H. E. Crampton 
129 
some respects from those which do not, and that the former are on the whole 
less variable. Consequently we are justified in concluding that elimination 
proceeds upon a selective basis ; for although we must admit that some of the 
eliminated, killed by " chance," may have been just as " fit " as the survivors, 
yet it is not possible that all of them were. Certainly, all of the survivors 
were " fit." Wherefore the observed differences between any two comparable 
measurements of the two groups afford tangible determinations of the instability 
of organization with which unfitness is associated. Natural selection is thus 
found to exist in the material under consideration. Two questions, however, 
immediately arise: How far is the process natural? and: In what respects is it 
selection ? 
The conditions for pupal elimination were certainly natural in every respect, 
and it is equally certain that the circumstances attending pupal-imaginal elimina- 
tion were not the same as those which the pupa meets in nature at the time of 
metamorphosis. For the pupae in the second case were kept in a cabinet in 
the laboratory for some weeks and were subjected to warmer and drier air than 
usual. And yet elimination and selection appear at the second period no less 
clearly than during pupal existence. As regards the relation which secondary 
selection bears to primary elimination, we have seen that the two sexes differ in 
an interesting way. In the male sex the direction taken by selection is the same 
in both cases. Therefore we may conclude, if the figures relating to the first 
period be accredited, that the selective agencies were the same during both 
periods ; or to put the matter in another way, we may say that the same physical 
characteristics of the pupae were tested in the two periods. But in the female 
sex it was found that, on the whole, secondary selection took place in the reverse 
direction, and we must conclude either that the conditions which exercise a 
restrictive effect at the time of metamorphosis are different from those which 
operate earlier, that in other words the organic test refers to other pupal functions, 
or else that the unnatural conditions of the laboratory produce a different result 
from that which would have occurred in nature. The facts are so different in 
the two sexes, that a direct investigation of this important problem should, and 
will, be made. Nevertheless, the fact of primary importance is, not that selection 
is here natural in the sense that it would ham occurred in nature, hut that the 
reduction in numbers proceeds hand in hand with a restriction in certain structural 
characteristics as regards type and variability. 
The next point to be noted is, that selection, ivhere it occurs, cannot be based 
directly upon the characters of the pupa luhich tue have described as showing 
selection. The pupa does not " use " its antennae, and no conceivable advantage 
can accrue to the pupa from a longer, wider, and stouter organ, and yet the 
survivors have on the whole longer, wider, and stouter antennae. Again, it is 
difficult to conceive how particular dimensions of the bust can serve advan- 
tageously or disadvantageously in the struggle for existence during the pupal 
period. What, then, is the actual basis for eliminative selection ? 
Biometrika iii 17 
