SEX-LINKED CHARACTERS IN DROSOPHILA. 1 93 



paternal. Thus I and 2 represented the two members of a pair 

 of homologous factors, as did also 3 and 4, 5 and 6, etc., to 35 

 and 36, in every case the odd number representing the maternal, 

 and the following even number the corresponding paternal factor. 



I now secured the ends of the chromosomes so that the proper 

 factors stood opposed to each other and began to twist the 

 chromosomes. 



It was at once apparent that if the twisting should not vary 

 in the members of one species, coupling would be invariable, nor 

 w r ould the coupling be that representing either parent alone, 

 but both, and the combination of factors would be always the 

 same. This is shown in Table IX. In this table the vertical 

 columns represent the gametes formed when a splitting follows 

 the number of tw r ists named at the top of the column. In each 

 case there are two different combinations of gametes; one formed 

 on the left and one on the right side of the split. 



A study of the above table will show that if there be, for 

 instance, no turn and the split falls between the two chromosomes 

 all the factors in the gamete on the right will be maternal. If 

 there be three eighths of a turn, the odd-numbered factors from 

 i to 21 (maternal) and the even-numbered factors from 24 to 36 

 (paternal) will always be in the gamete on the right. And so> 

 in every case, if the amount of twist be constant, the factors 

 present in any gamete w r ill be constant. This will be true even 

 if the factors are of different lengths, and the twisting not uniform 

 for all parts of the same chromosome, if only the lack of uni- 

 formity be constant in all the members of the species. 



Now r , the nature of the twisting and the amount of variation 

 that occurs can only be solved, if at all, by the cytologist. On 

 the face of the question as so far presented it must be that 

 variation in the twist occurs, or there would, on the basis of the 

 theory here discussed, be no such thing as " independent mendeliz- 

 ing," but a constant coupling. My question is therefore, " What 

 are the facts concerning this twisting? How uniform is it? ' 

 For I conceive it to be possible that if the twisting be nearly 

 definite, coupling of certain factors would generally occur, and 

 would depend not entirely on the nearness together of the factors, 

 but on the amount of the twist, and on the side of the splitting 



