GRAVITATION AS AFFECTED BY TEMPERATURE. 
353 
(4) Prof. C. V. Boys kindly gave me access to the note-books compiled by him in 
his research on A at the Clarendon Laboratory. I found there six complete experi¬ 
ments in which temperature was systematically recorded. Giving all the experiments 
equal weight, we find three whose value for A is below 5'52, the average being 5‘517, 
and mean temperature 15°’9 C.; while the other four which have A above 5'52 average 
5'528, with mean temperature 13°‘8 C. This would show a plus temperature coefficient 
for G of 1/500 for 2° C. rise in temperature. In fig. 1 a graph is shown of these results. 
(5) Von Sterneck* showed that the temperature gradient at a given depth 
in a mine varies from one mine to another, and that the mean density A for the 
whole earth, deduced from his pendulum experiments in the mines, increases as 
gradient increases. This again indicates a plus temperature coefficient. When 
5-532 
I 
-i-1- 
\ 
5-528 
1 
X 
X 
5-524 
\ * 
\ 
A 
5-520 
\ 
\ 
\ 
5-516 
X 
X 
_j_ 
5^512 
_1_ 
\ 
-!-1_ 
-1- 1 -1_ 
13° 14° Temp. C. 15° 16° 
Fig. 1. Shows the relation A/0 in C. V. Boys’ experiments. 
the pendulum is swinging in a mine the strata immediately above and below would 
have an influence on the pendulum period out of proportion to their masses, on 
account of their proximity to the pendulum, consequently their temperatures would 
influence the results for A, supposing the existence of a temperature effect. Let 
us suppose a plus temperature coefficient. The strata below would always be at higher 
temperature than those above and would attract more strongly than the latter. In 
the case of high gradient this difference in attraction would be greater than when the 
gradient is low, and the result of the experiments would be an apparent value of A 
greater for high gradients than for low. 
3. Of the five sources of evidence above, one is useless from uncertainty as to 
temperature ; but in the other four, in all of which the direction of temperature 
difference is known, an apparent plus temperature coefficient for G is found. There 
* ‘Akad. Wiss. Wien.,’ 1899. 
