12 
Journal o f Agricultural Research 
Vol. XXIX, No. 1 
The F lf which was apicallv-awn- 
letted, approached more closely to the 
awnless than the awned parent. If 
the data can be interpreted on a 1-fac¬ 
tor basis it is necessary to look for the 
25 per cent of recessives among those 
having more awn development than 
Fp In F 2 , classes 4 and 5 combined 
include 26 per cent at Mandan, 24.5 
per cent at St. Paul, and 21.7 per cent 
at Davis. The total number of plants, 
constituting 24.4 per cent, was shown 
to be not significantly different from 
the 3:1 ratio and suggests that the 
awned and short-awned classes rep¬ 
resent the recessive type. 
In F 3 , however, the awned or short- 
awned classes did not breed true within 
the limits of classes 4 and 5 combined, 
either at Mandan or at Davis. At 
both points only about 6.5 per cent of 
the whole F 2 bred true to these limits. 
Thus the simple hypothesis that classes 
4 and 5 are due to a single recessive 
factor is untenable. 
It may be assumed that those plants 
recessive with respect to the primary 
factor, if any, are distributed among 
classes 3, 4, and 5. As these classes 
make up 55.1 per cent of the F 2 popu¬ 
lation at Mandan and 44 per cent of 
that at Davis, it is necessary to assume 
that a large part of classes 3, 4, and 
perhaps even 5 (which does not wholly 
breed true within the limits of classes 
3 to 5) exceed the F! in amount of awn 
because of some other factor than the 
one being considered as primary. Only 
a part of these classes (3, 4. 5) breed 
true to their wide range. At Mandan, 
31.2 per cent of the F 2 was composed 
of such plants (classes 3, 4, and 5 pro¬ 
ducing only 3, 4, and 5) but only 15.1 
per cent at Davis. Evidently it is 
possible to find a recessive class in the 
Mandan group data, but the Davis 
group data indicate that some of these 
classes recessive in the most inportant 
factor must have been as low as class 
2 . It is very clear that there is no one 
outstanding recessive factor responsi¬ 
ble for the awned condition. 
Looking at the other end of the F 2 
series for the homozygous dominant 
class, it is shown in Tables V and VI 
that while 44.9 per cent at Mandan and 
56 per cent at Davis were of classes 1 
and 2, there were only 7.4 per cent at 
Mandan, and 11.6 per cent at Davis, 
of classes 1 and 2 of the F 2 population 
which produced only classes 1 and 2. 
Thus with reference to the most 
important factor homozygotes must 
be looked for in class 3 as well as 
in classes 1 and 2. Twenty-six per 
cent of the F 2 generation at Mandan 
and 35.9 per cent at Davis, were of 
classes 1, 2, and 3 producing only 1, 2, 
and 3. These can include all homo¬ 
zygous dominants. Thus it is possible 
that there may be a dominant factor 
present in classes 1 to 3 and a recessive 
factor present in classes from 3 to 5 at 
Mandan and in classes 2 to 5 at Davis, 
with heterozygotes appearing in classes 
from 1 to 4 and perhaps even 5. 
To account for this wide range and 
necessary overlap of the classes con¬ 
taining the homozygous dominant or 
recessive for the most important 
factor, it is necessary to assume the 
existence of a second factor, or a group 
of factors, nearly if not fully as impor¬ 
tant as the first. 
On the hypothesis that there are two 
factors of equal importance, it will be 
necessary to find 6.25 per cent of 
the F 2 strains at each extreme breeding 
true. Only 3. 3 per cent of the Mandan 
F 2 were of class 5 breeding true to 5, and 
5.4 per cent of the Davis F 2 were of that 
sort. It thus is necessary to suppose 
that some of the 2-factor recessives 
were in class 4. At the other extreme 
only 0.2 per cent of the Mandan F 2 were 
of class 1 breeding true to 1, and 4.1 per 
cent of the Davis F 2 were of that sort. 
Thus some 2-factor homozygous domi¬ 
nants must be looked for in class 2. 
There were 6.4 per cent of F 2 classes 1 
and 2 which produced only 1 and 2 at 
Mandan, and 11.6 per cent in the 
Davis F 2 . At the other extreme there 
were 6.5 per cent of Mandan F 2 and 
6.6 per cent of Davis F 2 of classes 4 and 
5 producing only 4 and 5. The 2-factor 
homozygous dominant and recessive 
(expectation 6.25 per cent) thus can be 
found in classes 1 to 2 and 4 to 5, re¬ 
spectively, but additional factors are 
necessary to account for the variation 
remaining./ 
To have arbitrarily assumed the 
awnless to be the recessive class, as 
Howard and Howard (22) had done in 
their 2-factor hypothesis and then 
grouped the remaining classes-as awned, 
6.25 per cent of the awnless plants 
should have bred true. As only 0.2 per 
cent of class 1 bred true for class 1 at 
Mandan, and 4.1 per cent at Davis, it is 
apparent that their findings could not 
apply to this cross. The writer, there¬ 
fore, assumes that, for the Kota-Hard 
Federation and reciprocal crosses here 
studied, two factors at least must be 
present in a dominant condition in awn¬ 
less strains, and that the awned plants 
should be represented at least by dou¬ 
ble recessive factors. This is a con¬ 
clusion opposite to that of Howard and 
Howard. 
The 2-factor hypothesis here ad¬ 
vanced does not entirely explain the 
