DR. P. E. SHAW ON THE NEWTONIAN CONSTANT OF 
386 
(5) If the temperature of the quartz torsion fibre rises during the experiment, its 
rigidity will change and the telescope readings will be affected. 
In Table III. the indicated temperature change in column 9 is 0'2 C C. from 
beginning to end of the experiment. There is reason to think that this change is 
almost entirely due to heat getting into the thermometer above the water cooler 
where it would not affect either the torsion fibre or the lower part of the suspension. 
But suppose for present purposes we allow a temperature rise of 0'3° C. Fuzed 
quartz has temperature coefficient — lxl0~ 4 ; thus the fibre would be stiffened by 
0'3xl0 -4 . Our temperature effect is 20xl0~ 4 . So we see that this error maybe 
considered negligible numerically and it acts in the wrong direction. 
The following list comprises the sources of error mentioned here and in § IY. The 
first four are due to heat entering the vacuum tube, the last six to the effects of heat 
on the masses, M, M, and other parts external to the vacuum :— 
1. (Temperature change in fibre). 
2. (Convection). 
3. Radiometric pressure. 
4. Radiation pressure. 
5. (Electrostatic forces). 
6. Magnetic forces. 
7. (Movement of base of vacuum tube). 
8. (Movement of top of vacuum tube). 
9. (Rotation of M, M). 
10. (Rise or fall of M, M, due to expansion.) 
The numbers of this list enclosed in brackets have already been dealt with and 
may be considered to have been eliminated by precautions already indicated. Nos. 3 
and 4 may be grouped together as they will invariably act in conjunction ; thus we 
have two errors left to deal with. 
No. 6. —The susceptibility of purest lead is — 1 x 10~ 6 ; that of iron for weak (earth) 
fields is, say 10. Thus a trace of iron, 1 part in 1,000,000, would mask any magnetic 
effect of the lead. Commercial lead has traces of iron varying from 1/300,000 to 1 /5000 
according to the source. # Suppose the lead used for M, M is of the worst commercial 
quality. The spheres, M, M, each weigh 50 kilos, so that a rod of iron 20 cm. long- 
weighing 10 gr. should have a greater magnetic influence than the presumed iron 
impurity in the lead. I placed iron rods of this mass in each sphere, and found the 
spheres, thus loaded, acted appreciably in the same way, as regards temperature effect, 
as when unloaded. Thus the temperature effect 011 the susceptibility is negligible. 
We have so far considered only the large spheres, M, M ; but it is evident that if 
the small spheres, m, m, are quiteun magnetised, the susceptibility of M, M would be 
immaterial. Separate tests were therefore made on the magnetism of m, m, and these 
* See Thorpe’s ‘ Dictionary of Applied Chemistry.’ 
