324 
Breeding Experiments with Foxgloves 
(5) In plants in which the general internal and external surfaces of the corolla 
are white, the spots which occur may be purple or brown. Sometimes spots were 
nearly absent and the corolla appeared to be practically pure white, but when such 
was the case some sign of brown spots could be found on some few of the flowers. 
An obviously intermediate condition between purple spots and brown spots was 
rare, and judged by ordinary inspection the spots were either purple or brown, 
seldom brownish purple. The brown colouring matter may be regarded as a 
decomposition product of the purple anthocyanin. In plants in which the corolla 
was coloured purple over the general surface the spots were never entirely brown. 
Thus it would appear from the present observations that the causes which lead to 
the total decomposition of the anthocyanin in the purple spots of otherwise white 
flowers are never wholly effective in flowers mth purple corolla. Accordingly 
although there is some independence between the two characters, the colouration 
of the spots and the colouration of the general surface of the corolla, yet never- 
theless there seems to be a certain relationship between the two, since the spots 
were never found to be comjiletely brown unless the general surface of the corolla 
was white. A feature of unknown significance may be mentioned, the cells of 
the spots containing the brown colouring matter were found to be densely 
crowded with typical starch grains, while the cells of the purple spots containing 
the anthocyanin in solution in the cell-sap contained little or no starch. 
Purple spots were dominant and brown spots recessive. 
The observed results in the different families totalled 69 plants with purple 
spots and 23 with brown spots. The Mendelian expectation calculated for each 
family and added together yielded exactly the same figures. Since completely 
brown spots were never found in coloured flowers, only those families in which 
plants with white flowers occurred were included in this calculation. 
(6) With regard to the general colouration of the corolla it was found that 
some of the parents were heterozygous and others homozygous. The results of 
the crossings were distinctly Mendelian in character. Purple was dominant and 
white recessive. There were 41 white and 139 purple offspring; the theoretical 
expectation, calculated for each family and added together, was 49 white and 
131 purple. 
A point of considerable interest was noticed, namely that in dealing with 
parents and offspring of similar gametic constitution (i.e. heterozygous and 
homozygous dominant) the inheritance of the intensity of colouration followed 
the usual statistical laws. A light purple plant crossed with a dark purple plant 
might give some perfectly white ofi:spring, but those offspring that were coloured, 
if of the same gametic character as the parent, tended to be intermediate in shade 
between the two parents. Thus the intensity of colouration is not a mere fluc- 
tuating uninheritable character, but its inheritance follows the usual laws found 
by statisticians to hold in the case of characters where a Mendelian relationship 
cannot be detected. The parental coefficient of correlation between heterozygous 
