No. 595] THE MECHANISM OF CROSSING-OVER 429 



been obtained, we must be cautious about accepting the 

 exact values shown in the curve of coincidence; this 

 applies not so much to the main portion of the curve as to 

 the right-hand end (shown in dotted lines), for in the case 

 of very long double cross-overs, very few kinds are even 

 theoretically possible, compared to the number of different 



DISTANCE 



positions in which short double cross-overs of a given 

 length may be found. Accordingly, the marked fall, fol- 

 lowed by great rise at the very end of the first curve has no 

 true significance. 



Certain points may be seen to stand out plainly, how- 

 ever. It is clearly evident that interference is great tor 

 short distances-?, e., that relative coincidence is low; as 

 distance increases the coincidence rises, at first, quickly, 

 but beyond a certain point the rise ceases. 



There is no indication of a usual length of loop of less 

 than half the length of the chromosome, as cytological ob- 

 servations on strepsinema stages would suggest, and as 

 would therefore be expected on the view that crossing-over 

 occurs at that stage. The fall seen near the right hand 

 end is entirely unreliable, as has been explained, tfut, 

 even if taken at its face value, the drop at this point can 

 have no significance for the question at issue, for a tall 

 due to the loop would have to be as long as the whole pre- 



