208 THE AMERICAN NATUBAZJST [Vol.L 



might become joined with a piece of the homologous chro- 

 mosome instead of with a piece of the same chromosome. 

 But on the fragmentation theory it must be supposed that 

 the fragments reunite in exactly the original order, and, 

 further, that the two homologous chromosomes break at 

 precisely the same point before interchanging— otherwise 

 one reformed chromosome would lack certain factors and 

 the other would have too many ; nevertheless, this point 

 can not be a fixed point, as interchange may occur any- 

 where. Since interchange, when it occurs, usually takes 

 place at one point only, it must also be assumed that the 

 frequency of the recombination just described is so nicely 

 reg-ulated that in about half of the cases it has happened 

 just once (and at one point in the chromosome) during 

 the sum total of resting periods of all cells ancestral to 

 any particular egg cell that shows interchange. For in 

 about half the eggs a particular chromosome has ex- 

 changed in only two sections, and in very few have there 

 been more than three points of interchange. Moreover, 

 in the ancestry of the rest of the eggs, no interchange 

 whatever can have occurred. Finally, the fallacy of the 

 fragmentation idea becomes obvious when we consider 

 that if interchange took place in the resting period of an 

 embryonic cell, most of the eggs derived from this cell 

 would show that particular recombination, and hence the 

 individual in question would give an unusually large pro- 

 portion of offspring of this sort. Thus different individ- 

 uals of the same strain would differ greatly in their link- 

 age values, there being scarcely any constancy at all. 

 Since this is not true it would have to be assumed that 

 interchange takes place only a short time before the mat- 

 uration divisions, owing to some peculiarity in the chro- 

 mosome processes occurring in the cells at this period. 

 Thus we return again to the conclusion that interchange 

 occurs during synapsis. 



Further evidence that interchange occurs during syn- 

 apsis is to be found in some results obtained with Bridges' 

 "non-disjunctional" flies. Non-dis junctional females of 



