326 Mil. C. F. M. SWYNNEKTON ON THE 



then tasted well H. ivahlbergi, being again quite obviously uncertain whetlier to 

 eat it or not, but ended by once more rejecting it, though she readily ate a Charaxes 

 achcemenes and a S Ar?-ugico hasuta. 



Ten minutes later she refused the Euralia ivahlbergi for a second or two, then 

 crushed and rejected it, and crushed well and rejected the Longicorn beetle, 

 Gymatura hifasciata. She then suddenly noticed a Mutilla that was climbing 

 uj) the side of the cage and leaning over seized it, the ant making no attempt to 

 escape, crushed it — its strong smell reached my nostrils two or three feet away 

 as she did so, — and thi-ew it away. 



Fifteen minutes later she crushed and ate Papilio demodoctcs, tasted and rejected 

 the Cymattbra, refused, then merely took and threw aside Hypolimnas ivahlbergi. 



Five minutes later she tasted and hesitatingly dropped H. wahlbergi, refused, 

 then tasted perfunctorily and rejected a Mycalesis campina, crushed and swallowed 

 a Papilio angolanus, the H. ivahlbergi, and a Catachrysojjs dolorosa, refused to touch 

 a Mycalesis campina, refused, then tasted unwillingly and rejected a Papilio 

 angolanus, subsequently refusing obstinately to taste it again, crushed well and 

 doubtfully and finally ate the Gymatura, and once more persistently refused to 

 have anything to do with Papilio angolanus. 



[Summary : — 



(1) If we take manner freely into account we have the order : — 



1 . G. peculiaris, C. parsimon, A . basuia 6 , P. titgela, P. artaxia, 



H, da^dalus, G. cloantha, G. achcemenes. 



2. P. demodocus. Gymattira bifasciata not above it, 



3. P. angolanus. 



4. H. wahlbei'gi {anight have gone higher, but for special 



suspicions). 



5. M. campina, Mutilla sp. 



(2) Against the above suggestions concerning //. ivahlbergi we have the following 

 note, made at the time of the experiment : — 



" The bird had thus accepted and eaten the Hyjjolimnas ivahlbergi exactly at the 

 Papilio angolanus point — at the moment when she could just bring herself to eat a 

 single Papilio angolanus, — but not a second. That is to say, the Nymphaline was 

 placed on a level with Papilio angolanus. I doubt whether this result was due to 

 an}' appreciable extent to the latter's resemblance to Amauris domi7iicanus (it had 

 a hindwing attached); it seems more likely (and accords with the fact that the 

 roller's tastings of it were characterized by hesitation) that though thoroughly 

 palatable this butterfly is slightly unwholesome and that the roller retained a 

 memory of its effects on itself on previous occasions."' 



I also recorded that I regarded G. peculiaris as having been hardly sufficiently 

 tasted for a sure result. 



Exp. 120. — December 12. I fed the bird on grasshoppers till she began to accept 

 and eat them with distinct disinclination. She then accepted a brightly coloured 

 bug, Gryptacrus comes, crushed it well, shaking her head greatly all the time, 

 though with long sweeps, as though to get her nostrils clear of the strong 



