that Matter takes up no Room in the yjEther, 847 



plane ^ arrangement were formed of cast iron, and (like the 

 corresponding parts of the bedplate) were machined. They 

 were cemented to stone slabs, carried in turn upon 12-inch 

 socket-drain-pipes which stood vertically upon the cement 

 floor; all contacts being of course " geometric." By the 

 mode of support just described, the position of the bedplate 

 was positively determined, and on the other hand, it was 

 impossible for small movements of the seating (due e. g. to 

 temperature changes) to impose any appreciable stresses on 

 the bedplate. Such movements, taking place slowly, might 

 conceivably produce minute tilts or azimuthal displacements 

 of the apparatus as a whole, but the effect of these on the 

 micrometer-readings was not detectable. 



10. The optical system used for reading the excursions of 

 the mirror-polished plate has already been described under 

 the name of "micro-azimometer"*, and need only be briefly 

 referred to here. The arrangement was an auto-collimated 

 one, with an achromatic lens fixed before the window of the 

 suspension-chamber, and a fine vertical illuminated slit in 

 the principal focal plane of the lens. The image of the slit 

 formed by light which had passed twice through the lens, 

 and had been intermediately reflected by the suspended mirror, 

 was viewed through an eyepiece which latterly consisted of a 

 single cylindrical lens (not achromatized) with generating 

 lines vertical. This had a focal length of about 1*5 cm., and 

 the image of the slit thus observed was sufficiently super- 

 magnified to show the diffraction pattern on a fairly open 

 scale. The position of the slit was micrometrically adjustable, 

 and the image could thus be brought into agreement with a 

 fixed vertical wire, upon which the eyepiece was focussed. 

 The thickness of the wire was such that, when the setting- 

 was correct, the central band of the diffraction pattern was 

 almost wholly eclipsed, leaving only a narrow residual strip 

 of light visible on either side of the wire. A very slight 

 deviation from central setting caused these residual strips to 

 appear of unequal brightness, so that in favourable circum- 

 stances very close settings could be made. 



11. Tests were made of the delicacy of the optical system 

 by fixing the little polished plate, and observing how con- 

 sistently the micrometer controlling the position of the slit 

 could be set and reset. It was thus found that, with the first 

 mirror (fig. 1) and using an ordinary eyepiece, the probable 

 error of a single setting (expressed in terms of the azimuth of 

 the mirror) was rather less than 0*5 second of arc; while 

 with the second mirror (fig. 2) and using a cylindrical lens 



* Phil. Mao;. March 1912. 



