304 Non-Isolablc Races. 



amounts employed in this experiment (about 80 grams 

 per square meter) should not be exceeded, that is to say, 

 that the result cannot be improved by still heavier ma- 

 nuring. 



For the experiment with sand I dug in my experi- 

 mental garden a bed of 13 square meters in extent and 

 one-half a meter deep, and filled it with ordinary fine 

 sand. On this bed and on a neighboring one of the same 

 size I sowed seed of Oenothera Laniarckiana in the sum- 

 mer of 1899. The control bed was not manured but 

 contained a very fertile soil ; the seed was sown in the 

 middle of April. 



1lie sand of the bed bordered immediately on the 

 rich soil of the path which surrounded it.-" Therefore 

 the plants at the margin could thrust their lateral roots 

 into this, and thus obtain richer food than the more 

 central rows. This circumstance showed very important 

 results during the course of June, for wdiile many flower- 

 ing stems were produced towards the outside of the l)ed, 

 hardly any occurred in the middle. It was not until the 

 middle of July that the development of stems set in here 

 also. Curiously enough this occurred in almost every 

 instance at exactly the same time. In the middle of 

 August among the 82 plants of the outer rows about 

 60% had developed stems, whilst in the middle there 

 were 133 rosettes amongst 203 plants, that is to say 

 about 24% of annual specimens. We see that the dis- 

 tances between the plants in this experiment were very 

 considerable, for on 13 square meters there were only 

 285 plants. Even at the end of the summer they hardly 

 touched one another. In the control experiment in which 



^ In subsequent years I have separated the sand from the earth 

 by boards. 



