THE CAUSE OP MIMETIC RESEMBLA.]S'CE. 575 



the latter are almost invariably dif»tasteful. I exhibited tlie 

 former larvae in the second stage at the Entomological Society ou 

 June 3, 1891. The following account is given in the Proc. Eat. 

 See. 1891, p. XV : — "At this period [viz. the early stages of 

 growth] the larvae arrange themselves in small groups upon the 

 leaves and leaf-stalks of the birch, and when disturbed they 

 raise the anterior part, bending the head over the dorsal surface 

 of the posterior part of the body. In this attitude they strongly 

 remind the observer of those Tenthredo larvae which, when 

 irritated, bend the tail forwards over the anterior part of the 

 body. The fact that the head is raised in the one, and the tail 

 in the other, does not cause any conspicuous difference when the 

 larvae are seen from a little distance. The common Tenthredo 

 larva, Croesus septentrionalis, is about the same size as these 

 small Lepidopterous larvae, feeds in similar small groups when 

 large (w'hen small the groups contain far more individuals), and 

 also frequeuts the birch." In my experience, however, the 

 Croesus feeds much later in the year. Mr. "W. Holland also 

 noted the same resemblance in the ' Entomologist's Eecord ' 

 for Oct. 15th, 1891 (see vol. ii. p. 228). Mr. Holland has also 

 kindly lent me his notes made at the time, and I see that he 

 observed the saw-fly-larva-like movements which follow dis- 

 turbance. The groups have never been figured before, so far as 

 I am aware. In May 1896 we reared some larvae from the 

 Q^g in the Hope Department, and I was able to get some 

 excellent coloured drawings by Mr. P. J. Bayzand. The:se 

 drawings, reproduced on Plate 40. figs. 2 & 3, show the attitude 

 taken up on disturbance better than any description. They also 

 prove that, like the saw-fly larvae, the groups contain far more 

 individuals in the younger stages than later. The conspicuous 

 orange-coloured true legs suggest the appearance of the orange 

 ventral glands of the Croesus, which are everted when the larva 

 is irritated. 



Thus the causes of the resemblance we are discussing may be 

 deep-seated or may be superficial, or, more generally, may be due 

 to several kinds of causes in each category. It is in the latter 

 extremely complex cases, and these are far the commonest, that 

 the argument for natural selection becomes irresistible. This 

 wiU be more thoroughly dealt with in the succeeding Section ; 

 but even in the case of the simplest element in the resemblance, 

 viz. the similarity in colour and pattern taken alone, the theories 



41* 



