534 



THE AMERICAN NATURALIST [Vol. XLYIII 



case of the second chromosome, where the total percentage 

 of crossing over is about 32, there is probably one or two 

 per cent, of double crossing over. The diagram of the 

 second chromosome is in this respect only tentative, and 

 the plotted position of cretin will be moved a little farther 

 to the right when the amount of double crossing over 

 between light and cretin has been found. The value 4.4 

 for sterile anther light axil is not affected by double 

 crossing over, since the section of chromosome between 

 these two loci is so short that a double break would prob- 

 ably not occur between them at all. The amount of double 

 crossing over between any two loci can only be found 

 when there is a gene between them. Thus if a gene 

 should be found which lies between light and cretin, either 

 by indirect calculation or, better, by direct experiment, 

 the amount of double crossing over could be found. The 

 more genes which can be worked with in the same chromo- 

 some, the more accurate becomes the diagram. 



All the values found for these cases in sweet peas and 

 primula are based upon such small numbers that they 

 can be used only as illustrations of the way in which one 

 would apply to new cases certain principles worked out in 

 Drosophila. While they serve as examples in line with 

 these principles, they are entirely inadequate as proof. 

 A very interesting case of variation in linkage is pre- 

 sented by some of the families involving chromosome II 

 of the sweet pea. In this article I have avoided such data 

 as far as I could, but it is possible that the order in which 

 I have aligned these genes will be found to be incorrect 

 when data upon all three genes in a back cross are ob- 

 tained. Such data would show, through the phenomenon 

 of double crossing over, what the order of genes is, even 

 though variations in the linkage should occur. 



