50 



A. H. STURTEVANT 



lated from the figures given in the diagram of the chromosome. 

 Table 3 includes all pairs of factors given in table 2 but not used 

 in the preparation of the diagram. 



It will be noticed at once that the long distances, BM, and 

 (C, 0)M, give smaller percent of cross-overs than the calculation 

 calls for. This is a point which was to be expected, and will be 

 discussed later. For the present we may dismiss it with the 

 statement that it is probably due to the occurrence of two breaks 

 in the same chromosome, or 'double crossing over.' But in the 

 case of the shorter distances the correspondence with expectation 

 is perhaps as close as was to be expected with the small numbers 

 that are available. Thus, BP is 3.2 less than BR, the difference 



TABLE 3 



BP 



BR 



BM 



(C, 0)R. 

 (C, 0)M 



OBSERVED PER CENT 

 OF CROSS-OVERS 



32.2 

 35.5 

 37.6 

 33.7 

 45 . 2 



expected being 3.0. (C, 0)R is less than BR by 1.8 instead of by 

 1.0. It has actually been found possible to predict the strength 

 of association between two factors by this method, fair approx- 

 imations having been given for BR and for certain combinations 

 involving factors not treated in this paper, before the crosses 

 were made. 



DOUBLE CROSSING OVER 



On the chiasmatype hypothesis it will sometimes happen, as 

 shown by Dexter ('12) and intimated by Morgan ('11 d) that 

 a section of, say, maternal chromosome will come to have paternal 

 elements at both ends, and perhaps more maternal segments 

 beyond these. Now if this can happen it introduces a com- 

 plication into the results. Thus, if a break occurs between B 

 and P, and another between P and M, then, unless we can follow 

 P also, there will be no evidence of crossing over between B and 



