely rather broad) strip or groove of non- 

 scutellate membrane. " Middle toe usually united to outer toe for about 

 half but in several genera for the whole of its basal phalanx, in two 

 genera (RliyncJwcyclus and Terenotriccus) with at least the basal half of 

 the second phalanx thus united, and usually adherent to inner toe for less 

 than half, sometimes nearly the whole, of its basal phalanx; lateral toes 

 always unequal in length, the outer more or less longer than the inner, 

 reaching (without claw) to about the middle of the subterminal 

 phalanx of the middle toe, never to the end of the latter; hallux about 



a There are, in fact, many types or modifications of the exaspidean tarsus in this 

 group, though the extreme variations are connected by structures intermediate in 

 every degree. In a majority of the genera the acrotarsium completely incloses the 

 tarsus, at least for part of its length, the two edges being in actual contact for a greater 

 or less distance. The opposite extreme is shown, so far as I have been able to ascertain, 

 by a single species, Myiodynastes audax, which, at least in the specimen examined, is 

 separated entirely by a space 1 mm. wide at its narrowest point, the outer portion of 

 the acrotarsium extending beyond the posterior margin of the outer side of the tarsus. 

 The plantar space is essentially nonscutellate, but the lower portion shows, on close 

 examination, indications of small scutella. In this case the tarsal envelope may be 

 said to be semi-pycnaspidean, other species of the same genus (M. solitarius, M. 

 luteiventris . M. bairdi. and M. chrysocephalus) having truly exaspidean tarsi. 



In some of the genera which have the two edges of the acrotarsium actually meeting 

 on the inner side the usual acrotarsial segments or scutella are obsolete, the tarsus being 

 "booted" as in certain Oscines, examples being the genera. Perissotriccus, Platyrhin- 

 chus, Craspedoprion, Myiobius (as restricted by me), and some species of Mionectes; 

 other genera showing nearly the same condition, at least in some of the species. Cni- 

 podectes also has a booted tarsus, but the edges of the acrotarsium are separated by a 

 membraneous interval. Unfortunately, these variations in the tarsal envelope do not 

 seem to be correlated with variations in the degree of cohesion between the anterior 

 toes, form of bill, or other external characters, nor do they conform in all cases to in- 

 dubitable relationships; consequently they have disappointed me in the hope that 

 they might greatly simplify the classification of the group, for they seem of little value 

 l)eyond the definition of genera (even sometimes failing here!) or minor supergeneric 

 groups; indeed, it has been found that each of them is more or less variable within what 

 appear to be proper generic limits. 



