THE COTTON PLANT 93 
are planted here and there through an ordinary field of 
a variety which is roughly similar in size and habit of 
growth; they obtain no special treatment whatever, 
but are carefully observed day by day. 
Having mentioned the subject of plant-breeding, 
we should make a serious omission from this article if 
we neglected to mention the possibilities of Mendelism. 
The work which has already been carried out in this 
direction on cotton, under the limitations already men- 
tioned, shows promise of useful results in the future, 
but any serious investigation of heredity, as such, in the 
cotton plant, is beyond the powers of ordinary agricul- 
tural experiment stations, for the following technical 
reasons, The plants are large, and even a small 
“‘ family ’’ occupies a great deal of ground; they are 
subject to sufficient natural crossing to obliterate entirely 
the demonstration of the more subtle Mendelian ratios 
unless the flowers are artificially protected at consider- 
able expense from such contamination; the work of 
the writer in particular has shown that such ratios, 
indicating complex inheritance, are far from uncom- 
mon; further, many or most of the characters which 
have economic importance are not matters of colour 
or form, but of dimension, and before these can be 
studied adequately in their inheritance, the student 
must have intimate knowledge of the possible fluctua- 
tion which such dimensions may undergo as the result 
of mere environmental changes. Lastly, the practical 
application of such results as have been obtained has 
been hindered because the grower has been working 
in the dark, on account of the absence of detailed know- 
ledge on the consumers’ side of the trade; there has 
