242 Studies in Indian Cotton 



The petals may be of one of two sizes, either small, when they lie 

 within the bracteoles whose length they do not exceed, or large, when 

 they project beyond and are about double the length of the bracteoles 

 (vide Figs. 3 and 4). The exact size of the larger petal varies somewhat 

 with the particular type but in no case approaches that of the smaller, 

 and the two stand in marked contrast without intermediate form. 

 There appears to be complete correlation between the size of the petal 

 and the colour. The smaller petal is invariably white and the larger 

 petal invai'iably yellow. Among the plants under experiment, which 

 now amount to over a hundred thousand, and among cottons under 

 cultivation in the field no single exception has been observed. The 

 correlation holds with the simple yellow and white types and also with 

 those types in which a red colour is superimposed. It follows from 

 this that all plants with a 7-ed on yellow flower, such as type 3, have 

 large petals, while plants with a red on white flower, such as type 11, 

 have small petals. The cross between types 3 and 9 illustrates this 

 point well ; in all cases both plants with red on yellow, and those with 

 yellow flowers, whether pure or impure, have large petals, while in the 

 plants with red on white flowers, whether pure or impure, and in those 

 with white flowers, the petals are small. 



A further instance of correlation, and one which is of considerable 

 importance both practically and on account of its bearing on the argu- 

 ment of section 3 {d), has been found to exist between the presence of 

 the red colouring matter and an increase in the length of the vegetative 

 period. There is a distinct retardation of the commencement of the 

 flowering period when the red sap colour is present. This is shown in 

 Table XXIX. In this table the unit is a plant of the F^ generation 

 and the figure is, for the pure forms, taken as the average of the F^ 

 offspring and, for the impure forms, as the average calculated from 

 only those F^ offspring which are, judging by the depth (to lamina), or 

 absence, of the red colour, pure in this character. 



In the light of this correlation it is necessary to reconsider the 

 results detailed in section 3 {d). In that section attention was drawn 

 to the monomodal curve as indicating incomplete resolution. No 

 distinction was, however, made between plants with, and plants without, 

 the red colouring matter. It would appear possible that a separation 

 of the plants into two groups dependent on the presence or absence of 

 the red colouring matter might disclose two trimodal curves, whose 

 presence is rendered obscure through superposition. Table XXIX, 

 however, in which such a separation is effected, shows no such trimodal 



