AND COLLEGE, PUSA, FOR 1914-15. 41 



eating the existence of a number of different colour factors. 

 Carthamin is absent altogether in some of the types while 

 in others it is but feebly developed. Dr. Marsden of 

 Madras, who carried out some dyeing tests with the various 

 types, found that the best colour bearing variety was eight 

 times better than the worst. In the percentage of oil in 

 the seeds, the range in values is nothing like so great as 

 in the case of the carthamin content of the faded flowers. 

 With one exception, the percentage of crushed seeds ex- 

 tracted by ether varied from 20-77 to 30-19. Nineteen of 

 the twenty-four types contained over one quarter of their 

 weight of oil. Oil and carthamin content were not found 

 to be antagonistic and in several types high oil and high 

 colour occurred together. 



Indian mustard (rai). The most interesting feature 

 of this crop, when studied in pure culture, was found to be 

 the extraordinary range in form. One hundred and two 

 pure types were isolated which varied from thirty inches 

 to nearly ten feet in height. Almost every conceivable 

 intermediate form between these extremes was represented 

 and so close was the resemblance that in many cases the 

 types would only be distinguished by the massed habit. 



Self-pollination was found to be the rule in rai but 

 crossing, to the extent of about fourteen per cent., occurred 

 when the types were grown next to next in lines. The 

 pollination details were found to agree in the main with 

 those previously described in other species of Brassica. A 

 certain amount of evidence was obtained on the inheritance 

 of characters in this crop which indicated the existence of 

 numerous factors. Time will not admit of this aspect of 

 the work being continued. 



Both in safflower and in rai ? the results obtained in 

 this study point to the overwhelming importance of 

 selection in the improvement of crops like these in which 

 some crossing takes place and where the range of form is 

 so great. Form separation, if conducted on a broad basis, 

 would almost certainly lead to the isolation of any desired 

 type which could be multiplied at once and distributed to 



