Ernest Warren 
319 
Standard 
Standard 
Coefficient 
No. of 
No. of 
No. of 
devia- 
deviation 
Relationship 
parents 
off- 
flowers 
tion of 
of 
of 
spring 
parents 
offspring 
correlation 
Mid-parent and offspring (in- 
15 
170 
1000 
29-626 
49-220 
•554 
cluding self-fertilized families) 
Mid-parent and offspring (ex- 
11 
127 
774 
25-886 
47-164 
-509 
cluding self -fertilized families) 
Male parent and offspring (in- 
15 
17G 
1000 
38-783 
49-236 
-341 
cluding self-fertilized families) 
Male parent and offspring (ex- 
cluding seK-fertilized families) 
1 ) 
127 
774 
38-898 
47-358 
-300 
The parental correlation is thus low as in the case of the area covered by spots. 
Such a condition increases the probable error of the numerical results, and many 
more parents would be necessary in order to obtain more reliable figures. Never- 
theless it may be noted that the statistical results obtained with the various 
characters are very fairly consistent with one another and the parental inherit- 
ance of the characters investigated tends to be somewhat low. 
III. The Variability of the Race and of the Families. 
According to the pure-line theory the variability within a family consisting 
of individuals of similar gametic constitution and arising by self-fertilization of 
the parent should be small, supposing that no new mutation should arise, and 
it is of such a nature that it is incapable of being transmitted by inheritance. 
Selection within such a family would not be effective. In other words, families 
obtained by self-fertilizing individuals at the two ends of the range of variation 
would exhibit the same mean. The material at present to hand will not throw 
direct light on the question, but indirectly it lends no support to the theory. 
The comparative variabilities of the race and of the individual families are 
given in the table on p. 320. The colour of the different flowers on an axis 
was found to be essentially uniform; but in the two other characters there was 
some variability among the flowers on the same axis. Therefore in these two 
characters (the area covered by spots, and the ratio of the width to the length 
of the corolla) the mean of the six basal flowers of the axis was taken as expressive 
of the nature of the plant. 
Thus the mean variability of a family is 67 % of the variability of the whole 
race. Owing to the fact that tlie majority of these families arose through cross- 
fertilization this large family variability is not a proof against the pure-line 
theory. It may be stated here that with the four self-fertilized families raised 
there was some tendency for a reduction in the variability, but nevertheless it 
remained very considerable. 
The average percentage variability of the self-fertilized families is 50 % of 
the variability of the race. This is to be compared with 67 % for cross-fertilized 
