COTTON-SPACING EXPERIMENTS 3 
which assured weights from rows of the same length. In some cases 
the length of the sections was determined by the size of the plants, 
with a view to including only plants of comparable size and develop- 
ment. Large productive plants at the ends of the rows were elimi- 
nated. The yield of seed cottun from each section of each row was 
recorded separately. This system of subdividing the experiments 
gave an increased number of row yields and 4-row block comparisons 
and afforded a basis for comparing statistically the yields from the 
different spacings. Usually only two, and never more than three, 
spacings were compared in any one experiment. 
Before the cotton was picked, boll samples were systematically 
taken from each block of each spacing for laboratory examination. 
From these samples data relative to boll size, seed weight, percentage 
of lint, and length of lint were obtained in order to study the relation 
of spacing to these characters. 
After picking, the plants were counted, and these counts are 
recorded for each row section in the tables of yield. 
Since conclusions based upon the total yield of each spacing are 
likely to be misleading, the interpretation of yields is based upon 
consistency of behavior in the side-by-side comparisons. Though 
the data could be presented and analyzed in various ways, it is prob- 
able that nothing of value would be added to the conclusions by 
doing so, and this feature has been intentionally avoided on account 
of the number of experiments reported. Where there are enough 
comparisons, the figures are handled on the basis of differences of 
the sum ef the yields of the two inside rows of adjoining blocks, 
which is a simple, reliable method of showing differences. 
DIFFICULTIES IN SPACING EXPERIMENTS 
As will be seen from the data of the experiments, many irregular- 
ities were encountered, both in the stands obtained and in the yields 
of individual rows where the stands were not unequal and the condi- 
tions were as uniform as could be provided. In view of these irreg- 
ularities it will be understood that misleading inferences may easily 
be drawn if only the total yields of spacing experiments are reported 
and used as a basis of comparison. 
One of the facts to be considered in spacing experiments is the 
sharp contrast that may appear between the outside rows of adjoin- 
ing blocks that represent different spacings or between outside and 
inside rows of the same block. 
It often happens that the outside rows of the close spacing give 
higher yields than the adjoining outside rows of a wider spacing and 
also higher yields than the inside rows of the close spacing. It ap- 
pears that the outside rows of the close spacing often rob the outside 
rows of an adjoining wider spacing. Allowance can be made for this 
robbing effect by limiting the comparisons of yields to the mside rows 
of the blocks, although it is a mistake to disregard the evidence 
afforded by the relations of the outside rows that the close-spaced 
plants so often gain a pronounced advantage over the wide spacing. 
Also it is possible that the inside rows of the close-spaced block may 
suffer to some extent from the competition of the more robust plants 
in the outside rows. On the other hand, a restriction of the plants 
in the outside rows of the open-spaced cotton may afford some advan- 
tage to the inside rows. ) 
