316 Dr. T. A. Chapman on the Evolution of the 



(euphemus, areas, alcon), but, as they hibernate as half- 

 grown larvae, the probability that they have the usual 

 number of moults and hibernate in the third instar is very 

 great. Nor, even should they deviate from this as regards 

 one moult, as one or two Lycaenines occasionally do in one 

 or other of these particulars, they equally form a base from 

 which we may believe arion started in developing its ex- 

 traordinary habits. In that case we may with considerable 

 certainty assume that at first arion was carried into the 

 nest in the third instar, and there hibernating initiated 

 the remarkable indifference the ants show towards the 

 larva, which would then be quite harmless, though they are 

 now enemies that one would expect to be dreaded and 

 combated. 



It is probable that some of the larvae so carried in chose 

 to take another moult before hibernating, as occurs in 

 A. pheretes (Trans. Ent. Soc. 1912, p. 403), and, feeding 

 in the interval, their cannibal tastes would enable them to 

 get on by eating the larvae of the ant. Once this habit was 

 acquired, further progress would in a sense be simpler, 

 though there would remain the difficulty of two moults 

 to be made in the ants' nest, and contentment with the 

 diet of ant larvae in the spring to be acquired and fixed. 

 If we assume an occasional variation in the instar for 

 hibernation from the third to the fourth, accompanied 

 frequently with an actual diminution of the size of the 

 larva in these instars, we make a further step. It is very 

 likely that the variation in the instar of hibernation from 

 the third to the fourth is really due to some privation of the 

 larva, that made it badly grown and nourished in the third 

 instar and ill-equipped for hibernation. My experiments 

 on the larvae of Agrotis comes * and others showed that 

 starvation often led to an extra moult ; here, of course, the 

 moult is not an extra one, but an antedating that would be 

 of similar effect. 



These larvae in the fourth instar would not be much larger 

 than the usual third-stage larva, and would be equally 

 within the power of the ants to carry. We do not know 

 whether the ants would prefer a larger or smaller larva, 

 but probably the latter, as that has been the direction 

 in which arion has varied. A tendency to vary towards 

 diminished size would also be involved in the causes leading 



* E. M. M., vol. xxxii (1896), pp. 54, 80. 



