388 
DR, P. E. SHAW ON THE NEWTONIAN CONSTANT OF 
Table IV., vary for spheres from 209 to 175. This is due to the fact that as more 
wrappings are used, the masses, M, M, can he swung through less angle ; hence the 
range is reduced. Thus we have some eight degrees of wrapping indicated in the 
above table. 
(e) The outer layer of cotton wool was covered with tin-foil, and several experiments 
were performed in which the parts of the helix coverings, near the hot spheres, M, M, 
were sprayed with tap water every two minutes. The experiments on dates 18th 
August, 1915, and 19th August, 1915, in Table IV., were conducted thus. In this 
way the helix coverings remain cool always. 
Seeing then that similar results are obtained under all these varying conditions as 
to temperature and conductibility of the surface of the central tube containing the 
vacuum, it seems permissible to state that the effect observed is not attributable to 
radiometric pressure or other forces due to the entrance of heat through the walls 
of the vacuum tube. A further reason why radiometric pressure should be counted 
out is that like effects have been found for vacuum pressures varying from 2 mm. 
to 14 mm., so that the corresponding radiometric pressures, if existing, would drop 
in the ratio about 5 to 1. Yet Table IV. shows no material difference in the 
results throughout these changes. 
In like manner, none of the other possible errors, scheduled above, seem capable of 
giving the observed temperature effect. So, unless some other error has been over¬ 
looked or some agency, at present unknown, comes into play, we must conclude that, 
at least for lead there is a temperature effect of gravitation. The plotted results in 
fio-s. 13 and 14 do not lie on smooth curves. But it will be observed that the scale 
© 
of ordinates is very open. 
Still, provisionally, the foregoing numerical results have been worked out on the 
basis of a linear relation, 
f = G (l+aO) Mm/d 2 
Where a is a temperature coefficient of amount 
+ 1‘2 x 10- 5 per 1° C. 
It will be observed that my results overlap the weight experiments of Poynting 
and Phillips (§ I., 3) for range of temperature 20° C. to 100° C. I cannot speak 
with any certainty as to the effect in my apparatus for this small range. If the 
effect be linear, the scale movement would be under 0‘2 mm., i.e., too small an amount 
to be sure about. But the effect may increase much faster than temperature, in which 
case one could not expect to observe any result for a range 20° C. to 100° C. It 
must also be remembered that (as stated in § II.) these experiments are not strictly 
comparable with the weight experiments, and so, for various reasons, we see that the 
two results, which appear to differ considerably, do not necessarily clash. No other 
investigations on the subject have yet been made. 
