272 
Pearsall.—Studies in Growth. 
rapid growth of the fruits takes place four (or four and a half) weeks after 
the most rapid growth of the flowers. Thus, roughly, the food demands of 
the fruits will follow those of the flowers by about four (or four and a half) 
weeks, instead of by seven weeks, and to represent the effect of fruit 
development upon the flowering curve, from this point of view, the fruiting 
curve needs to be transposed three (or two and a half) weeks earlier (i. e. 
seven to four). The drop in the flowering rate then appears to be directly 
related to the increase in the numbers of fruits, and the horizontal portions 
of the curves also coincide, representing a condition of equilibrium when 
constant proportions of the food supply are used respectively by flowers 
and fruits. This balance seems to be maintained throughout the final 
decline in growth rates, which according to Balls is due to the rising water 
table, root asphyxiation becoming marked at the beginning of September. 
The smaller variations can be eliminated and the main features of the 
curves retained if the figures are presented as total stem length, number of 
flowers and of fruits produced by the average plant at different times. The 
dependence of stem length upon the time of flowering, and of the total 
number of flowers produced upon the appearance of fruits, is equally clearly 
indicated by this alternative method of presentation. 
The same results are obtained from the other sets of figures given by 
Balls and Holton ( 1 , 2 ), and also from those of Harland ( 3 ), for cotton 
grown in St. Vincent, though in the latter case rather abnormal rainfall 
occurred in the middle of the flowering period. Harland’s data are of 
interest, since they show a higher rate of flower production when decreased 
fruit production took place, showing that fruit production had depressed the 
rate of flowering. 
We may pass on to consider other aspects of the development of cotton- 
plants which seem to fall in line with the views put forward here. The 
valuable data of Harland ( 3 ) are of particular interest in this respect, since 
he analyses statistically the growth of various parts of the plants, though 
attempting no explanation of his results. 
Referring to Fig. 5, on any branch B the order of flowering is F lt F 9 , 
F 3 , &c. If the growth rate of the branch is uniform then time intervals, Fj- 
F 2 , F 2 -Fo, F 3 -F 4 , &c., should be uniform. Harland shows statistically that 
actually in cotton the time interval between the flowering of adjacent nodes 
(Fr-F, F 2 -Fo, &c.) increases as the distances from the central branch 
increase. Hence a point is reached when the rate of growth is practically 
zero. When it is remembered that each flower may produce a fruit, 
i.e. a growing-point nearer the source of supply (the stem), it seems 
possible to explain the falling off in growth rate of later nodes as being due 
to the growth of fruits lower down the branch, which serve to reduce the 
rate of food supply to the branch apex. 
Harland also shows that, in a precisely similar way, the time interval 
