THE AMERICAN NATURALIST [Vol. XL V 



not much impressed with the consistency of his results. 

 Not only are his means and standard deviations much 

 higher in 1909 than in 1908, but the relative variabilities 

 are also higher in ten out of fourteen cases. To be sure, 

 he has used the conventional precautions. He describes 

 his field as a "gleichmassiger humoser Lehmboden," and 

 states that the fertilizer was mixed with soil before appli- 

 cation and distributed as evenly as possible. But for an 

 organism so responsive to environmental influences as 

 the garden pea, 17 and in a problem of this delicacy, these 

 precautions are not at all sufficient. It is quite clear 18 

 that the seeds from each parent were planted together 

 in rows, and if the soil differed at all from one part of the 

 field to another the tendency would be for this hetero- 

 geneity to induce a differentiation in the crop. 



If it be urged that we do not know that the differentia- 

 tion in Roemer's means are due to environmental hetero- 

 geneity, the reply is simple. It is the duty of those who 

 claim genotypic rank for observed differences to prove 

 that their results are not due merely to faulty experi- 

 mental conditions. 



Third. One or Both of Roemer's "Populations" are 

 in Reality "Pure Lines." 



By definition a pure line is the offspring of a single 

 individual of a continually self -fertilizing or vegetatively 

 propagating organism. They may or may not be differ- 

 entiated. Genotypists assume that generally they are 

 both differentiated and highly constant. Selection within 

 the line is absolutely incapable of effecting any change. 



Now the curious thing about Roemer's material is that 

 at least one, and possibly both, of the two populations 

 suitable for our examination are in reality pure lines. 

 Whether the ' ' Individualauslese " by which the "G-elbe 

 Viktoria Erbse" was developed gave rise to a pure line 

 depends largely upon the stringency with which it was 



"Eoemer observes that the pea is exceedingly susceptible to environ- 

 18 See Eoemer, I. c, pp. 404-405. 



