360 Mr. C. Vernon Boys [June 8, 



maximum effect, and if the levels of the gold balls are not precisely 

 the same as those of the lead balls, again the departure is from a 

 position of maximum effect. All these three eccentricities can be 

 determined with an accuracy of 1/1000 inch. Errors of 1/100 inch 

 would make a barely perceptible effect upon the result. The design, 

 therefore, is such that a great number of measures which are difficult, 

 and can at the best only be made with a second quality degree of 

 accuracy, are of so little consequence that this degree is more than 

 abundant. The final result depends directly upon a few measures 

 which, as I hope to show, can be made with facility and most 

 accurately. These are the horizontal distance from centre to centre 

 of the wires by which the lead balls are suspended, the horizontal 

 distance between the centres of the quartz fibres by which the gold 

 balls are suspended, the angle through which the mirror is deflected, 

 the masses of the lead but not of the gold balls, and the natural time 

 of oscillation of the mirror when the balls are suspended and when a 

 thin cylinder of small moment of inertia, but of the same weight as 

 the balls, is suspended axially in their stead. 



Before going more into detail and showing how the operations are 

 carried out so that all the quantities may be known with a sufficient 

 degree of accuracy, it will be convenient to project upon the screen 

 a drawing of the vault in which the experiments have been made. 

 Prof. Clifton has kindly allowed me the free use of the vault under 

 the Clarendon Laboratory at Oxford. This is shown in Fig. 2, of 

 which the upper portion represents an elevation and the lower part 

 a plan. The instrument itself stands upon the table A x in the 

 corner, where it is screened from temperature disturbances, which 

 my presence in the distant corner and a very small flame produce, 

 by an octagon house of double wood lined with cotton-wool, and by 

 double felt screens f x f 2 . On the second table A 2 are placed a large 

 astronomical telescope T, through which the large scale S is seen by 

 reflection from the mirror in the apparatus, a small reading telescope 

 t to read the angle of the lid and vernier, a pulley-wheel p 1 and a 

 driving-wheel d. The pulley-wheel <p± keeps the cord b which 

 passes round p 2 and p 3 , and is attached to the cart g, always tightly 

 stretched, so that the observer at the telescope can always keep a 

 little flame carried by the cart immediately behind the particular 

 division under observation. The driving wheel d is made with a 

 very large moment of inertia, and the handle is near the axis, so that 

 its motion is necessarily steady. A very light cord passes round 

 this, across the room, and after passing through a hole in the screen 

 passes also round the little wheel D, Fig. 1, and thus serves to drive 

 the train W W, and so carry the lid and balls round almost in- 

 sensibly. Two hundred and thirty turns of d are required to move 

 the lead balls from the -f- to the — position. I generally turn the 

 handle a hundred and thirty times, and then when the mirror is 

 approaching an elongation, turn the handle the remaining hundred 

 times, finally stojiping when the lid reading, as observed in the small 



