266 



D. F. JONES. 



of the two kinds of seed in each member of the pair stated as 

 per cent. The closest perfect proportion is calculated from this, 

 based on the results from the A and B plants but not from the 

 C plants, and this subtracted from the actual proportion found 

 gives the deviation in per cent, which appears in the last line. 

 This description applies to mixtures I to 12 inclusive. Of the 

 remainder no correction for misplaced seeds is made. 



TABLE III. 



NUMBER OF PLANTS GROWN AND THE PER CENT. OF WRONGLY CLASSIFIED SEEDS 

 IN THE POLLEN MIXTURES INVOLVING YELLOW AND WHITE, STARCHY ENDO- 

 SPERM. 



The total number of seeds, the numbers in each class and the 

 deviations in per cent, from the closest perfect proportion for 

 the 20 pollen mixtures are summarized in Table II. The devia- 

 tion, if it is an excess of self-fertilized seeds, is given as plus and, 

 if the opposite, as minus. The deviations can range from + 50.0 

 to - 50.0 per cent. The extremes indicate complete non-func- 

 tioning of each kind of pollen on one set of plants and exclusive 

 functioning on the other. 



Altogether the number of seeds amounts to 63,694. Large 

 numbers are, of course, necessary to be convincing in any experi- 

 ment on selective fertilization as such investigations are largely 

 studies in sampling. Mixtures number 6, 7, 13, 17 and 18 are 

 less reliable than the others, because they have rather small 

 numbers from either the A or B plants, less than 200 in each case. 

 Like the chain with the weak link the value of each set of figures 



