MATROCLINIC INHERITANCE 
be remembered that mut. debilis, when self-polHnated, was marked by 
great seed steriHty. This steriHty was much reduced when pollen 
from one of the other forms was used, and the effect of foreign pollen 
was greatest of all when that of mut. hilonga was used. Doubtless 
several factors are concerned with the increase of fertility on crossing, 
but it seems not unwarranted to call attention to the fact that if our 
hypothesis were true such an increase would be expected, because good 
embryos would result from the fertilization of female /3 gametes by 
male « gametes. All the hilonga individuals in the mixed progeny 
from debilis X bilonga would be represented in a self-pollinated prog- 
eny by aborted seeds. 
On the whole, the facts point to the truth of the hypothesis of non- 
equivalent gametes. The facts to be explained are sufficiently orderly 
to demand more than a superficial criticism at the hands of those who 
see in the mutation phenomena merely evidence of Mendelian segre- 
gation. It seems to the writers that the work with Oe. Reynoldsii 
affords very convincing evidence of De Vriesian mutation. 
Quantitative Evidence of Matroclinic Inheritance 
Although no one who has had an opportunity to examine the 
mutation crosses has doubted the fact of matroclinic inheritance, it 
was of course essential to obtain quantitative data that would con- 
vince one of the accuracy of our observations. Leaves and capsules 
from self-pollinated and crossed progenies were therefore measured, 
both in order to establish the fact that the several forms differed widely 
from one another and to show that the mutation crosses resembled the 
pistillate parent. In most cases a large enough number of plants was 
at hand to give satisfactory data. 
Mature stem leaves were measured from plants of all the pure 
strains and mutation crosses, five leaves being taken at the same part of 
the main stem from each plant as it came in the row, without selection. 
The leaf lengths are summarized in Table III, the widths in Table IV. 
The two tables are based upon the same material, but individual 
leaves were frequently imperfect, so that one or the other measurement 
could not be made. On this account the number of measurements 
does not always tally in the two tables. It is very clear that the 
modes of the variation curves lie very close together in the cases of all 
progenies having the same pistillate parent. There are some dis- 
crepancies, to be sure, the most notable being the failure of a closer 
