352 The Philippine Journal of Science 1915 
plant is impossible when seed selection is based upon a composite 
sample of the ordinary varieties of rice. Considering that 
there is an almost unlimited number of Philippine varieties, 
that planting and harvesting operations for the past centuries 
have resulted in a great many combinations of the forms, and 
that there is more or less variation within a supposedly pure 
variety, standardization or purification of varieties cannot be 
expected when the seed is taken from the common market stock. 
One of the ordinary samples of rice—only one of many examined 
at the College of Agriculture—was found to include three dis- 
tinct varieties, differing in texture, shape and size of grain, and 
color of hull and of cuticle. Frequently, samples which had been 
carefully graded and were apparently uniform were found to” 
be, when hulled, nonuniform in color of cuticle and in texture 
of endosperm. 
Selection must be confined to the product of a few individual 
plants. In this way it is possible to govern the variation by 
eliminating many of the undesirable characters possessed by 
ordinary individuals and to standardize or purify the varieties. 
The work done by H. O. Jacobson,’ Bureau of Agriculture, 
on the ‘“‘Head-to-the-row test of rice’ is ample evidence of the 
fact that selection must be limited to a few exceptional plants 
within a variety, and that this method results in rapid improve- 
ment, high uniformity in yield, and a consequent accumulation 
of desirable characters (Plate II, figs. 1, 2; Plate VI). 
TIME TO SELECT SEED 
The practice of delaying seed selection until planting time and 
of using either a part of the former crop for seed or of 
securing seed from a neighbor’s general crop does not provide 
for seed improvement nor insure results above the average. 
Seed selecting should be done at harvest time. The farmer 
should go through his field and remove enough of the most 
desirable plants to furnish his seed for another year. This will 
enable him to base his final selection upon plants having a large 
number of similar panicles of good size, large stools, stiff stems, 
or any other characters which he considers desirable (Plate V, 
fig. 1). 
The production of pure pedigreed varieties of uniformly high 
yield demands careful grading and rigid elimination. Only a 
few exceptional plants with the most desirable combinations of 
characters should be used. For this reason the number of plants 
* Phil. Agr. Rev. 7 (1914) 346-851. 
