120 
for selecting plants with reference to the quality of the wheat itself. My 
object, then, in starting this work was to find some method of sampling the 
plant that would give an indication of its quality. 
In my work a Dumber of plants of Turkish Red wheat were analyzed, each 
head of the plant being harvested separately, and the analysis being made 
separately of each head, all kernels of each head being used. The data 
obtained show that there is quite a difference in the proteid content of These 
various heads, and that it would not he safe to take the analysis of any one 
spike as representing the composition of the plant on which it grew. 
Analyses were also made of groups of spikes on the same plant — that is. 
Supposing that a plant bore ten spikes, we would take five of those in one 
group and five in the other group, then shell out all of the kernels in cadi 
group and analyze each group separately. In other words, we made an 
analysis of half of the kernels on half of the heads of the plant. The results 
attained in this way were very much closer than the results of analyzing the 
separate heads. Thus by taking half the number of heads of the plant and 
analyzing them we got a fairly good sample of the whole plant. Separate 
analyses made of the two rows of kernels showed small difference in the total 
nitrogen and proteid content of the samples. 
In carrying on our work we have adopted the practice of sampling each 
head by taking one row of spikelets as a sample for analysis and the other 
row of spikelets as a sample for planting: then by sampling each head on the 
plant we get what we consider to be a good average sample of the plant for 
analysis. 
The proteid nitrogen content of 800 spikes of Turkish Red wheat selected 
in the field in 1902 with special pains to secure heads that were thoroughly 
matured and free from disease was determined and the seed planted. From 
the crop thus obtained selections were made and kernels from entire plants 
were sampled and analyzed. The minimum content of proteids, that is. 
proteid nitrogen multiplied by 5.7, was G.3S : the maximum, 28.21; showing a 
very large range and indicating the possibility of increasing very largely the 
nitrogen content by selection. Analyses of 351 plants grown from the above 
in 1903 showed a minimum proteid content of 5.84 per cent and a maximum of 
.".."...".4 per cent. 
When the analyses of the original S00 spikes were made determinations were 
also made of specific gravity of the kernels in a considerable number of the 
heads. The weight of the kernels on the low nitrogen content heads was found 
to be somewhat larger than it was on those of high nitrogen content. While 
the decrease was not regular, there was a slight tendency for the yield to 
decrease as the nitrogen content increased. The weight of the average kernel 
also agreed with the weight of the kernels on the spike, there being in the 
main a slight decrease in the weight of the average kernel as the content of 
proteid nitrogen increased. The number of kernels on the spike did not vary 
particularly. The volume occupied by the kernels decreased, or had a slight 
tendency to decrease, as the proteid nitrogen content increased. The specific 
gravity decreased regularly as the content of nitrogen increased. The proteid 
nitrogen in all the kernels increased as the percentage increased, and the proteid 
nitrogen in the average kernel increased as the percentage increased: so that, in 
spite of the fact that the kernels became somewhat smaller and lighter, the 
total production of nitrogen increased as the percentage increased. In other 
words, a selection based on increase in percentage of nitrogen would increase at 
the same time the total production of nitrogen. 
Similar figures for the nitrogen content of all of the heads of the plant of the 
crop of 1903. instead of one head as in 1902, also showed that the number of 
kernels decreased slightly as the percentage of nitrogen increased: the weight 
of the kernels also decreased as the percentage of nitrogen increased: the weight 
of the average kernel seemed to vary more than it did before, but the tendency 
was for the weight to decrease with an increase in the percentage of proteid 
nitrogen ; the grams of proteid nitrogen in the kernels increased in the main 
as the percentage increased : and the grams of proteid nitrogen in the average 
kernel increased as the percentage increased. The increase in the average 
kernel was very large. 
We discarded all kernels that were net fully developed because it is well 
understood that kernels that had not fully ripened would have a higher nitrogen 
content than those that had. 
In the crop of 1903 we also determined gliadin and glutenin in most of the 
plants. Our object in doing this was not very definite. Of course we were 
