390 
DR. P. E. SHAW ON THE NEWTONIAN CONSTANT OF 
the window, which necessitates a break in the water-jacket system to allow light to 
enter the window and return to the telescope ; and though an elaborate system of 
w T ater helix and lagging is provided in the window region (not fully shown in any 
figure) one feels that here there is a joint in the armour through which heat may 
enter. The ideal vacuum tube would have no front window on the wall of the tube. 
The light from the scale would enter the top window and pass by two totally 
reflecting prisms to and from the mirror back to the telescope. The optical difficulty 
in realising this plan would no doubt be great, but if it were possible we should have 
the great advantage of a continuous water-jacket, without joints, from end to end of 
the vacuum tube. 
Suppose, however, the ideal arrangement could not be made to work, one seems to 
have only one alternative if the window defect, above indicated, is to be minimised. 
This alternative is to reduce the bore of the vacuum tube to, say, 36 mm., and the beam 
to, say, 25 mm. The window would be reduced proportionally and it might then be 
possible to have a continuous water-jacket with a small slit opening at the window. 
In reducing the beam as just suggested we should have the further advantage of 
reducing the period of oscillation to one half. Sensitiveness would be slightly lessened, 
but spurious heat effects, if any, would be greatly reduced. 
2. General Summary. —I. It has been found possible :— (a) to obtain consistent 
cyclic readings in a gravitational experiment of the Cavendish type, even though the 
large masses are maintained for hours above 200° C., while the small masses remain at 
ordinary temperature ; (b) to carry on this investigation in the centre of a city at any 
time by day or by night, in spite of the attendant tremors and the special disadvantage 
of having' the torsion balance in a vacuum. 
II. The conclusion reached is that there is a temperature effect of gravitation. 
When one large mass attracts a small one , the gravitative force between them increases 
by about .lj'500 as temperature of the large mass rises from, say , 15° C. to 215° C. 
At present the result is provisionally stated as being +l'2x 10~ 5 per l C. ; but the 
readings are not steady enough to justify the statement that there is a linear relation 
for G/0. It. seems possible that time may be a factor in the effect; but the net result 
has not been shaken by a long series of tests. 
III. The above result, though new, is not entirely unsupported by other experiments, 
for previous gravitation experiments give indirect evidence of a positive temperature 
coefficient . The weight experiments of Poynting and Phillips, which yielded negative 
results, are not strictly comparable with mine. 
IV. As a bye-product of these experiments, it was found that silver balls of the 
highest purity, after being heated to 130° C. and placed in a strong magnetic field, were 
permanently, though weakly, magnetised, and that the coercivity was considerable. 
This is probably due to residual iron, see § VII. 6. 
V. Several technical troubles overcome during the research are summarised 
above. 
