214 Melafions hetioeeti Growth and Emiironment 



of both with the mean maximum, if some contrary tendency were not 

 in operation. This is brought out by the partial correlations 



C.r^^ = -.3638 



The correlation between mean maximum and mean minimum for con- 

 stant sunlight is + -6534, slightly greater than before, showing that 

 variations in sunlight affect these two variables in opposite directions; 

 the correlation between mean maximum and sunlight for constant mean 

 minimum is also greater, showing that the variations in mean minimum 

 have tended to obscure this high correlation, a high mean minimum 

 tending to occur with high mean maximum but with low sunlight. The 

 third partial correlation, that between mean minimum and sunlight for 

 constant mean maximum is strongly negative, and explains how the 

 other two correlations have been obscured. The weeks of high sunshine 

 are evidently the weeks of much radiation and night cooling; this 

 partially counteracts the effect of high day temperatures produced by 

 sunlight. 



The small effect of sunshine on the range of mean minimum tem- 

 peratures is brought out in Fig. 3, which shows that with considerable 

 increase in total sunshine for a specified period (11 weeks) the highest 

 mean minimum temperature is very little raised and the total range of 

 weekly mean minima is little affected, at the same time that the highest 

 mean maximum is pushed up many degrees and its total weekly range 

 is greatly extended. It will be shown later that this relatively constant 

 range of mean minimum temperature over prolonged periods is of great 

 importance to the healthy growth of plants under these conditions of 

 life, and that a comparatively slight increase above a certain level is 

 very detrimental, but fortunately does not often occur. 



Method. 



It was necessary to select a species of plant for experiment that would 

 grow under water culture conditions throughout the year and would 

 not show too great a variation in the growth of individuals of the same 

 age from a single sowing. Several years experience had shown that 

 barley and peas are among the most satisfactory subjects for treatment 

 by this method. In many respects barley is the better of the two. as if 

 a pure line is used and the seeds are graded, very even growth can be 

 obtained. Unfortunatelv during the wintci' months, in the "dead"' 



