A NEW CURRENT WEIGHER, ETC. 487 



of 2 millims. on the scale ; a difference of I/* was thus read with ease. As there was 

 no silk on the wire to interfere with precise measurement, this high sensitiveness was 

 well worth attaining. The contact planes of the pistons were tested for parallelism 

 with satisfactory results ; tests were also made which indicated that these planes were 

 normal to the axes of the pistons. 



To facilitate the diametral measurements, the marble cylinder under observation 

 was supported on a turn-table provided with ball bearings and levelling screws 

 (fig. 11). The original intention was to support this table on a separate platform and 

 so avoid the bending of the girder which results when it supports the load ; this, 

 however, proved to be unsatisfactory, and ultimately a small wooden platform was 

 bolted to the girder, and on this the turn-table and cylinder rested. The traces of 

 the axial planes on the ends and ungrooved portions of the cylinder, and the knowledge 

 that the ends were at right angles to the axis, enabled the coils to be rapidly set in 

 position so that their axes were vertial ; at the same time the adjustment ensured 

 that the plane edges of the touch-pieces would come into contact with the copper 

 wires at opposite ends of a diameter. Two spirit levels at right angles were used for 

 the levelling of a cylinder, and it was usually found necessary to make a slight 

 adjustment for every measurement made in a different axial plane. In general, 

 observations were made in eight approximately equidistant diametral planes, and in 

 each of these, 14 measurements were taken in equidistant axial planes; at the 

 conclusion of the 14 observations the first was repeated as a check on the constancy 

 of the apparatus. The method used was not a " null" one ; the zero reading, i.e., that 

 when a mirror was against a stop, was observed from time to time, and a constant 

 deflection of 10 millims. from this was adhered to throughout the measurements. The 

 apparatus worked very smoothly, the readings being easily reproduced to I/*, and 

 only in a few cases of uncertainty was more than one observation made of any one 

 diameter. The temperature of the room was very nearly constant and equal to 

 15 0< 5 C. ; a Richard's thermograph recorded the variations. 



At the commencement of a series of measurements, tlie Whitworth steel gauge 

 (square section, flat ends) was placed in position, and the uprights carrying the 

 micrometer heads were bolted to the girder. A mass equal to that of the turn-table 

 and cylinder was next placed on the small platform between the micrometer heads, 

 and the observations on the gauge were then made, the latter being displaced and 

 reset between every two measurements. The cylinder was then placed in position 

 and measured, and afterwards the gauge was again set up. With respect to the 

 latter measurements, the difference in the readings of the micrometer heads never 

 varied by more than 2/i from the commencement to the completion of a series of 

 observations. When measuring a diameter, the touch-pieces made contact with one 

 wire of each helix and the mean of the observations gave, therefore, the mean outside 

 diameter of the two coils. To determine the difference of the mean diameters of the 

 coils, one of the micrometer heads was raised -fa inch and a few observations of 



