94 



THE AMERICAN NATURALIST 



[Vol. XLV 



is that the mean potential (that is to say, congenital) 

 value of every character is about the same for the two 

 lots. I fully realize that the study of genetic problems by 

 the use of mass averages has recently received a decided 

 set-back, largely through the labors of some of those who 

 have contributed to our present program. But until 

 some one is ingenious enough to produce a strain of par- 

 thenogenetic or self-fertilizing mice, I fear that my only 

 practical method of procedure in these experiments is to 

 deal with mass statistics based upon "heterozygous" 

 stock. 



It must also be pointed out that the technique of the 

 problem which I am discussing i s inevitably different 

 from that involved in the endeavor to find, or to produce, 

 "mutations" or single abrupt deviations from the parent 

 stock, which appear at once in full force, if they appear 

 at all, and thereafter breed true. On the contrary, the 

 distribution of the lengths for the tail, ear and foot, 

 within each of the temperature groups in my experi- 

 ments, appears to follow the normal probability curve, 

 just as in the case of the so-called "fluctuating varia- 

 tions," whose heritability is nowadays so much in 

 question. 



in the splendid paper of Professor Johannsen, to which 

 we listened yesterday, occurs the following statement: 

 "as yet no experiment with genotypically homogeneous 

 cultures has given any evidence for the Lamarckian view, 

 the most extreme 'transmission '-conception ever issued." 

 Leaving aside for the time being the question whether 

 results such as mine, even when every possible defect of 

 technique has been eliminated, are to be regarded as 

 "evidence for the Lamarckian view," let us consider for 

 a moment whether the fact that T have not mvself found 



between results from pure and 



