190 
DE. A. VEENON HAECOUET ON THE VAEIATION WITH 
By and by the contents of the reagent tube were only a little darker than the pale 
standard, then the second observation was made in the same way. The comparison 
was more difficult because the rate of change was less. It was found helpful to have 
two similar standard tubes, one on each side of the reagent tube. Generally speaking, 
the first observation was not liable to an uncertainty of more than 10 seconds ; but 
when the rate was slow the uncertainty attaching to the second observation extended 
over a minute or more. By a repetition of a set of observations or of a particular 
observation, and taking the mean of the Intervals observed, the error is lessened; 
but the method does not admit of an accuracy approaching that of observing the 
appearance in a colourless liquid of the intense colour of the iodide of starch, whose 
results led to the discovery of what we believe to be a general law. The present 
observations furnish only another case of agreement within the limits of experimental 
error with numbers calculated from the formula already established. 
The following are the times in which at different temperatures the depth of colour 
changed from that of the dark to that of the pale standard, that is to say, in which a 
definite piece of chemical work was done 
Temperatures . 
9° 
12° 
15° 
18° 
21° 
24° 
27° 
CO 
0 
0 
Time 
[Set I.. . . 
. 50-0 
37-25 
27-75 
20-5 
15-17 
in 
8-62 
6-5 
in < 
Set II. . . 
. 447 
34-2 
26-2 
18-2 
14-5 
10-83 
7-83 
5-8 
minutes 
^Set III. . . 
. 47'0 
33-3 
237 
18-67 
14-0 
10-33 
7-5 
5-92 
Mean . 
. 47-2 
34-9 
25-9 
19-1 
14-6 
10-8 
8-0 
6-1 
The corresponding numbers calculated by the method explained below, and set 
forth in the following table, are 
47-2 34-9 25-9 19-3 14-4 WS 81 Gl. 
Using the formula xjx' = (T'/T)’”, x, x' being consecutive observed values of the 
times at temperatures T, = 273 + ^, and U, = 273+ ^', a value of log is calculated 
from the derived formula 
log log xjx'—\og log T'/T = log m. 
In the table, column 7 contains the values of log m calculated by subtracting the 
values of log log T'/T in column 6 from the corresponding values of log log xjx' in 
column 5. The mean of these values of log m is taken to be the true value of log m. 
Assuming this value of log m, values of log log xjx' are calculated from the formula 
log log xjx' = log log T'/T + log m. 
Column 8 in the table contains the values of log log xjx' calculated by adding the 
