206 MR. C. F. M, SWYNNERTON ON THE 



Scbaum), a Tenebrionid beetle, a dragon-fly {Anaa; speratiis, Hagen), and 

 Asilidse (mainly Alcimus ruhiginosus, Gerst.). 



A very large number o£ experiments were carried out on European and 

 Lesser Stripe-breasted Swallows {Hirnndo rustica, Linn., and H. puella, 

 Temm.), but the preference results were hardly in proportion to the time and 

 labour expended, as it was before T had learned to experiment effectively. 

 Other insect experiments that were to some extent wasted for the same cause, or 

 else were very scanty, were on Cercopitliecus pygerythrus, F. Cuv., Papio 

 cynocephalus, Geoff., Crossarchus fasciatus, Desm., Petrodromus tetradactylus, 

 Pet., Dendromys spp., Hypliantornis jamesoni, Sharpe, CoUopasser ardens, 

 Bodd., Lanius colhmo, Linn., Teleplionus senegalus, Linn., Phyllastreplms 

 fiavistrialus, Sharpe, Apalis thoracicus, Shaw & Nodder, Irrisor erytltro- 

 rJiynchus, Lath., Centropus hurclielli, Swains., Loplioceros leiicomelas, Licht., 

 Chrysococeyx cupreus, Bodd., Buho maculosiis, Vieill., Francolinus coqui, 

 Smith, Pternistes humboldti, Peters, Numida mitrata, Pall., Lobivanellns 

 lateralis, Smith, a toad, a lizard (Agama sp.), Mantidse, Harpactor 

 erytlirocnema, and one or two other Reduviidse. 



Unconfined predaceous arthropods tested to some slight extent have been 

 Sjiiders and driver-ants {Dorylus sp.). A list of the unconfined birds will be 

 given later. The main experiments here have been on Bee-eaters (Merops 

 apiaster, Linn.), a ^ushShrike (I)ry oscopus guttatus, Hartl.), a Robin [Tarsiger 

 stellatiis, Vieill.), a Drongo [Dicrurus ludioigi, Smith) — all of these actually 

 wild ; also tame but unconfined Ground-Hornbills {Bucorax cafer, Boc). But 

 attacks, in some cases numerous, were watched in the case of thirty-five 

 species of unconfined birds in all. I have some other evidence bearing on 

 the question " Do Birds eat Butterflies ? '■* besides that contained in Section 1 

 of this paper, and I will postpone any really full discussson of the point until 

 I shall have published it. 



Some attempt has been made to carry out the experiments generally on a 

 sufficiently large scale. They number at the moment of writing about 1600 

 in all, including the 120 experiments and special observations on wild birds 

 that are for the most part comprised in Section 1. I believe that approximately 

 20,000 insects (not including the very numerous small grasshoppers that were 

 so often given simply to alter the state of an animaPs appetite) have been 

 used in all in the course of the experiments, and of these all but two or three 

 thousand were butterflies. And some were offered many times over. Here 

 I would like to say, in view of some recent criticism (Proc. Ac. Nat. Sci. 

 Philad. 1912, pp. 281-364), that Professor Poulton was from the start foremost 

 in urging on me the necessity for testing the question of preferences on no 

 small scale but with masses of material, not with one or two offerings of each 

 species but with many. 



[ comment later on the carnivorous and other animals used. Some of the 

 results were very interesting, I was much disappointed at not securing a 



