DlAMAGNETIC CONSTANTS OF BlSMUTH AND CALC-&PAR 195 



results because of the iron contained in them as an impurity. The 

 bars were each about 15 mm. long and about 2 mm. in cross section. 

 The force to be measured being only about -00000001 of that exerted in 

 the case of iron it was necessary to carry out the experiments with the 

 very greatest care. 



In order to obtain bars free from iron, very fine crystals of chemically 

 pure substances were selected and the bars cleaved from them. They 

 were then polished with their various sides parallel to the cleavage 

 planes by rubbing on clean plates of steatite with oil. In order to 

 remove any particles of iron that might have collected upon them 

 during these processes, they were carefully washed with boiling hydro- 

 chloric acid and with distilled water and then wrapped in clean papers, 

 and never touched except after washing the hands with hydrochloric 

 acid and distilled water. 



In order to reduce to a minimum the causes that might interfere 

 with the accurate determination of the times of vibration of these bars 

 the poles of the magnet were encased by a box of glass. From the top 

 of this a tube four feet long extended up toward the ceiling, and inside 

 this was hung a single fibre of silk so small as to be barely visible to 

 the naked eye. The bars were placed in little slings of coarser silk 

 fibre and suspended by this. Outside the glass case was a microscope 

 placed horizontally and having a focus of about six inches. This was 

 directed toward the suspended bar, and when the latter was at rest the 

 cross hairs of the microscope fell upon a little scratch in one end of the 

 bar. Near by was a telegraph sounder arranged to tick seconds. The 

 bar was set swinging through a small arc by making and breaking the 

 current, and the interval between two successive transits of the little 

 scratch on the bar by the cross hairs of the microscope was measured 

 in seconds and tenths of a second by the ear. By keeping count through 

 a large number of successive transits the time of a single swing could 

 be determined with very great accuracy. The bar was caused to swing 

 only through a few degrees of arc and such small correction for ampli- 

 tude as was found necessary was applied. The time of swing was deter- 

 mined first with the axis vertical and then with it horizontal. But 

 besides the time of swing of each bar it was necessary to measure : the 

 length ; area of section; moment of inertia in each position ; and for the 

 calc-spar bar the angle it made with the equatorial plane of the magnet 

 when in its position of equilibrium. This was not necessary in the 

 case of bismuth, because its position of equilibrium lay in the equatorial 

 plane. 



