202 The American Naturalist. [ March, 
(2) Some plus- or minus-variations must also originate from 
the union of germ-cells. If the same character is strongly 
developed in both parents, it may appear still more strongly 
developed in the offspring; the same rule applies conversely to 
weakly developed characters. But this simply puts the question 
one stage back, for variations which are indifferently plus, minus, 
or mediocre are certainly not definite, although the union of two 
similar variations produces a definite result. 
Before considering the possible origin of definite variations we 
must consider whether there are such variations. 
2. What Variations are Definite and What Indefinite ? *—This 
is really the most important and central question. Its solution 
has a vital bearing upon Weismann’s principle as well as La- 
marck’s. Following Huxley,” Geddes * has most clearly stated 
these bearings : 
“In the absence of any theory of definite and progressive 
change,” -and in the presence of multitudinous variations under 
domestication and in nature which we can neither analyze, ration- 
alize, nor hardly even classify, we are not only justified but 
logically compelled to regard variation as spontaneous or indefi- 
nite,—7. e., practically indeterminate in direction, and ‘therefore 
unimportant, except as the groundwork for Selection to act on.’ 
Conversely, variation must be indefinite, else the paramount im- 
portance of natural selection must be proportionally impaired as 
this becomes definite. . . . It would exchange its former 
supremacy as the supposed determinant among the indefinite 
possibilities of structure and function for that of simply accelerat- 
ing, retarding, or terminating the process of otherwise determined 
change.” 
We cannot emphasise too strongly these cardinal factors of 
indefinite Variation (so far as adaptation is concerned) and 
paramount Selection as two of the foundation stones of Weis- 
Scag Selection trusts to the chapter of accidents in the matter of variation.” 
7 Article '‘ Evolution,” Enc. Brit., Vol. VIII, 
% Article ‘ Variation,” Enc. Brit., Vol. XXIV. 
*9 Such as has been postulated ee ren Mivart, or based upon the La- 
_marckian principle by Spencer, Cope, and others. 
