163 
Several of these families consisted of but one individual eacli^ 
and these may be left out of account as having no significance. The 
most interesting is 09258 in wbich a bursa-pastoris parent produced 
a Single Hejgeri offspring^ thus showing that tliis parent was 
heterozygous in respect to capsule-character. Six of the eleven 
families derived from hursa-pastoris parents bred true to the 
hursa-pastoris type of capsule, this being slightly in excess of 
expectation on the assumption that this character is determined 
independently by two genes^ and considerably in excess of the 
one-third which should have bred true if but one gene differentiated 
the Heegeri capsules from the bursa - pastoris capsules of the 
P^. However the number of families is wholly inadequate to permit 
the attachment of any special significance to this closer agreement 
with the requirements of the two-gene hypothesis. All of the 
families from parents having Heegeri capsules have bred true to 
the parental character^ as they should do to agree with Mendelian 
interpretation. The four families which split into the two parental 
types show the ratios, 4.67 : 1^ 15.6 : 1^ 24.0 : 1, and 63.5 : 1^ all 
of these ratios dififering in the same direction but in quite various 
degrees from the two available ratios 3 : 1 and 15 : 1. Two of 
the families show a suggestive approximation to the expected 
ratios^ while the other two depart widely from the nearest 
available ratio 15 : 1. All of these families as well as the three 
families of the have shown a smaller proportion of Heegeri 
plants than required by the hypothesis that the bursa-pastoris 
capsules are determined independently by two genes. 
It appears to me that the explanation of these results is to 
be sought in some modifying influence acting upon the normal 
Mendelian processes. The ratio 4.67 : 1 may then represent a 
modified ratio of 3 : l, and the other three ratios may be referred 
to the ratio 15 : 1. These three families taken together give 
a ratio of 22.2 : 1, essentially identical with the observed ratio 
21.9 : 1 in the Fg, and showing almost exactly the same pro- 
portional departure from 15 : 1^ that 4.67 : 1 shows from 3 : 1, 
for 4.67 : 3 = 23.3 : 15. 
The nature of the modifying cause or causes which may be 
operating to produce these defective ratios need not be discussed 
at length here, as the matter is capable of experimental treat- 
ment and is being investigated ; but it may be })oiutcd out that 
either „selective fertilization" favoring the union of unlike gameles, 
11* 
