294 THE AMERICAN NATURALIST [Vol. L 



point). It follows from this that interference must in 

 any case be due to one of the following three general 

 causes : (1) Either the chromosomes are not likely to bend 

 across each other twice at points near together (i. e., the 

 loop tends to be long), or (2) breakage at one point for 

 some reason interferes with another breakage nearby 

 (even though the threads are crossed at both of these 

 points), or (3) fusion of chromosomes at one point in 

 some way interferes with fusion of threads which are 

 crossed in a neighboring region. That fusion at one point 

 could interfere with fusion at another point can scarcely 

 be imagined. And if crossing-over occurs according to 

 scheme number three, the ' 'loop explanation" must also 

 be thrown out. Consequently, if crossing-over occurs at 

 a stage of tight twisting the breakage of the threads at one 

 point must somehow be considered to prevent another 

 break near by. In explanation of this, breakage might be 

 thought of as resulting from the tightness of the twisting, 

 for then a breakage of the threads at one point would re- 

 lieve the tension of the filaments for some distance along 

 the line and so tend to prevent another breakage from 

 occurring near by. (Later, when threads reunited at the 

 point of breakage, pieces from homologous chromosomes 

 would be as apt, or more apt, to lie end to end, and there- 

 fore to join, than pieces of the same chromosome. As a 

 partial explanation of why the fragments should join again 

 at all, it might be supposed that only the chromonemas 

 break, the fused sheath which envelops the pair still hold- 

 ing the pieces together.) 



It is fully realized that the above discussion is highly 

 speculative. It is intended, however, not as a presenta- 

 tion of conclusions, but as a tentative suggestion of pos- 

 sibilities, in order to obtain some system of ideas that may 

 furnish a temporary basis for a real attack— experimental 

 and observational— upon the subject. 



Tests for These Alternatives 

 Is there any way of obtaining evidence as to which of 

 these three schemes of crossing-over is the more probable 



