358 PROCEEDINGS OF THE AMERICAN ACADEMY. 



accuracy of one-tenth of a per cent in the measurement of four cubic 

 centimeters of a ^^jj normal solution by volumetric analysis would be 

 very good work. In the opinion of the two instructors in chemistry 

 consulted by the writer, it was considered very improbable that so great 

 accuracy could be obtained. Taking the set of observations above in 

 which the main current was 0.2820 amperes, we find that an error 

 of one-tenth of one per cent would mean possible errors of 0.004 in 

 each measurement of the y's. The difference obtained on reversal of 

 the magnetic field is 0.030. The error in w might then be 25 per cent. 

 The possible error would be greater in the case of some sets of obser- 

 vations where the amount of sodium hydroxide formed was less than in 

 the above case. In general iu the above table smaller values of q cor- 

 respond to smaller values of I. The error in the measurement of la 

 would therefore in general be greater for the series in which / is 

 small. Accordingly, the values of w obtained from the above table for 

 different values of / seem to the present writer to be not very reliable, 

 particularly in the case of the smallest values of I. There is an ob- 

 vious error in the record of the results given for the primary current 

 0.0000285. 



The present writer used in his measurements the same general 

 method as that adopted by Professor Hall and described in the account 

 of the latter's work at the beginning of this paper. 



In order to measure the transverse current with more accuracy, the 

 present writer made a sensitive Thomson galvanometer of low resistance. 



Description op Astatic Galvanometer. 



The general plan followed in constructing this galvanometer was 

 that which Paschen describes in the Zeitschrift ftir Instrumentenkunde, 

 1893. In order to make the suspension nearly astatic the more easily, 

 about sixty small magnets were used in the two systems. These magnets 

 were made of fine hair-spring. They were 2 to 3 mm. long and about 

 0.2 mm. wide. The distance between the systems was 5 cm. A mirror 

 8 mm. long, 2 mm. wide, and 0.4 mm. thick, was made from glass which 

 had been ground plane on both sides by Zeiss of Jena. A very fine 

 quartz fibre was used to suspend the system, which system as finally used 

 had in the earth's field a period of eight seconds. Number 22 copper wire, 

 Brown and Sharp gauge, was wound on a cone of 1.4 cm. height, the radius 

 of the base being l.t) cm. This was wound nearly full with wire, so that 



