August 9, 1894] 



NA TURE 



36; 



and no multiplication of observations or taking of means would 

 have eliminated it. It is on this account that it is so important 

 in experimental work to vary the conditions in every way, so as 

 to discover unsuspected consistent errors. 



The microscope scale was made by Zeiss, and is a most 

 perfect example of scale construction. In order to test the 

 •accuracy and find the errors of the scale, I took advantage of my 

 visit to CardilT, for the meeting of the British .Association, to 

 -compare it with a series of Whitworth standard bars on Prof. 

 Viriamu Jones's very perfect Whitworth measuring machine. 

 for this class of work sunshine or dust give great trouble, but I 

 was fortunate in having splendid weather for my purpose, as 

 visitors will probably remember. It rained without ceasing 

 ■during the two days that I was making these measurements. 



Having now very imperfectly described the apparatus and 

 <he place in which I have carried out my experiments, I will 

 rnext show a series of photographs, which I took by magnesium 



tube, which tube is also seen in Fig. i. This tube en- 

 ables me to control the motion of the mirror from the 

 telescope without approaching the corner in which the ap- 

 paratus is set up. This is done as follows : the back window 

 at the level of the mirror is made of metal, with a hole in it in 

 which is screwed a metal tube lightly filled with cotton wool. 

 This is not central, but opposite one end of the mirror. The 

 tube on the tabic does not fit the screw, but is merely bent up 

 and enters it loosely. By gently drawing air from the end of 

 the tube at the telescope a very feeble draught is produced in 

 the apparatus, for nearly all the required air is supplied by 

 leakage round the pipe near the screw, very little entering 

 through the window pipe, in consequence of the resistance 

 offered by the cotton wool. In this way, if the mirror is 



light, to give a better idea of the appearance of the apparatus 

 and its surroundings. Fig. 5 is a view of the vault showing 

 <he clock, the eye end of the big telescope, and the little telescope. 

 In the distant corner is seen the felt screen with a long slit, 

 through which the scale and telescope can be seen from the 

 mirror of the instrument. This, of course, is on the table be- 

 hind the screen. Fig. 6 is a view of the corner itself, with the 

 screen drawn back. The octagon protecting house, which sur- 

 rounds the apparatus, is seen in position. Here again a slit is 

 cut large enough fo allow the scale and telescope to be seen 

 from the mirror. Fig. 7 is a view of the instrument with the 

 two halves of the octagon house separated. Here a further 

 system of screens consisting of concentric brass tubes may 

 be seen, but the lower one, which surrounds the window, 

 has been removed and placed upon the table. The driving 

 gear is also seen in this photograph, and a tube coming 

 ■from'the screw under the instrument which holds the central 



Fio. 7. 



moving it may be gently brought to rest without impact, or it 

 may be given a swing of any desired amplitude. So perfectly 

 does this work, that the mirror may be steadied very 

 quickly so as to move through less than a scale division, 

 an amount which corresponds to six or seven seconds of arc, or 

 to a force of less than one thousand millionth of the weight of 

 a grain. 



The operations for any complete experiment are fourteen in 

 number. I do not intend to go through these seriatim, as 

 time will not allow me to do so. It is sufticient now to say 

 that the first eight are necessary to get the instrument and scale 

 relatively fixed and .adjusted, the vertical measures made, and 



NO. 1293, VOL. 50] 



