Sine Galvanometer 



387 



scope settings were then made on both sides of each wire, and the coil was shifted until 

 the scale came into position. The slide was then adjusted until two divisions of each 

 scale (150 mm. apart) came to the centers of the fields (or approximately so) and mi- 

 crometer settings made upon them. From the two sets of readings the axial distance 

 between the wires was readily determined. 



Systematic measurements were made in this way along eight lines parallel to the 

 axis of the cylinder and distributed uniformly around its surface with an angle of 45 

 between successive lines. One set of lines, 90 apart, is designated as A, C, A', C; the 

 other, and intermediate, set, as B, D, B', D'. These measurements were made at tem- 

 peratures not differing greatly from 23 C; a few additional measurements at tem- 

 peratures about 5 or 6 lower on two pairs of wires, along the lines A, C, A', C, made 

 it possible to determine the axial temperature coefficient of expansion with sufficient 

 precision. 



The axial distance at 23 C. between corresponding wires for the two pairs of Helm- 

 holtz spirals, 1-4 and 2-3, is given in Figure 16 for each of the eight lines as a function 

 of the spiral number from one end of the coil. 



The mean axial distances at 23 C. for all the lines, together with the average de- 

 partures, and the final means for the two Helmholtz pairs are given in Table II. 



Table II. Axial Distances. 



Measurements along line C" were made both by Dr. Ives and by myself. Our means 

 agreed within less than 1m, while our average difference without regard to sign was 

 between 3m and 4m. The departures in the second and third columns are considerably 

 greater than the experimental error. The close agreement in the case of each pair of 

 coils between the means for the two sets of lines shows that it was unnecessary to increase 

 the number of lines. 



The mean coefficient of axial expansion was found to be + 10x10"* per degree C, 

 the average departure of the individual values for the four lines A, C, A', C, from this 

 mean being 3 X 10" 6 . As remarked above, such variations are to be expected with 

 marble. A part of the discrepancy is doubtless due to the experimental error. 



In making the axial measurements each pair of wires was referred to a different 

 pair of lines on the standard scale. The distance between each of these pairs was cer- 

 tified as correct within 2m, and as probably correct within a much smaller amount. From 

 these facts and the way in which the measurements were made, it is not probable that 

 the error hi the final means of Table II is greater than 3m or 4m. 



17. In testing for magnetic impurities the materials used in the construction of the 

 instrument, three methods were used. 



