466 



THE AMERICAN NATURALIST [Vol. XLIV 



a clue as to how the paired characters in the germ cells 

 of the hybrid separate. Prior to this division there are 

 two chromosomes of each kind; one member of each 

 pair being maternal in origin and the other paternal. 

 The members of each pair come together just before 

 maturation division, reducing the number of chromo- 

 somes to half. Later these paired chromosomes divide 

 so that each germ-cell gets one half of each pair. 



Sutton first pointed out in 1903 that if each character 

 that mendelizes is carried by a particular chromosome 

 the mechanism of reduction gives an explanation of the 

 way in which there may come to be two kinds of germ 

 cells with respect to each particular pair of characters. 

 This hypothesis has been championed by Wilson in 

 1903, and later by Boveri in 1904. 1 If we analyze the 

 facts further we find that the hypothesis requires in 

 order that pure gametes are to be formed by the hybrid 

 that each particular character, or whatever it is that 

 produces the character, be confined to a single chromo- 

 some ; otherwise the separation will not be complete and 

 pure gametes will not be formed. Do the facts of reduc- 

 tion fulfil this condition? 



When the reduction in the number of the chromosomes 

 takes place we find that the homologous pairs of chro- 

 mosomes fuse completely, so far as we can judge by our 

 modern methods of technique. Observation gives no 

 evidence in most cases that the chromosomes only ad- 

 here side by side, but on the contrary conjugation ap- 

 pears to be a complete fusion, and if this is what really 

 takes place, what guarantee is there that subsequently 

 the members of a pair will separate along their line of 

 fusion? It seems all the more remarkable that such a 

 process should take place, if, as is often assumed, the 

 separation division is not the first, but the second divi- 

 sion of the paired chromosome. In other words it is ad- 

 mitted that in such cases the first division is at right 



'In 1902 Boveri referred to a possible relation between reduction and 

 inheritance in hybrids, but he did not point out how this idea could be 

 applied to explain the numerical results of Mendelism. 



