No. 539] DISTRIBUTION OF PURE LINE MEANS 



697 



characters of the parent plants which furnished the seed 

 from which his various pure lines were grown, but he 

 neither determines whether there is any parental corre- 

 lation in the population nor gives us the data from which 

 this might he determined. To be sure, one may determine 

 from his data that the variability within the individual 

 line is less than that of the population, but this might be 

 as easily attributed to the much-scouted "(lalton's Law" 

 as to genotypic heredity; possibly it is to be referred to 

 some factor quite outside inheritance. 10 In fine, it is 

 impossible to determine anything at all about inheritance 

 from Roemer's data. 



Second. The Possibility of Environmental Hetero- 

 geneity is not Adequately Taken into Account. 



One of the first precautions of the experimental 

 breeder should be to make sure that the environmental 

 conditions to which his materials are subjected are not so 

 heterogeneous as to vitiate entirely any conclusions con- 

 cerning innate factors. Two watch glasses of Paramecin 

 may present highly different environmental conditions. 

 Food and housing are important factors in the egg 

 records of poultry. Peas differ from row to row, or 

 within the row, because of slight differences in the sub- 

 stratum. But experimenters nowadays are so obsessed 

 with the idea of rigid ' ' unit characters, ' ' ' ' determiners ' ' 

 and ''genes" that little attention is paid to environmental 

 influences; they are so absorbed in "analyzing" the 

 "germ plasm " that they forget to make proper allowance 

 for factors which may so modify the soma — which is the 

 only available index of the germ plasm of an individual 

 — that it is misleading as a guide to hereditary tendencies. 



From my own experience in experimental cultures it 

 seems quite possible that the differences in Roemer's lines 

 are due to lack of uniformity in the substratum. One is 



