





FLOWER AND SEED COLOUR IN LUPINUS 329 
TABLE 4. ' 
1 ET > DT Leen CO elie ey ° © an, 
be | Frequency of pheno- © _ 2 ISLE) 3 = 5 | 
|- 28 types So eu: He |. € we S 
| 5.2 | Genera- bn ee ss ten bP 1-3 > 2.2 Standard | 2 — 5 | „© 3% 
2 € tion | | | = 5328 error | 8 BE 3 ER 
BE [VEDur | Wir) 5 | eee M'1#88l 8. | BE | 
PBR S| SEN TARDE SAP ed he |e oe 
| 
Repulsion. 
1 F, 77; 32} 41| 61156 |: Oo | -0,1878 | |, Ojazse | 1:1,8 | 35,7: 
2 F, |2491119| 96| 5 1462| 0,8034 | 0,0837 | 0,8072 | 1: 3,6 | 21,7 | 
3 | F,+ F, |3236|151] 137 | 11 | 625] - 0,7046 0,0827 0,7003 |1:2,7 | 27,0 | 
| 
Coupling. | 
4 | F, | 38| 21" 6| 2 | 38} 0,6170 0,3188 036807 2,3 21} 33,3 4 
zygotic groups to be expected in case of di-hybrid numbers in general, 
according to the facts stated on pag. 318 at once shows that the 
numerical relations of nos. 1, 2 and 4 are very doubtful. The inner 
groups are rather unequal in all three cases, and the proportions 
between the outer terms are also irregular. The relations between 
VF — (vF + Vf) and vf are listed in the following and compared to 
the relations to be expected at the numbers secured by the summing 
up of these two quantities. 

1 46 10 expected 7,5235 
A en, » 29,25: 9,75 
RONDES sae » 10,5 
The irregularities of the repulsion values in F, and F, become 
smoothed out when -summed up, and the numerical relations (ref. 
no. 3) representing the sum of the repulsion values of both genera- 
tions agree rather satisfactory with the general values just listed. The 
gametic ratio obtained from the summed up values is 1:2,7 The 
standard error of the association coefficient of the observed values in 
ref. no. 3 is 0,082, which indicates a fluctuation of the gametic ratio 
between 1: 2,3 and 1:3,, or — if the threefold standard error is 
used — between 1:1,s and 1:8. The uncertainty is thus seen to be 
very great. The deviations obtained between the different values of 
the gametic ratio lie within the allowed range. 
The determination of the gametic ratio according to the F,-method 
gives the following result. The number of families of different segre- 
gating types are listed in table 5; they are grouped in the same way 
