226 MR. 0. F. M. SWTNNERTON ON THE 



reader to the end o£ my introductory note to Section 3, and, again, to the 

 commencement of Experiment 494. 



Note. — I had already long drafted this introduction, with its reply to a hypothetical 

 critic and its description of the precautions taken to render the experiments and their inter- 

 pretation as reliable as possible, when my "critic" suddenly materialized. For I was lent 

 by Mr. Marshall a copy of Mr. W. L. McAtee's very comprehensive if somewhat over- 

 vigorous criticism (Proc. Ac. Nat. Sci. Philad. 1912, pp. 281-364) of " The Experimental 

 Method of Testing the Efficiency of Warning and Cryptic Coloration in protecting Animals 

 from their Enemies." 



Having read it, I decided, for the sake of completeness, to expand my original statement 

 by the inclusion of points raised by Mr. McAtee that had not occurred to me as difficulties, 

 but of which it seemed to me that my experiments had in any case supplied the explana- 

 tion, and I hereby gladly acknowledge my indebtedness for them. I have, I think, omitted 

 none of any importance, and I believe that Mr. McAtee, in view of the facts stated and yet 

 to be published that were not in his possession when he wrote, is likely to be the first to 

 agree that the experimental method is perhaps less unreliable, and those on whom he chiefly 

 concentrated his criticism less indifferent to the necessity for abundant evidence than he 

 had supposed, that the "strongly contradictory conclusions different experimenters have 

 drawn " (p. 357) were probably merely the result of their not having carried their experi- 

 ments quite far enough, and that any unintentional misinterpretation that may have resulted 

 from the same cause has been, at worst, by no means confined to those who favoured the 

 selectionist view. 



2. EXPERIMENTS ON CAPTIVE EUROPEAN ROLLERS 

 {CoBAciAs GAERULus, Linn.). 



A. Introduction. 



1. Probable food in the wild state. We have four species of Coracias (and 

 one of Eurystomus) at or near Chirinda, and of these Coracias garrulus is 

 usually by far the commonest during its stay. 



The other three species are apparently residents, and the four together 

 (or even C. garrulus alone while here) are probably sufficiently numerous to 

 contribute somewhat to the selective factors operating on whatever insects 

 they chiefly prey upon. 



Migratory locusts, both larval and winged, various grasshoppers (including 

 Phymateus), Locustids, weevils and other beetles including Anthia pachyoma, 

 lizards, small snakes, mice (the natives often bait their roller-traps with these), 

 crabs, centipedes, scorpions^ cicadas, Mantidse, fly-maggots, and termites have 

 been found by various observers in the stomachs of birds of this genus. At least 

 three of its local members possess much the same feeding-habits, remaining 

 perched on the summit or on one of the more prominent side-branches of a 

 tree or shrub, quietly waiting for their food to pass. Probably more often 

 than not it is a beetle or grasshopper wandering over the ground that is 

 pounced on, but migratory locusts are followed and captured on the wing : 



