CHI ASM AT A AND CROSSING-OVER 281 



Bridges, crossing-over between the homologous dislocated segments 

 will give a reversible inversion. If the segment is short enough for 

 localised gene action to make a difference beyond it (as Muller 

 found), position effect change will be possible at the same time 

 (Ch. VIII). Thus mutation may be possible by secondary structural 

 change in normal diploid Drosophila. 



6. THE PARALLEL INVESTIGATION OF CROSSING-OVER 

 (i) Method. Having found an agreement in all particular cases 

 that are available for study between chiasmata and crossing-over, 

 we can now profitably consider how far their general properties 

 show a parallelism. We can use the one method of study to eke out 

 the other. This is the more important because, although the results 

 are parallel, the techniques are complementary, which is just why 

 it has taken so long to establish a connection between the two. 

 Table 44 shows how the one reveals precisely what the other fails 

 to reveal. 



(ii) Diploids (General). In organisms with chromosomes of 

 uniform size the average frequency of chiasmata in each diplotene 

 bivalent is, so far as we know, usually between one and three. In 

 Primula sinensis it must be greater than two, and is probably less 

 than three. \Vhere large size-variations occur the larger chromo- 

 somes have a higher average frequency or it ceases to be pro- 

 portional to the length. In certain Liliaceae and Leguminosae the 

 chiasma frequency is greater than 3 ; in the Tettigidae it seems 

 to be usually less than two. Further, where pairing is by 

 chiasmata a single chiasma must always be formed to secure 

 regular pairing. 



These observations show definitely the amount of crossing-over 

 to be expected in chromosomes in general and in particular instances, 

 on the chiasmatype hypothesis. The average occurrence of one 

 chiasma must correspond to 50 per cent, of cross-overs in the length 

 of chromosome in which it occurs, i.e., to 50 corrected units of 

 crossing-over distance. Therefore the lengths of chromosomes 

 completely measured in cross-over units should be greater than 50, 

 and they should frequently be over 100. 



