March 1 8, 1886] 



NATURE 



475 



(5) upon Sa'ix vimitialis : white (i) but immature upon S. 

 7'iiiniui/is, sewn to sliow the under sides ; yellowish-intermediate 

 (2) but immature upon ^. alba; whitish-intermediate (4) but 

 mostly immature upon 5'. Sniiihiaiia and similar leaves ; inter- 

 mediate (i), yellowish-intermediate (5^, and yellow with traces 

 of the red spots (i) upon S. ciitLi-ea ; intermediate (l) upon 

 Pofnlus nigra ; intermediate (6) and yellowish-intermediate (2) 

 upon S. triandra ; yellowish-intermediate (10) upon 5'. triandra 

 without the bloom on the under sides of the leaves ; yellowish- 

 intermediate (4) upon 5. bihyloiiica (when the larvce were young 

 and more numerous, one of them in the fourth stage possessed 

 traces of the upper row of rust-coloui'ed spots, and was a whitish 

 variety : it was put upon apple and died as an intermediate 

 variety when advanced in the last stage) ; intermediate (i) and 

 yellowish-intermediate (3) upon ^. rubra. The results of these 

 experiments were mostly what might Iiave been anticipated 

 from the colour of the leaves. The especially interesting results 

 were : the effects produced by the sewn leaves of S. viminalis 

 and the bloomless leaves of 5. triandra as compared with tlie 

 normal leaves in both cases, and the occurrence of red spots on 

 two larva; — one, the only yellow variefy obtained, and the 

 other an intermediate variety. The indications of individual 

 variation in the .5". cin rca larv.x were also interesting. Series IV. 

 All these larva; died before trustwo-'thy observations could be 

 made ; they had been reserved for experiment : the ocelli of 

 many were covered with a harmless opaque layer (lamp-black 

 and McGuilp) in the attempt to isolate the sensory surface which 

 is affected by the colour of the environment. Others were fed 

 upon apple and .S". rubra, and a certain number were changed 

 at every stage, so as to find the periods d-jring which the Iarv;e 

 are most susceptible to the surrounding colour. The failure of 

 all these was due to the season an<l not the conditions of the 

 experiment. StriesV. The female parent was bred from a yellow- 

 ish-intermediate larva, the male parent from a yell iw larva. The 

 resulting larvae were, intermediate (l) upon S. viminalis sewn to 

 show the upper sides of the leaves; intermediate (l) and yellow- 

 ish-intermediate I'l) upon S. cinerea. The first result was inter- 

 esting, but the second shows that the larvce did not tend much 

 towards yellow. The hereditary influence towards yellow in 

 llii^ case depends chiefly on the male parent, and how far this 

 clement asserts itself in opposition to the other se.\ is quite 

 unknown in this class of experiment. 



Results of the Experiments. — The existence of hereditary in- 

 iliience is, on tlie whole, demonstrated. The parent laiv.1; 

 tended towards white, and out of the 75 larvae of the next 

 i^eneration, there was only one yellow variety. Yet the latter 

 were, on the whole, rather more inflaenced in the direction of 

 yellow than the parents, when the plants tended this way. The 

 results were the same as in the parents when very powerful white 

 influence was used (apple, &c.) The comparison of the different 

 series was less satisfactory, but the hereditary differences were 

 mostly delicate, except between V. and the other series. 



Series I. and IIL compare favourably, while in II. the 

 parentage is very obscure. There is conclusive proof that it is 

 the colour of the leaf, and not its substance when eaten which 

 influences the larval colours. 



Conflicting evidence as to the effect of plants is cleared up. 

 6'. triandra produces yellow ; and Mr. Boscher'.-i white larviE, 

 said to have been found upon this plant, occurred upon the very 

 similar, but much whiter, .?. alba. Previous conclusions as to 

 .S. Siiiithiana and 5'. bah)donica are cmfirmed. The existence of 

 individual variation with similarity of parentage and conditions 

 is now proved, although it is rare and slight (8 out of 75 in 1885, 

 o out of 23 in 1SS4). Thus it cannot explain the extreme differ- 

 ences met with in the field (1SS4), e.g. yellow upon apple, &c. 

 The colours of the larvje are determined by (i) the food-plant ; 

 (2) hereditary influence ; (3) individual variation. Slight differ- 

 ences may be caused by the latter ; extreme differences by the 

 two former. The uncertain action of (i) will be shown later. 

 The two red-spotted larvae were very interesting, showing how 

 the character tends to appear on the yellow varieties (the only 

 yellow one produced), and yet that it may appear upon the other 

 varieties (which is a new experience). 



Obseruations in the Field in 18S5. — (1) White variety upon 

 ordinary apple ; (2) white and (i) intermediate upon 6'. vimi- 

 nalis : (l) yellowish-intermediate upon S. line iris (Forbes); 

 (i) yellow upon S. incatia in Switzerland; (l) white upon .J. 

 all'a (ordinary) in Switzerland ; (i) yellowish-intermediate upon 

 var. 5. alb I C* vitellini] in Switzerland ; (2) yellow upon var. .S. 

 alba in Switzerland ; ( i) whitish-Ln'ermediate upon .9. Smithianrii; 



(4) yellow and (i) whitish-intermediate upon .J. cinerea ; (4) 

 yellow upon .9. triandra; (l) yellowish-intermediate upon S. 

 ftziiy'&KzVa ; (12) yellow and (3) yellowish-intermediate upon S. 

 rubra. All, except the five in Switzerland, were found at 

 Oxford. 



Conclusions from Captured Larvte ; Reconciliation of Conflicting 

 Evidence. — The results recorded above were very uniform. Only 

 in the case of 5. cinerea was there any great difference between the 

 larva; on the same tree, for the Swiss varieties of S. alba, which 

 produced yellowish larva;, had leaves resembling .S". rttbra rather 

 than the ordinary English S. alb.i, with one exception (which 

 produced the white larva). At the same time there were two 

 instances which perplexed me for a long time, and finally sug- 

 gested the explanation which clears up the greatest difficulty in 

 the way of the theory^the conflicting evidence (Mr. Boscher's 

 and my own) with regard to the action of .9. viminalis. These 

 instances were, the yellowish intermediate larva upon 6'. linearis 

 and the yellow one up:)n .S". incana. Both the trees had small 

 narrow leaves with very white under sides, and yet the larvje 

 were not white. S. linearis is whiter than apple or any sallow 

 that I have seen. Then I remembered that the single yellow 

 larva I had found upon i". viminalis (in 1S84) was upon a tree 

 with very small leaves ; finally, I had the opportunity of looking 

 at twigs from the trees upon which Mr. Boscher had found 

 about eighteen yellow larv£e. These, too, bore very small 

 leaves, although they were as white as usual on the under sides. 

 This association of yellow larvje with small leaves (although 

 white) suggested the following explanation. It is the immediate 

 environment almost in contact with the larva which has the 

 greatest eftect upon it, and the longer it acts the more extreme 

 will be the result. The larva; habitually rest upon the under sides 

 of the leaves, until their size and weight render it impossible 

 for them to do so longer ; then they retire to the stem. Hence 

 the larva: rest for a much longer period of their lives upon large 

 and strong leaves than upon small ones, and therefore, in the 

 forhier case, the effect of the white under sides is much more 

 powerful, for after the larva has reached the stem the immediate 

 environment is less exclusively white, and maybe largely formed 

 by the green or yellowish upper sides of the leaves (depending 

 upon the arrangeaaent of the latter). This completely explains 

 the conflicting results upon^. viminalis, for there is an immense 

 difference in size between the leaves of the two forms upon 

 which the two varieties of larva have been found ; and the 

 arrangement is also different, tending to produce a white en- 

 vironment after the larva has gone to the stem in the case of the 

 large-leaved trees, but not in the small-leaved forms. This ex- 

 planation is valid for other food-plints in exact proportion to 

 the difference in colour between the two sides of their leaves. 

 Thus it is probable that it explains in a great measure the 

 yellow eftect of the small-leaved S. cinerea, and the much 

 whiter eftect of the large-leaved S. Smithiaua ; both having 

 very similar leaves with white under sides. It also explains the 

 very powerful effect of apple, with its large strong leaves, which 

 are arranged so as to give a maximum white eftect after the 

 larva has retired, very late in life, to the stem. Another difii- 

 culty is also cleared up by this suggestion — the fact that bred 

 larvae in 18S4 and 1S85 became intermediate upon large-leaved 

 5'. vimina'is, although they tended towards white (as shown by 

 the effect of other plants). The long leaves were disarranged 

 when crowded into the glass cylinders in which the larv?e were 

 kept, and so the immediate environment of the larvce was arti- 

 ficially altered when they were on the under sides of the leaves, 

 and also on the stem. Furthermore, they were often disturbed 

 by changing the food, and so did not rest upon the white sur- 

 faces for such long periods as in the natural state. The 

 larvae found in the field upon 5. rubra and plants tend- 

 ing towards yellow were much more extreme varieties 

 than those produced by breeding. This is partly due to 

 the hereditary influence towards white in the latter case, 

 but also probably to the white muslin which was tied 

 over the breeding cylinders, and to the less amount of direct 

 sunlight obtainable indoors and among the crowded leaves. In 

 the field the larvce habitually rest on the most exposed and 

 tallest boughs, of which the colours are most brightened by sun- 

 light, and such an environment therefore produces a very strong 

 influence upon them. Such considerations suggest that it will 

 be very interesting to breed the larva; under coloured light, and 

 I intend to make the experiment this summer. 



Conclusion. — The whole evidence for the theory of colour- 

 relation advanced in the present paper c jnsists of 204 instances, 



