274 Darwin, and after Darwin. 
involves a grave error of reasoning to suppose that 
this question can be answered deductively from the 
theory of natural selection itself, as I shall show at 
some length in the next volume. 
A still more extravagant, and a still more un- 
accountable fallacy is the one which represents it as 
following deductively from the theory of natural 
selection itself, that all Aereditary characters are 
“necessarily” due to natural selection. In other 
words, not only all adaptive, but likewise all non- 
adaptive hereditary characters, it is said, sst be due 
to natural selection. For non-adaptive characters are 
taken to be due to “correlation of growth,” in con- 
nexion with some of the adaptive ones—natural 
selection being thus the zzdzrect means of producing 
the former zerever they may occur, on account of its 
being the direct and the ox/y means of producing the 
latter. Thus it is deduced from the theory of natural 
selection itself—tist, that the principle of natural 
selection is the only possible cause of adaptive modifi- 
cation: 2nd, that non-adaptive modifications can only 
occur in the race as correlated appendages to the 
adaptive: 3rd, that, consequently, natural selection is 
the only possible cause of modification, whether 
adaptive or non-adaptive. Here again, therefore, we 
must observe that none of these sweeping general- 
izations can possibly be justified by deductive reasoning 
from the theory of natural selection itself. Any attempt 
at such deductive reasoning must necessarily end in 
circular reasoning, as I shall likewise show in the 
the following paragraphs, therefore, ‘“‘ adaptations,” “‘ adaptive modifica- 
tions,” &c., refer exclusively to such as are hereditary, i.e. phyletic. 
