I']. II. Saundkhs 141 



that doubles on the whole develop more rapidly and vigorously than 

 the singles, and that where the penod of development is sufficiently 

 prolonged, selection based on this difference can be used as a means 

 of securing a higher proportion of doubles in the beds than corresponds 

 with the actual output from the parent plants. 



The following figures are taken from the F^ and F^ generations 

 obtained from a cross between two Intermediate (East Lothian) strains, 

 both throwing doubles and yielding the normal output. The marked 

 difference in the proportion of doubles obtained among the more and 

 the less vigorous individuals is apparent in whatever way the figures 

 are marshalled for comparison. The seedlings in each culture were 

 numbered in order of vigour as described above and the fiowers recorded. 

 The total number of singles and doubles obtained among the plants 

 belonging to each grade of vigour is shown in Columns 1 and 2 where 

 it is seen that the proportion of doubles occurring among plants of the 

 highest grade of vigour (grade 1) is far in excess of expectation, being 

 about 3D : IS instead o{ 9D :7S. As we descend the scale of vigour 

 the proportion of doubles diminishes until we reach equality or even a 

 slight excess on the side of the singles. The figures in the two columns 

 are compiled from the same data treated in diiferent ways as set forth 

 below. By either method this relation between degree of vigour and 

 the proportion of doubles is clearly brought out. Column 1 gives the 

 number of singles and doubles obtained in the different grades when 

 each individual in a culture is counted as belonging to the grade 

 corresponding to its position in its own series. According to this 

 method the most vigorous individual in every culture is counted in 

 grade 1, the second most vigorous in grade 2 and so on. It is obvious 

 that on this method of plotting the results a gap will occur in the 

 record whenever an individual dies before flowering. The result will be 

 a certain discrepancy in some or all of the grades between the number 

 of singles and doubles taken together, and the number of seedlings 

 originally classed in that grade. Column 2 gives the figures obtained 

 when the records are plotted so that these gaps are closed up. Since 

 such mortality as occurs may be taken to be purely haphazard we may 

 disregard those members in a series which die young and renumber the 

 surviving members so that they stand in the same order as before but 

 in consecutive positions. We may illustrate this point by the following 

 case. In a culture of 9 seedlings in which those numbered 6 and 8 die 

 before flowering the individuals numbered 7 and 9 might either be 

 counted as still belonging to grades 7 and 9 (the method adopted in 



