268 



On the Cavendish Experiment. 



[June 20, 



obtained, nevertheless the means of the summer and of the winter 

 observations differ by about 1 per cent. 



I have not referred to the various methods of determining the 

 constant of gravitation in which a balance, whether with the usual 

 horizontal beam, or with a vertical beam on the metronome principle, 

 is employed. They are essentially the same as the Cavendish method, 

 except that there is introduced the friction of the knife-edges and the 

 unknown disturbances due to particles of dust at these points, and 

 to buoyancy, without, in my opinion, any compensating advantage. 

 However, it would appear that if the experiment is to be made with a 

 balance, the considerations which I have advanced in this paper 

 would point to the advantage of making the apparatus small, so that 

 attracting masses of greater proportionate size may be employed, and 

 the disturbance due to convection reduced. 



It is my intention, if I can obtain a proper place in which to make 

 the observations, to prepare an apparatus specially suitable for abso- 

 lute determinations. The scale will have to be increased, so that the 

 dimensions may be determined to a ten-thousandth part at least. 

 Both pairs of masses should, I think, be suspended by fibres or by 

 wires, so that the distance of their centres from the axis may be 

 accurately measured, and so that in the case of the little masses the 

 moment of inertia of the beam, mirror, &c, may be found by alter- 

 nately measuring the period with and without the masses attached. 

 The unbalanced attractions between the beam, &c, and the large 

 masses, and between the little masses and anything unsymmetrical 

 about the support of the large masses, will probably be more accu- 

 rately determined experimentally by observing the deflections when 

 the large and the small masses are in turn removed, than by calcula- 

 tion. 



If anything is to be gained by swinging the small masses in a good 

 Sprengel vacuum, the difficulty will not be so great with apparatus 

 made on the scale I have in view, i.e., with a beam about 5 cm. long, 

 as it would with large apparatus. "With a view to reduce the con- 

 siderable decrement, I did try to maintain such a vacuum in the 

 first instrument, in which a beam 1'2 cm. long was suspended by a 

 fibre so fine as to give a complete period of five minutes, but though 

 the pump would click violently for a day perhaps, leakage always took 

 place before long, and so no satisfactory results were obtained. 



With an apparatus such as I have described, but arranged to have 

 a complete period of six minutes, it will be possible to read the scale 

 with an accuracy of one ten-thousandth of the deflection, and to 

 determine the time of vibration with an accuracy about twice as 

 great. 



