8io 
Journal of Agricultural Research 
Vol. XXI, No. ii 
Table II .—Means of number of culms per plant with initial culmage from I to 4 
Date. 
Mean number of culms. 
June 19. 
I 
2 
3 
4 
24. 
i- 57 
2. 92 
3. 61 
4. 27 
T , 2 9 . 
3 - 46 
5. 08 
6. 18 
s-8 7 
July s . 
5 - 79 
8. 46 
12. 11 
I2 - 33 
IO. 
6. 71 
9.67 
14. 46 
! 5 - 53 
15 . 
7 - 75 
12. 29 
17. 18 
18.73 
20. 
9 . 71 
15.00 
19.96 
21.93 
25. 
14 . 04 
00 
00 
M 
c* 
25-43 
28.93 
31 . 
20. 29 
29.42 
35 - 79 
39.80 
Aug. s . 
24- 39 
32 - 58 
41. l8 
46.73 
IO. 
3 i* 5 ° 
39-63 
52. 29 
58.27 
15 . 
38-18 
49. 21 
65 - 39 
68. 00 
22. 
48. 18 
62. 83 
83-71 
86. 60 
26. 
54-32 
71.67 
95 - 43 
97 - 47 
0 3 1 . 
63. 86 
83-63 
109. 07 
109. 60 
Sept, s . 
75-64 
94.21 
122. 82 
123. 93 
II. 
90. 04 
IO7. 42 
143. 04 
140. 40 
A casual examination of the four series of means in Table II shows 
that, so far as number of culms per plant is concerned, there is a gradual 
increase in differences except between the two series initialing at 3 and 
4. The series initialing at 4 has fewer culms per plant at the third and 
last count. Apparently the last series shows a lag, in comparison 
with what one observes among the first three series. A possible explana¬ 
tion of this lag will be given later, but attention is called again to the 
fact that the last series included only about half the number of plants 
found in the other series. To render the four groups more graphic they 
are plotted and shown in figure 2. 
It seems strange at first sight that two groups of plants differing by 
only a single culm at an early stage of their existence should later come 
to show such a marked difference in this regard. It seemed possible 
that an explanation might be forthcoming on the basis of different 
degrees of heterosis; that is, the different degrees of growth vigor due 
to the differences in heterozygosis or amount of hybridity. 
Two assumptions would be necessary, in connection with the well- 
known fact that brome grass is a normally inter-pollinating plant. One 
assumption would be that in a random sample of brome-grass seed¬ 
lings some of them would possess a considerably higher degree of heterosis 
than others and that this difference in heterosis would result, in one 
direction, in a varying number of culms produced per plant. It is 
extremely probable that brome-grass seedlings taken at random have 
genotypes of sufficient diversity to be responsible for certain differences 
in vigor of growth which might easily find expression in a difference in 
number of culms. Moreover, one would look for an initial difference 
of this kind to be cumulative in character, as the difference in vigor 
would persist through the life history of the plant. The foregoing expla- 
