August 2, 1894] 



'NATURE 



333 



The driving wheel d\% made with a very large moment of inertia, 

 and the handle is near the axis, so that its motion is neces- 

 sarily 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. i), and thus serves to drive 

 the train w \v, and so carry the lid and balls round almost in- 

 sensibly. Two hundred and thirty turns of J are required to 

 move the lead balls from the + to the - position. I generally 

 turn the handle 130 limes, and then when the mirror is 

 approaching an elongation, turn the handle the remaining 100 

 times, finally stopping when the lid reading, as cbserved in the 

 small telescope, is correct. The large scale, s, is 9 feet long, and 

 is divided into 50ths of an inch. There are 4800 divisions. 



Two beams, /j /,, are seen in Fig. 2. The upper surfaces of 

 these are straight, and are adjusted by screws until they are 

 truly level. These are used when the true optical distance from 

 the mirror to the scale is being measured. A steel tape, on 

 which I engraved a fine line near each end, rests upon the 

 beams. At one end a slider carrying a microscope is placed so 

 as to see a fine line at the centre of the mirror accurately in 

 focus, while at the other a corresponding slider is placed so that 

 a projecting brass rod rests against the scale. At the same 

 time cross lines engraved upon the plate-glass bases are placed 

 exactly over the lines engraved on the steel tape. When after- 

 wards the microscope is focussed upon the end of the brass rod, 

 the distance between the cross lines, as measured by a s^ale, is 

 the amount that has to be added to the distance between the 

 engraved lines upon the tape, in order to obtain the distance 

 from the scale to the mirror. 



Overhead wheels are shown in Fig. 2, fastened to the roof 

 above the apparatus, and again close to the end wall These 

 serve many purposes, as will appear later. Among others, the 

 middle one of each carries a cord fastened at one end to a 

 crossbar joined at its ends by guys to the pillars R of the 

 lid (Fig. I), and at the other to heavy balance-weights to 

 counterbalance the balls M M and part of the lid. Thus the 

 friction is greatly reduced, and the tremor set up by rotating 

 the lid is in a corresponding degree slight. 



All time observations are made chronographically upon a 

 drum by the Cambridge Scientific Instrument Company. This is 

 placed in the adjoining vault. Two time-markers record with 

 their points less than i/ioo inch apart, one of them marking 

 every second of the clock, with special marks for minutes and 

 half-minutes, and the other every depression of the key at my 

 right hand. The late Prof. Pritchard kindly lent me an astro- 

 nomical clock for the purpose, to which I fitted time-marking 

 contacts ; but into the details of these I must not enter. He 

 also allowed me to make use of one of his assistants to keep 

 me informed of the rate of the clock from time to time. 



I have up to the present spoken vaguely of the large lead 

 balls and of the small gold balls, but have given no indication as 

 to how they are made and how I can be sure of the truth of 

 their form and their homogeneity. Mr. Munro, whose capacity 

 for turning accurate spherical work is well known, made for 

 me two moulds of hard cast-iron, which I have on the table. 

 One of these is for a 4i-inch lead ball, and one for a 2i-inch 

 lead ball. F.ach mould is made in two halves, so truly as to 

 shape and size that the thin steel disc that was used as a tem- 

 plate would distinctly rattle when in its place, but when a strip 

 of cigarette-paper was inserted on one side it could not be got 

 in at all. The upper half of e.ach of these moulds is provided 

 with a cylindrical steel plunger accurately fitting a central hole 

 in the mould, and with its end turned to the same spherical 

 surface when it is pressed home upon its shoulder. The lower 

 half of each mould has a J -inch central cylindrical hole, 

 into which the lug of the brass ball holder exactly fits. 

 There is also a small hole at the side which can be stopped with 

 a brass plug. The balls are made as follows : — The interior of 

 the mould is smoked and then screwed up as tight as possible. 

 It is then heated until a piece of lead upon it begins to melt. 

 The necessary quantity of pure lead melted in an earthen pot 

 is then carefully skimmed and poured in until the cylindrical 

 neck is full. The mould is then made to rest upon a cold iron 

 slab, and a blowpipe is directed upon the upper part so that it 

 cools Irom below upwards, and not from the surface inwards; 

 more lead is added to keep the neck full. As soon as the lead 

 in the neck solidifies the plunger is inserted, and the whole is 

 placed in a hydraulic press. The plunger is forced down upon 

 its seat, the lead, already free from bubbles and vacuous cavities, 

 is compressed until at last the excess of solid metal flows 



NO. 1 292, VOL. 50] 



through the small side hole in the form of wire. The ball i-! 

 thus made true in form, necessarily homogeneous, which no alloy 

 is likely to be, and definite in size. When cold it can be lifted 

 from the .mould, when after cutting off the wire which projects 

 from its equator, it is ready for weighing. 



The small gold balls are made by melting the required quantity 

 of pure gold in a hole in a bath brick, and, as in the case of the 

 lead, letting it cool from below upwards, so as to avoid cavities. 

 It is then inserted in a pair of polished hemispherical hardened 

 steel dies, which Mr. Colebrook made for the purpose, and 

 beaten, being turned between each blow, and annealed once or 

 twice until a perfect polished sphere, without a mark upon it, is the 

 result. I make these in pairs of exactly the same weight, and, as 

 in the case of the lead balls, thus obtain truth of form, accuracy 

 of size, and homogeneity all in a very perfect — more than suffi- 

 ciently perfect — degree. These are each suspended from a 

 quartz fibre of the necessary length, to the other end of which 

 a hook and eye is fastened. Into the very important details of 

 these operations it is impossible, for want of time, for me to 

 enter. The gold balls are '2 and '25 inch in diameter, and a 



? 



o 



Fig. 



pair of gold cylinders were made in a similar tool '25 inch in 



diameter, and about the same length. 



Perhaps the most important detail in the whole apparatus is 

 the " beam mirror," which is of the form shown in Fig. 3. It 

 is necessary, as far as possible, to reconcile the following in- 

 compatible conditions. It should be as light as possible, 

 have as small a moment of inertia as possible, the optical de- 

 finition should he as perfect as possible, and, almost most im- 

 portant of all, the form should be such that the resistance offered 

 by the viscosity of the air should be reduced to the smallest 

 possible degree. By cutting the middle portion out of an 

 optically perfect round mirror all these conditions are realised 

 in some degree, and the optical definition is actually more per- 

 fect in the horizontal direction than that due to the whole disc. 

 This is fastened to a cross-shaped support of gilt copper. The 

 ends of the mirror have vertical grooves of microscopic fine- 

 ness cut in their thickness, so that the quartz fibre hanging from 

 the cross-arm above may rest definitely in them. The central 

 hook is for the purpose of hanging the " counterweight," i.e. a 

 slender silver cylinder of exactly the same weight as the gold 

 balls with their fibres and hooks. By this means the unknown 

 moment of inertia of the mirror may be eliminated with the 



