84:8 Dr. C. V. Burton : An Experiment indicating 



as eyepiece, the probable error of a single setting was reduced 

 to 0*05 second. 



12. The first plate, when suspended, had a complete 

 torsional period of 8'5 seconds, and the readings obtained 

 were sufficiently consistent to make the full optical sensi- 

 tiveness significant. The second plate was so suspended as 

 to have a complete torsional period of 58 seconds, and in this 

 case the accidental error was seen to be considerably greater 

 than the error of setting. Everything was done which 

 suggested itself as likely to reduce thermal disturbances 

 within the suspension-chamber ; in addition to the successive 

 protecting envelopes already referred to, a stout wooden 

 screen was used to intercept radiated heat from the observer 

 or from the lamp, the source of light being a very small 

 acetylene flame enclosed as completely as was possible, and 

 the slit being of necessity exceedingly narrow. Several stops 

 were also used to ensure that no light outside the limits of 

 the useful beam should reach the suspension-chamber ; and 

 finally a shutter was arranged to cut off all light from the 

 apparatus except when the field was being actually observed. 



13. It may be remarked that, with such high magnification, 

 and especially when tha cylindrical form of eye-lens is used, 

 the demands on the optical quality of the suspended mirror 

 are severe, and the polishing was found very troublesome. 

 The thinness and consequent flexibility of the plate made it 

 necessary to do all the grinding and polishing on dry tools, 

 and considerable practice was needed before the correct 

 manipulation was obtained. The chief difficulty, however, 

 was in testing the figure of the polished or partly polished 

 surface by applying it to a quartz plane ; for the plate very 

 readily attached itself limpet-wise to the quartz, its flexibility 

 enabling it to come into close contact, even when the form 

 of the unstrained surface was far from flat. But these 

 difficulties were at length overcome, and satisfactory flatness 

 obtained. The second plate (fig. 2) was the one used in 

 conjunction with a cylindrical eye-lens, and the only defect of 

 figure which it finally showed was such as would have been 

 produced in a naturally flat plate by holding it at top and 

 bottom and minutely wringing it about a vertical axis. This 

 caused the image of a vertical slit (viewed through a 

 cylindrical eye-lens) to be slightly inclined to the vertical ; 

 a trouble which is not to be abated by any tilting of the slit, 

 the fiducial wire, or the eye-lens. The difficulty, however, 

 was readily removed by placing close to the object-glass a 

 very weak piano-cylindrical lens which could be turned round 

 in its own plane until the desired compensation was effected. 



