220 EN TOMOLOGY 
direct effects of food, light or other primary factors. Such cases, 
then, are ina sense accidental. For example, many inconspic- 
uous green insects are green merely because chlorophyll from 
the food-plant tinges the blood and shows through the skin. 
If it be argued that natural selection has brought about a thin 
and transparent skin, it may be replied that the skin of a green 
caterpillar is by no means exceptional in thinness or trans- 
parency. Moreover, many leaf-mining caterpillars are green, 
simply because their food is green; for, living as they do within 
the tissues of leaves, and surrounded by chlorophyll, their own 
green color is of no advantage, but is merely incidental. 
oe 
Again, in the “ protectively’ colored chrysalides experi- 
mented upon by Poulton, their color was directly influenced 
by the prevailing color of the light that surrounded the larva 
during the last few days before pupation. Of course, it is 
conceivable that natural selection may have preserved such in- 
dividuals as were most responsive to the stimulus of the sur- 
rounding light; nevertheless the fact remains that these resem- 
blances do not demand such an explanation, which is, in other 
words, superfluous. 
Indeed, a great many of the assumed examples of “ protec- 
tive resemblance”’ are very far-fetched. On the other hand, 
when the resemblance is as specific and minutely detailed as it 
is in the Kallima butterflies—where, moreover, special instincts 
are involyed—the phenomenon can scarcely be due to chance; 
the direct and uncombined action of such factors as food or 
light is no longer sufficient to explain the facts—although these 
and other factors are undoubtedly important in a primary, or 
fundamental, way. Here natural selection becomes useful, as 
enabling us to understand how original variations of structure 
and instinct in favorable directions may have been preserved 
and accumulated until an extraordinary degree of adaptation 
has been attained. 
Value of Protective Resemblance.—The popular opinion 
as to the efficiency of protective resemblances is undoubtedly 
an exaggerated one, owing mainly to the false assumption that 
