562 The American Naturalist. [July, 
There is naturally a diversity of opinion as to how far each of 
these principles is operative ; not that they conflict. 
2. If both principles operate upon the origin of the fittest we 
should find in every individual two classes of variation, both in 
respect to new characters and to modifications of the old : — First, 
chance variations, or those which, with Darwin and Weissmann, 
I attribute to the mixture of two diverse hereditary strains. 
These may or may not be useful ; if useful they depend entirely 
upon selection for their preservation. Second, variations which 
follow from their incipient stages a certain definite direction 
towards adaptation. These are not useful at the start ; thus while, 
as they accumulate, they favor the individual, they are not 
directly dependent upon selection for their preservation. These 
I attribute to the Lamarckian principle. 
My present purpose is to show that variations of the second 
class are of an extent and importance not suspected previous to 
our recent palseontological discoveries, and that the Lamarckian 
principle offers the only adequate explanation for them, 
3. The general theory as to the introduction and transmission 
of variations of the second class may be stated as based upon 
the data of palaeontology — the evolution of the skeleton and 
teeth. 
In the life of the individual, adaptation is increased by local 
and general metatrophic changes, of necessity correlated, which 
take place most rapidly in the regions of least perfect adaptation, 
since here the reactions are greatest. The main trend of varia- 
tion is determined not by the transmission of the full adaptive 
modifications themselves, as Lamarck supposed, but of the dis- 
position to adaptive atrophy or hypertrophy at certain points. 
. The variations thus arising are accumulated by the selection of 
the individuals in which they are most marked, and by the ex- 
tinction of inadaptive natural groups. Selection, in so far as it 
affects these variations, is not of single characters, but of the 
ensemble of characters. 
The evidence is of a direct and indirect character. The direct 
evidence is that by actual observation in complete palaeontological 
series, the origin of adaptive structures is found to conform 
