36 Leitch . — Some Experiments on the Influence of 
this temperature, would therefore be an analysis of less than half the 
material. Indeed, this fact, together with the fact that above 30° a few 
similar curvatures still take place, shows that this point is a critical point 
at which, even if measurements in numbers were easy, an accurate analysis 
would be made very difficult, if not impossible, by the necessity for referring 
each individual to one or other of two types, according to whether a time- 
factor is or is not operating, as will be clear later. 
Above 30° measurements are again easy ; a few curvatures still occur, 
but the proportion is small enough to be negligible— about 1 in 15. 
From 30° onwards, an entirely new set of phenomena appears. I had 
expected that here it would be possible to get by short-time readings values 
above the readings at 29 0 , and that a time-factor would be found operating 
in accordance with Blackman’s theory. A time-factor does operate, but the 
relation between the rate of growth and the time is not a simple one. Here, 
owing to the difficulty of the analysis and the necessity for many experi- 
ments, I considered it best not to experiment at random temperature inter- 
vals as before, but to restrict the experiments to a few temperatures. They 
are, approximately, 30°, 35 0 , 40°, 43-5°, and 45 0 , and Table IV gives the 
results for these temperatures. The results for 30°, 35°) and 40° respectively 
are plotted in Text-figs. 7, 8, and 9. 
For the experiments at 30° the standard deviation and probable error 
have been calculated for the readings in the first ten minutes, and for those 
in the first half-hour. They are : 
Mean . 
Standa?'d Deviation. 
Probable Error. 
No. 
First ten minutes. 
o*68 
0*128 
0-033 
15 
First half-hour. 
i-74 
0*279 
0*043 
43 
In both cases the probable error is, considering the small number of 
readings, satisfactorily small, and compares very well with the results in 
the earlier experiments, 
