Organic Selection 141 



horns afterwards develop, and in certain small changes in 

 the evolution of mammalian teeth — and afterwards progress 

 regularly from one generation to another until they become 

 of some utility. While it is not clear that organic selec- 

 tion completely accounts for these cases, yet it is quite pos- 

 sible that it aids us in the matter ; for the assumption is 

 admissible that in their small beginnings these characters 

 were correlated with useful functions or variations, which, 

 by the operation of organic and natural selection in a 

 progressive way secured the survival and accumulation of 

 the former. Indeed, it is part of the imperfection of the 

 paleontological record that the evidence of such correlations 

 would not be preserved — say, for example, muscular ad- 

 justments such as those which Weismann cites as illustrat- 

 ing intra-selection. It is possible that the development of 

 muscular adjustment and strength compensated for the 

 wearing-off of the teeth both in individual development and 

 in evolution — as is supposed elsewhere,^ — although the 

 fossil teeth taken alone would give no inkling of it. 



The laws of organic correlation are so little known, 

 while yet the correlation itself is so universal,^ that no dog- 

 matism is justified on either side ; the less perhaps on the 

 side of the paleontologists who assert that these cases can- 

 not be explained by natural selection even when supple- 

 mented by organic selection; for when we inquire into 

 the state of the evidence for the so-called 'determinate 

 variations ' which are supposed in these cases, we find that 

 it is very precarious.^ 



1 Chap. XIV. § 3, and Appendix A, note to the quotation from Osborn. 



2 Instances of it are cited in Chap. XIV., below. 



' The only way to establish * determinate variations ' would be to examine 

 all the individuals of a given generation in respect to a given quality, and 

 compare their mean with the mean of their parents — not with the mean of all 



