380 BULLETIN 15 3, UNITED STATES NATIONAL MUSEUM 



The subspecific study of the present species is rendered difficult 

 by three factors: The nongeographic variability of the birds; the 

 apparently great amount of local shifting and wandering of indi- 

 viduals-, resulting in the occurrence of two forms in the same place 

 at times; and the lack of sufficient material. Add to these the un- 

 usually confusing and contradictory statements in the literature of 

 this roller and the task becomes almost hopeless. In the present 

 study I have examined a series of the 22 birds of the following 

 forms: afev 6 (Cameroon and Belgian Congo); aethlopicus 3 

 (Ethiopia) ; praedi 2 (Ethiopia) ; rufohuccalin 2 (Uganda) ; sua- 

 helicus 9 (Kenya Colony and Tanganyika Territory). The Angolan 

 form pulchcr^nmus I have not seen. However, the races about which 

 the greatest difference of opinion prevails are those of northeastern 

 Africa and are the ones which are of immediate concern in the 

 present study. They are afer^ aefhiopicus^ praedi^ and su'Cihelicus. 



Neumann "* was the first to review the forms of this bird and at 

 the time included the Madagascan glaucums as a race of afer. Be- 

 sides glaucurns he recognizes all those listed by Sclater '^^ except 

 yraedi (which Avas not described until 1921), three of the five being 

 new at that point {aethiopicus^ suahelicus, and pulchemmMs) . Scla- 

 ter and Praed *"^ agree with Neumann's results, but do not consider 

 glaummis as a form of afer. The thii'd, and most recent review is 

 that by Bannerman "^ whose conclusions are accepted by Sclater,*^- 

 and, for want of sufficient material, by me in the present work. The 

 races and their ranges as given by the latter are correct as far as 

 I can judge. However, geography and subspecific characters do not 

 always work together. Thus, to take but a few examples : Granvik "^ 

 writes that three s})ecimens from Soy, Kenya Colony. '' * * * 

 exhibit rather great differences in regard to the colours of the rectrices 

 and upper tail coverts. And yet they are shot out of one and the 

 same flock, whence it must be considered very likely that, in spite 

 of the differences, they belong to the same form." When it is 

 remembered that these are among the supposedly most reliable racial 

 characters, the statement has all the more significance. Likewise 

 Gyldenstolpe '"''^ writes tliat the majority of birds from the eastern 

 Belgian Congo are intermediate between afer and 7-ufobuccalis^ but 

 typical examples of both occur there as well. Lastly, I would draw 

 attention to the fact that Mearns collected a specimen of praedi in 

 the same spot (Loco) and on the same date as one of aethioplcus. 



•".lourn. f. Ornith., 1905, pp. 184-186. 



"^ Syst. Avium Ethiop., 1924, pp. 208-209. 



»" Ibis, 1919, pp. 67.3-674. 



«' Rov. Zool. Afritaine, vol. 10, 1922, pp. 149-151. 



•« Journ. f. Ornitli., 1923, Sonderhoft, p. 97. 



•«Kungl. Sv. Vet. Akad. Ilandlgr., 1924, pp. 280-282. 



