E. R Saunders 365 



division exists, since all grades occur, some arbitrary classification must 

 be made for the purpose of comparison, and the line is therefore drawn 

 between those pods where at least half the number of seeds sown 

 germinated, and those in which less than half proved to be good. The 

 results may be summarised thus : 



Nnmber of Namber of 



seeds sown seeds sown 



where less Namber of Number Namber where st Number of Namber Namber 



th&D half seeds which of of least half seeds wbkh of of 



Type germinated germinated singles doables germinated germinated singles doaUes 



In both cases the fruits containing the less good seed have given 

 a higher percentage of doubles. 



As to the proportion of doubles actually obtainable from the various 

 types the numbers quoted in seed catalogues range from about 50 per 

 cent, to as much as 90 per cent. In the case of the Erfurt Ten Week 

 strains from 60 to 75 per cent, is given. This is a rather higher 

 proportion than was found to occur in the breedings here described, 

 where the average ranged between 53 and 57 per cent., though 

 a considerably higher proportion might now and again be obtained 

 in individual sowings. Chate^ believed his experiments to show that 

 a larger percentage of doubles could be obtained from the pods on the 

 main stem and from the lower ones on the primary laterals than from 

 those on the laterals of a higher order; and similarly from the seeds 

 from the lower region of a pod as compared with the upper : the 

 difference is given as 20 per cent, only of doubles from branches of a higher 

 order as compared with 65 per cent, from those of a lower order, and 

 30 to 35 per cent, from the upper region of the pods as compared with 

 75 to 80 per cent, from the lower region. These two latter numbers 

 would give an average of 55 per cent, for the fruit as a whole, which 

 agrees very closely with the observations contained in the present 

 paper, and with the avei*age which the theoretical considerations here 

 advanced would lead us to expect. No indication of the aggregation of 

 seeds giving rise to doubles in definite regions of the fruit was obtained, 

 although a number of observations were made with a view to testing 

 this point. 68 pods belonging to three different strains (red, marine 

 blue, and Princess May) were halved transversely, the seeds from the 

 upper and lower halves being sown separately. The same result was 



' loc. eit. p. 79. 



