196 PROCEEDINGS OF THE ACADEMY OF [April, 



bases of the loral feathers pale brown instead of pure white; sides of 

 head and neck, together with entire lower svu'face, excepting the anal 

 region, much darker and more uniform. 



Geographical distribution. — Tahiti Island, Society Islands. 



Description. — Type, female adult; Tahiti Island, Society Islands, 

 November 14, 1899; C. H. Townsend. Pileum clove brown; cervix, 

 back, upper tail-coverts, and part of rump rather lighter and slightly 

 more rufescent with scarcely any metallic gloss; rump with a very 

 light brown band, slightly whitish along the edges of the feathers, and 

 similar to that of CoUocalia francica francica, though apparently not 

 quite so broad or so pale, the feathers with somewhat though not 

 very conspicuously defined deeper brown shafts; wings and tail darker 

 than upper parts, and somewhat blackish with a slight purplish or 

 bluish sheen, but the basal portion of the rectrices and the inner edge 

 of the wing-quills decidedly more brownish — much like the back; 

 median and lesser wing-coverts with the tertials also rather lighter, 

 more brownish than the quills, and somewhat glossed with greenish; 

 sides of head clove brown almost as dark as the crown, the feathers of 

 the lores with lighter brown bases ; sides of neck brown like the cervix ; 

 lower surface the same, but lighter, becoming still a little paler on the 

 abdomen, considerably so and appreciably more rufescent on the 

 crissum; lining of wing clove brown. 



Wing, 122; tail, 58; exposed culmen, 5; tarsus, 10 mm. 



The single specimen of this new and remarkable species was ob- 

 tained by Mr. Townsend during the cruise of the "Albatross" already 

 mentioned. It differs very strongly from CoUocalia francica spodio- 

 pygia, the light-rumped form from the Samoan Islands, in its larger 

 size, lighter, much more brownish upper parts, rather more narrow 

 rump band, darker, more uniform ventral surface, and especially, as 

 from all other species of the entire genus excepting C. leucophaea and 

 C. ocista, by the brownish instead of pure white bases of the loral 

 feathers. From CoUocalia leucophaea it may of course be easily dis- 

 tinguished by the pale band across the rump, as well as by somewhat 

 less brownish upper surface and decidedly paler anal region. From 

 CoUocalia ocista it differs in lack of tarsal feathers, in more brownish 

 upper surface, pale rump band, and lighter anal region. 



The only published name that can by any possibility apply to this 

 species is Hirundo peruviana Forster,-^ based on a specimen said to 

 have been obtained on Tahiti. But unless the description of this bird 

 is very erroneous it cannot refer to thespesia, as the following quotation 



^^Descript. Anim., 1844, p. 240. 



