230 MR. 0. f*. M. SWYNNBRtON ON THE 



I extracted and reoffered it, but it was again taken hesitatingly and once more 

 escaped. I then offered a second yellow $ , this time stripped of wings ; the bird 

 ate it readily enough after mastication. 



I now offered a Danaida chrysippus with wings. The roller's behaviour on 

 seeing it was most striking. He ruffled up the feathers of his crest and mantle, 

 drew himself up and looked most ferocious, shaking his head with the greatest 

 vehemence as I continued to hold the butterfly towards him. 



On my offei'ing immediately afterwards an Acrcea acara $ with wings, the bird 

 at first simply ignored the butterfly and later shook his head repeatedly. On my 

 continuing to press it, he indulged in his own eqtiivalent of the owl's " apologetic 

 nibbling " several times, pretending to close his bill lightly over the butterfly and 

 each time drawing back with a shake of the head and a clap of the bill and looking 

 straight at me as though to explain that what I was offering him was not 

 acceptable. A moth {Xanthospilopteryx superba), without smell but with wings, 

 was now refused without tasting. On its being reoffered without wings, the bird 

 leant forward, barely closed his bill on it, and at once drew back. 



I now turned and offered the moth to Lanms collurio, which ate it readily 

 enough. Immediately afterwards the roller accepted, crushed, and readily ate a 

 Papilio lyoeus and five Precis natalensis in rapid succession, all without wings. 



[Apparently, at the commencement of the experiment, only just hungry enough 

 to eat wingless P. lyoeus, eaten up to Gharaxes point on Feb. 24. Wingless Precis 

 natalensis and wingless Catopsilia Jlorella would appear to have been at least as 

 acceptable, in the same state of appetite, as the Papilio. I surmised that the 

 roller's behaviour towards the Precis with wings perhaps indicated that it reminded 

 him of, e. g., Acrcea acara. This is exceedingly possible, especially in view of the 

 results of other experiments in which these two butterflies were used, but an 

 alternative possible explanation exists — " wings versus no wings," — which is still 

 more applicable to the bird's treatment of stifi"- winged C.jlorella. 



I also at the time of the experiment judged the roller's refusal to taste Danaida 

 chrysippus, Acrcea acara, and the Xantliospilopteryx to indicate a probable pre- 

 vious acquaintance with those insects. Once more this is exceedingly possible. 

 Alternatively it might be argued that the roller was perhaps so replete as to 

 regard with boredom the eating of any large-winged butterfly with the wings 

 attached. Nevertheless, the special refusal of the Danaida, with every sign that in 

 my subsequent long series of experiments on birds of the same species I came to 

 associate with the very greatest unacceptability, should be given full weight. 



Usually acceptable butterflies of distinctive appearance should have been offered 

 with wings after the refusal of the Danaida, etc. 



The Catopsilia refused earlier is useless to us, seeing that yellow $ C.jlorella with 

 wings has frequently been an object of suspicion in the course of my experiments, 

 even when the birds were quite hungry enough to be in nowise deterred by wings 

 pure and simple. 



At the end of the experiment a preference was shown for wingless P. lyceus and 

 Precis natalensis as against wingless Xanthospilopteryx superha. The fact that 

 no less than five of the Precis were eaten in addition to the Papilio shows that, 

 actually, the roller had not been quite replete.] 



