WATER-STRESS BEHAVIOR OF PIMA COTTON. 
19 
It will be seen from Table VIII that the greatest number of 
flowers per plant was produced on plat 1, to which the greatest 
quantity of water had been added. However, on account of the 
greater amount of shedding on this plat there was a smaller number 
of bolls set by October 1 than on plat 3, which had received the least 
water. The ' ; potential seed-cotton production " on October 1 of 
plants on plat 3 exceeded that of any other plat, but the " potential 
seed-cotton production " is not in accord with the actual production, 
as determined upon picking. This discrepancy can be accounted for 
by the records of frost damage to bolls, as shown in Table VII, the 
percentage destroyed having been greater in plat 3 than in plat 1. 
Bolls set late in September were not able to develop fully, and those 
which cracked open sufficiently to be picked were lighter in weight 
than those set earlier in the season. 
Table VIII. — Comparison of the flowers produced, the percentage of shedding, 
the bolls set, and the potential seed-cotton production in each of four plats of 
Pima cotton with the total yield of seed cotton per acre in 1919. 
Plat. 
Mean num- Mean per- 
ber of flowers ; centage of 
produced, i shedding. 
Mean num- 
ber of bolls 
set by 
Oct. 1. 
Mean 
Yield of 
No. 1 ". 58. 2± 1.79 26.5±0.38 
No.2 50.9±1.78 21.6±.34 
No.3 J 57.0±2.16 ! 16.7± .21 
No. 4 1 48. 2± 1.30 I 17. 8± .26 
43.1 ±1.71 
40. 0± .40 
47. 6± 1.89 
39. 7 ±,..36 
1,602 
1,437 
1,496 
b 1.397 
o Mean "potential seed-cotton production" in the sense here used is the mean weight of seed cotton per 
boll as determined upon picking, multiplied by the mean number of bolls set per plant on Oct. 1. 
t The presence of a "slick spot" of approximately 900 square feet in plat 4 was responsible in part for the 
low yield on that plat. This soil variation was not noted until its presence was indicated by the stunted 
condition of the plants. 
Such data, Tables VII and VIII, seem to indicate that every pre- 
caution should be taken in growing cotton in areas where there is 
danger of damage from early frosts to stimulate the setting of fruit 
to the fullest extent during the months of July and August. The 
data show that the first irrigation on June 10 was not so early that 
it produced an excessive number of vegetative branches but served 
the purpose of stimulating the plants into early fruiting, which was 
especially important for that particular season. The 16 days lost 
by the plants in plats 3 and 4 in the early processes of fruit produc- 
tion was a handicap that they never overcame. While the cessation 
of growth and flowering immediately after the 1st of September 
might result in a reduction of the yield, such loss of time under Salt 
River Valley conditions would hardly result so seriously as an equal 
length of time lost 'at the beginning of the fruiting season. 
The more pronounced slowing in late summer of the growth 
rate and flowering of cotton plants which had produced a heavy 
growth of vegetative material early in the season presents a very 
interesting physiological problem. The work of some recent investi- 
