The Inheritance of Quantitative Characters in Maize 57 
INHERITANCE OF WEIGHT OF SEEDS. 
The behavior of seed weight in the crosses (60-3x54), 
(60-5x54), and (60-8x54) may be considered together. The de- 
tailed data are shown in Tables 20, 21, and 22. In making the 
weights for this study, it was necessary to use a scheme by 
which the sugary or wrinkled seeds of the Black Mexican parent. 
No. 54, and of certain recessive segregates from the cross, could 
be weighed as starchy seeds. This end was accomplished by hav- 
ing starchy families growing near the wrinkled families. 
Crossed seeds which had become starchy thru xenia were usually 
obtained in sufficient numbers to make data upon 25 seeds pos- 
sible. An insufficient number of hybrid starchy seeds, however, 
accounts for the meager data in certain families. It must be 
understood that this method of weighing is correct in theory, 
since the size of the seed is governed by the development of the 
pericarp and since this is unaffected by the pollen from a starchy 
race. The pericarp is merely filled out with starch by the dom- 
inance of starchiness in the F-, endosperm. 
In discussing these results, the weight of 25 seeds is taken 
as the unit quantity. When the mean weight of any particular 
family is stated, therefore, this is what is meant. 
Tom Thumb No. 60 has an extremely small seed. The mean 
weight is 2.7 ± .03 grams. The mean weight of the Black Mexi- 
can is 8.3 ± .11 grams, but this is probably too high. These data 
were taken from plants raised from commercial seed, which was 
in a very mixed condition. The last three positive classes should 
probably be discarded for this reason. The ears upon which they 
were borne showed pl°' T ily that they had come from hybrid seed, 
— the male parent evidently bei^g some large dent variety. 
Small numbers in the frequency distributions of both parents 
and hybridity in the Black Mexican undoubtedly make the co- 
efficients of variability — 14.44 ± .90 per cent and 14.54 ± .96 per 
cent respectively — too high; but, even leaving this fact out of 
consideration, the ten F 2 families grown show increased varia- 
bility over that of the F 1 generation in every case. The range of 
variability in each case overlaps that of the grandparents. The 
extreme F 2 individuals do not reach the extremes of either parent, 
but if one leaves out of account the three higher classes of the 
larger parent, for the reason given above, the extreme size of the 
Black Mexican is generally obtained. This may be a question- 
able procedure, but, as a matter of fact, the individuals used as 
the male parents of the crosses were only average in this char- 
acter. The F 3 families often gave segregates with seeds as small 
as the smallest Tom Thumb seeds. 
The F 3 families having the smallest mean seed weight were 
