E. R. Saunders 139 



in the beds of a College garden in Cambridge. The gardener who 

 raised the plants saved his own seed, that of any one year providing 

 the plants for the next season but one. Each year the yield of doubles 

 in the beds was found to be far in excess of expectation. This 

 excess could not be accounted for on the ground of a differential 

 mortality between the singles and (hiubles after planting out, for 

 gaps in the ranks of the flowering plants were extremely few, and 

 in one season every individual planted out was accounted for : nor 

 on the supposition of intentional selection to this end on the part of 

 the gardener who, though pleased with his success, was yet unable to 

 account for it, his method being simply to select the best grown plants 

 for bedding out and to discard the rest. 



The numbers recorded in the beds in fijin- successive seasons were 

 as ft.illows : 



the proportion of doubles ranging from 80 to over 85 per cent., where 

 the expectation according to results obtained with all strains hitherto 

 investigated would be (as stated above) under 57 per cent. A control 

 experiment carried out with the same material proved conclusively 

 that the excess observed among the garden plants was not the out- 

 come of a corresponding excessive output of doubles by this strain but 

 must be explained in some other way. In the control expei'iment 

 seed saved from the 1911 plants was sown the following year, and — 

 an important point — an endeavour was made to bring every seedling 

 to flower in (jrder to obtain the fullest record possible from the sample 

 sown. The result was entirely in accord with previous expectation, 

 the numbers obtained being 185 singles and 224 doubles or about 

 55 per cent, of doubles, thus proving that this strain behaves as regards 

 output of doubles precisely as all other strains that have so far been 

 tested. There is, in fact, no doubt that the strain in question is a 

 perfectly normal one and yields only the usual slight excess of double- 

 flowered plants. At the same time it is clear that by a suitable 

 selection of the young plants it is possible to secure a better show 

 of doubles in the beds than is warranted by the actual output by the 

 germ cells. 



