310 Dr. T. A. Chapman's Observations completing an 



of the last larva, I examined the midden materials and 

 found therein the cast skins of the ants, shrivelled and a 

 little dirty, but otherwise sound and intact. One of these 

 skins is shown in photograph on Plate XLIX, fig. 2. 



The last excretum of avion (only one) was found on the 

 day preceding its death. I feel no doubt that these were 

 really the excreta of the butterfly larva. The ants did 

 not feed the larva, and its rapid growth could not possibly 

 be accounted for, if it fed on the ants' droppings, which 

 were, however, always plentiful enough in the midden, as 

 small oval pellets usually black but sometimes pale and 

 less than half the diameter of the arion pellets. In the 

 midden section of the nest the plaster was disfigured by 

 many small black spots, apparently excreta of some sort 

 of the ants, but the living portion of the nest was almost 

 free from any such disfigurement. By making errors in 

 keeping the nest properly moist or dry, one forced the ants 

 to change their residence from one cell to another, and at 

 the same time confused them as to which place was properly 

 the midden. 



On Plate XLVIII I have placed photographs of portions 

 of slides of the avion faeces, which show the very largest 

 portion of ant larval skin I could find ; usually the portions 

 are small, so as onlv to have a few hairs in each, much as 

 shown on Plate XLVII. 



Those larvae that died and whose interiors I examined 

 had, practically, no intestinal contents. The larva taken 

 in spring had the intestine rather loaded, which led me to 

 believe that it voided no excreta whilst living with the 

 ants. This is obviously not the case with the larva in 

 autumn, though it continues probable that it is so in 

 spring. 



Both my larvae and Mr. Donisthorpe's fed up in four or 

 five weeks to a length of from 7 to 9 mm. If this be com- 

 pared with the size of the larva in May, the difference is 

 not very great, and the extra growth was probably all 

 made in spring. At any rate from September to May the 

 larva grows comparatively slowly, and probably is quiescent 

 for most of the time. Whatever was the immediate cause 

 of death of my larvae, it seems not unlikely that the true 

 one was that the plaster nests afforded no proper shelter 

 for this purpose. Against this is the fact that my last 

 larva fed fairly freely up to a day or two before its death. 



The facts here reported give us in any case a fairly com- 



