353 



it worth while. Thus an interesting and convincing exhibit 

 might be made of the contents of, say, a dozen pellets, repre- 

 senting a dozen species of birds. In each case Coleopterous 

 remains might be placed in one row, Orthopterous in another, 

 butterfly-remains in another, and so on, and beside each might 

 be placed the complete insects, set, to indicate the relative 

 amount of disguise that had taken place. 



" While writing the above it struck me to hunt up a pellet 

 of Coradas gamilus L., brought up during an experiment on 

 April 2nd of last year. I see the envelope is undated, but my 

 recollection with regard to it seems pretty clear, apart from 

 the fact that this was, I feel sure, the only pellet belonging 

 to this particular roller (' C ') that I have kept. I thought, 

 I remember, that it might be interesting to examine it some 

 day, but am sending it to you intact, as I think that, if the 

 butterfly-remains in it are at all typical of what I have usually 

 seen, you will find it more interesting and convincing to have 

 broken it up yourself. Before doing this (breaking up) I find 

 it best to let the pellet absorb water freely. The small wing 

 fragments are then most easily detected by placing a small 

 fragment of the pellet at a time in a few drops of water ; many 

 of them float. The appearance of the wing-veins is often in- 

 teresting — mere fragments of rods with little or. often, no mem- 

 brane attached. 



" I am also sending you a pellet of Bu corax caff er. I have 

 no idea what it should contain, apart from what can be seen 

 on the outside, but as it forms one of a number collected with 

 a view to throwing light on the food of my unconfined ground- 

 hornbills, either its contents or a note of them should be kept. 

 I am sending it because I see that it contains fairly typical 

 remains of Coleóptera, and 1 think you will like to compare 

 these with those of the butterflies, grasshoppers, and flies con- 

 tained in the other pellets sent. 



" I had nearly forgotten to say that the roller ' C ' had eaten, 

 just before bringing up the pellet referred to above, several 

 grasshoppers, one Mclanitis leda F., two Hypanartia schœneia 

 Trim., one Precis antilope Feisth., one Precis artaxia, one Charaxes 

 zoolina Doubl. -Hew., one Tagiades flesus F., one Rhopaloca))ipta 

 jorestan Cram., one R. lihaon Druce, one Papilio nircus L. w 



45 



