302 Non-IsolaUe Races. 
apart had 162 plants which developed stems, whilst on 
that in which they were close together there were 145. 
The difference between the two reckoned as a percentage 
of the whole culture is of course more striking, viz., 
83% as against 20%. More important, however, is the 
fact that per each square meter in absolute numbers 
more annual individuals are produced when the plants 
are grown far apart than when they are grown more 
densely and therefore in larger numbers. When viewed 
in this way the result points to the great importance of 
sowing seeds thinly in experimental cultures. 
Experiments with shading are met with the difficulty 
that the young plants cannot stand it very well, even 
when, as in my experiments, the shadow is that of trees. 
The experiment was conducted at the same time as that 
of 1890, referred to above, on a similar scale and by 
growing the plants far apart; it produced about 46% 
annuals as against 72% in the control experiment al- 
ready mentioned. 
By far the best means, however, of increasing the 
proportion of annual plants or even of securing their 
exclusive production, is to sow the seed and keep the 
young plants under glass. In doing so the seed can be 
sown in March or April in un-manured sterilized soil, 
and the seedlings may be pricked out singly into pots 
containing richly manured soil after the appearance of 
the third or fourth leaf. In this condition they remain 
under glass until the end of May, at least during the 
nights and on cloudy days, and can then be turned out 
of the pots without breaking the ball of earth round the 
roots and transplanted to the place where they are to 
grow. Treate 1 in this way almost all the plants behave 
