328 CARL HALLQVIST 

The result is quite normal with regard to the appearance of expec- 
ted phenotypes as, contrary to the results obtained in cross 8, a dilution 
type of the fully coloured parent type (bluish red) originates. The 
numerical relations, however, do not agree well with the ratio 9:3: 
3:1. The discordance is still greater in F;, which gives the following 
di-hybrid segregation numbers. 
277 blue : 121ting.blue:102 bl. red : 7 ting. red 
expected 
acc. to 9:3:3:1 285,s - 95,1 95,1 31,7 
D/M 8,3/11,2 = 0,74 25,9/8,8 = 2,94 6,9/8,8 =0,78 24,7/5,5=4,49 
The deficit in the tinged red group is remarkably great, and the devia- 
tion does not lie within the allowed range. The deviations become 
more appreciated when the di-hybrid segregations in F, and F, are 
summed up: 
354 blue : 153 ting. blue: 143 bl. red : 13 ting. red 
expected 
acc. 10.3:3:3:1 372,3 124,3 124,3 41,4 
D/M  18,9/12,8—1,48 28,7/10,0 = 2,87 18,7/10,o—1,87 28,4/6,1 = 4,66 
There is a deficit in the blue and the tinged red groups, while 
there are too many in the tinged blue and in the bluish red groups. 
It is, therefore, very probable that we here have another case of re- 
pulsion, even if an ordinary di-hybrid Mendelian segregation with 
irregular segregation numbers would explain the matter. The great 
deficit in the recessives should then be traced back to poor vitality, 
although nothing else favours this assumption. Unfortunately, the 
analysis has not been carried out on a scale sufficient to guarantee a 
definite decision. The relatively small number of F,-families due to 
diseases, which happened to damage the F,-plots of this cross more 
than other plots, is particularly to be deplored. Otherwise the question 
could have been easily settled especially through the occurrence of 
families showing segregation characteristic of »coupling». It is true, 
one family has probably to be referred to this type, viz. no. 287—19, 
but the small number of individuals precludes a decision. The results 
of the segregation are by no means conclusive but some of the facts 
agree well with the assumption of the existence of repulsion in this 
cross, and this is further strengthened by the fact that each of the 
factors in question are coupled with a third to be discussed in the 
following. 
The degree of the linkage is seen from table 4. A scrutiny of 
the reliability of these figures — in view of the relation between the 
