390 



Special Reports 



No appreciable error is introduced through defective insulation. Shortly after the 

 coils were wound the resistance between the adjacent coils of either group was not less 

 than about 100,000 megohms. The resistance was much less some months later, when the 

 spool had long been in a very damp atmosphere, but was still far too great to make any 

 correction necessary. The original value can doubtless be restored and kept permanent 

 by thorough drying and subsequent coating with paraffin. 



The total error in the constant is probably less than 1 part in 30,000. 



The mean diameter of coils 1 and 4, which will be treated as one Helmholtz pair, is, 

 :iccording to Table I, d = 299.028 mm. at 23 C, while the mean axial distance between 

 corresponding turns for the same pair at the same temperature is, according to Table II, 

 a - = 149.963 mm. The number of turns N being assumed as exactly 20, equation (5) 

 gives for the constant G,, of this pair, at 23 C, 



(? 14 = 6.00316 



gauss 



electromagnetic unit current 

 In the same way we obtain for the constant G 2i of coils 2 and 3 



G! 23 = 6.00291 gauss 



electromagnetic unit current 



and for the coils connected in series the constant 



(9) 



(10) 



G = Gh+G^ 12.0061 



gauss 



electromagnetic unit current 



According to these equations, G u is greater than 

 G u by about 1 part in 24,000. As an experimental 

 check, the instrument was set up as a tangent galvan- 

 ometer, with the axis of the coils in the magnetic prime 

 vertical, and double deflections were observed (1) when 

 a current of 0.015 ampere was sent through the coils 

 1 and 4 in series in the same direction, and (2) when 

 a current of 1.50 amperes was sent through the system 

 with coils 1-4 opposed to coils 2-3. Deflections of 245 

 divisions in the first case and 1.4 divisions 

 in the second case showed that G u exceeded 

 G 2i by 1.4 parts in 24,500. 



The discrepancy between the calculated ^1 ____). r _ ^ 



and experimental differences, only 1 part in 

 60,000 of the constant for a single pair, is 

 exceedingly satisfactory. 



From equation (8) and the mean tern- 



(ID 



perature coefficients 



Wi-C 



= +7.4X10- 



as follows from 14, and ( +10 X 10~ B , 



given in 16, we find for the temperature coefficient 

 of the constant G 



*(tL- 



-9x10- 



19. Let us suppose that the instrument is 

 correctly leveled and that the coil, with current Fm " 1? " 



^-magnet 



coi l 



-Angles Involved in the Theory of the 

 Sine Galvanometer. 



