Ernest Warren 
325 
dominant parents and heterozygous dominant offspring including self-fertilized 
families was 47 for male parent and -4:1 for female parent. These figures are not 
far removed from Prof. Pearson's mean coefficient -45. 
(7) The remaining two characters dealt with, viz. the percentage area covered 
by the spots and the relative width of the corolla, exhibited no discontinuity in their 
inheritance, and consequently a Mendelian relationship could not be detected. 
The parental coefficients of correlation for the spotting, including and excluding the 
self-fertilized families, were 43 and -32 respectively, and for the width 'M and -30. 
The mean of all four coefficients is -35 which has to be compared with Prof. Karl 
Pearson's mean coefficient of 45 derived from various sources. Owing to the 
fewness in the number of parents the probable errors of the foxglove coefficients 
of correlation are large, but nevertheless there appears to be a distinct tendency 
for the parental correlation to be low. Similarly Prof. Karl Pearson found parental 
inheritance to be low in the Shirley Poppy: the mean value being -33. 
(8) The mean of the seven mid-parental correlations, calculated for the crown 
character, the spots, the general colouration and width of the corolla, is -57 and 
the mean of the six parental correlations is -38. 
(9) The experiment at the present stage cannot throw any direct light on the 
pure-line theory. The variability which occurred among cross-fertilized families 
amounted to 67 % of the variability of the race, and among several self-fertilized 
families it was 50 %. With such a high variability it would appear to be unlikely 
that the families, raised by self-fertilizing the two brethren at the two ends of the 
range of variation of the families, would exhibit the same mean. 
(10) The resemblance of the individuals of a family to one another with 
reference to several characters was calculated, and the mean of these fraternal 
correlations was found to be 47. Prof. Karl Pearson gives a mean value of -5 
or a little more for the fraternal correlation of a series of diverse characters. 
VI. Theoretical Conclusions. 
We thus see that among the characters examined we find typical Mendelian 
and typical non-Mendelian results, and the relationship between the two types 
of inheritance is by no means clear. It has been said that Mendelian inheritance 
occurs in the case of characters which do not mix, but this view is not warranted 
by the present results. For example, the peloric character can be diluted by crossing 
with the ordinary type, but such dilution does not usually take place; also, the 
general purple colouration of the corolla is diluted when a dark parent is crossed 
with a pale parent ; but, nevertheless, a very pale general colouration is exceedingly 
rare, and the flower is as a rule either distinctly purple (although perhaps rather 
pale) or white. Thus we have discontinuity even when characters are capable 
of dilution. 
From the experimental data which have been accumulated during the last 
decade there is no doubt that segregation of characters may occur in crossing 
