136 Mr. E. B. Poulton. Colour-relation bettceen the [Feb. 4, 



taken in 1884 in which the larvae of Smerinthus ocellatus were fed 

 upon various food-plants, and the resulting larval colours were care- 

 fully compared. I also described the colours of captured larvae of the 

 same species, and mentioned the trees upon which they had been found. 

 I was extremely anxious to continue the investigation in the following 

 year (1885) in ^rder chiefly to throw further light upon the two 

 following points : — 



(1.) By means of experiments and observations I had been enabled 

 to show that the colour of the larva was in most cases affected by the 

 food-plant, but there remained a certain number of exceptions which 

 had to be accounted for. I suggested that these might be explained 

 by supposing that the tendencies towards certain colours which were 

 produced by particular food-plants in one generation became inde- 

 pendent larval tendencies in the next, which might modify or over- 

 c< me the usual effects of the food-plants ; and that such transmitted 

 influences augmented as the number of generations upon one food- 

 plant (or others producing similar effects) increased. I was desirous 

 of testing this theory by breeding from moths of which the history in 

 the larval state was accurately known. 



(2.) In my last paper I also pointed out that there was considerable 

 evidence for believing that the influence of the food upon the larval 

 colour is not a comparatively simple phytophagic influence, but one 

 which is exceedingly complex, being brought about by the colour of 

 part of the leaf (the part which the larva resembles), acting as a 

 stimulus to some larval sensory surface (presumably the ocelli) and so 

 through the nervous system regulating the amounts and kinds of the 

 vegetal pigments absorbed and made use of, and that of the larval 

 pigments deposited. 



I wished to test this theory by feeding the larvae under such con- 

 ditions that they could only be affected by the colour from one side 

 of the leaves of their food- plant, and it seemed that the best way of 

 achieving this object was by sewing the leaves together. In the sub- 

 sequent experiments the edges of the leaves were at first pared with 

 the scissors to make them correspond, but it was afterwards found 

 better to double each single leaf longitudinal and sew together the 

 opposite margins which were thus brought into contact. The inten- 

 tion of the experiment was to compare the larvae which had been 

 exposed to the colour of the under sides of the leaves only, with those 

 which had been exposed to the upper sides only, and with those which 

 had been fed upon the normal leaves. If the larval colours varied 

 according to these three sets of conditions, it would be quite clear 

 that the larvae were only influenced by the colour of the leaf-surface, 

 because the leaf-substance eaten (from the edge through its whole 

 thickness) must have been identical in all three cases. I also wished 

 tJ vary the experiment by feeding some larvae upon leaves which had 



