No. 592] THE MECHANISM OF CROSSING-OVER 211 



as determined by their ' 1 total frequencies of interchange." 

 The length of the first, or sex-linked, group of factors has 

 been found to be about 66 units; a unit, it will be re- 

 called, is a section of the chromosome of such length that 

 breakage occurs within it, on the average, one time in a 

 hundred cases. The evidence then shows that, in a hun- 

 dred cases, the first group breaks 66 times. This does 

 not mean that it breaks in as many as 66 cases out of 100, 

 for it may break two, or, very rarely, even three times, 

 coincidently (at different points along the chromosome) 

 in the same case (" double or triple crossing-over"). As 

 previously explained, when two breaks thus occur coinci- 

 dentally, the extremes of the chromosome come to lie on 

 the same side, and so a factor at one end of the first group 

 does not separate nearly 66 times in a hundred from a 

 factor at the other end; owing to these coincident breaks 

 it really separates in only about 45 per cent, of cases. The 

 number 66 is consequently not obtained by merely deter- 

 mining the frequency of separation from each other of the 

 two most frequently separating factors, but, as mentioned 

 above, it must be derived by adding together the fre- 

 quencies of all the smallest parts of the chain (frequencies 

 of AB + BC + CD, etc.). In the case of the first group, 

 the determination of this - total-length" has been accom- 

 plished bv the combined efforts of a large number of 

 people, although by the work of Morgan, Sturtevant and 

 Bridges particularly. 



Group II has a much greater length. It is probably 

 over a hundred units long, and is certainly over 90. This 

 result has been obtained principally by the work of Bridges 

 and Sturtevant, although, as before, others have helped 

 very materially. Mention must here be made of the fact 

 that Sturtevant has discovered in this group specific 

 mutant factors which, when heterozygous, lower the fre- 

 quency of separation in certain regions of the group very 

 much,* although the order in which the factors are linked 

 is not changed (16). The variation certainly proves, h— 

 ever, that (if the groups represent 11 



the chromosomes) 



