CONSTITUTION OF THE CHROMOSOMES 1 47 



series two genes lie, the greater would be their frequency of 

 separation, and consequently their exchange ratio. 



The suggested answer to the questions is then as follows: 

 The exchange ratios are different for the different pairs of 

 characters, because the genes to which these characters are 

 due are at different distances apart on the thread-like chro- 

 mosome. Two genes that are close together give a small ex- 

 change ratio; two genes that are farther apart give a larger 

 exchange ratio. The greater the ratio the farther apart are 

 the genes. These relations therefore yield a series of di- 

 verse ratios, as seen in the list on page 145. 



Is this suggestion borne out by the known facts as to the 

 chromosomes and as to the crossing-over of different genes? 

 Let us examine the facts from this point of view. 



In accordance with this suggestion, the genes on which 

 depend the characters in the list on page 145 would be ar- 

 ranged in series in the order of the ratios which they give 

 with white. If we think of white as at one end of the series, 

 then the genes would be in the order: white, abnormal, 

 bifid, club, vermilion, miniature, rudimentary, bar, as shown 

 in figure 33, II. 



Certain consequences of this idea appear at once, and 

 these may be used for testing whether the idea is adequate. 



I. Since the size of the ratio depends on the distance 

 apart of the genes, two genes that have nearly the same 

 ratio with another must be at about the same distance 

 from it. Thus vermilion, with a ratio of 30.5 from white, 

 must be at nearly the same distance from white as is minia- 

 ture, since the latter has a ratio of 33.2. If white is at the 

 end of the series, therefore, vermilion and miniature must 

 be close together. And if they are close together, they must 

 have with each other a small exchange ratio. This can be 

 tested; it has been done on a large scale. The exchange 

 ratio for vermilion-miniature Is found to be indeed small, 

 about 3.1 per cent. This same relation is found for other 



