156 



Messrs. Leake, Leventhorpe, and Whitehead. [Dec. 5, 



After describing Lord Kelvin's method of multiplying a comparatively 

 small known potential difference by means of a series of condensers, a 

 full account is given of the method adopted by Prof essor Ayrton and 

 Mr. T. Mather for the calibration of their electrostatic voltmeters. 

 The apparatus is arranged for the determination of alternating 

 potential differences up to several thousand volts, either by measure- 

 ment of the current passing throngh a non-inductive manganin resist- 

 ance of about 20,000 ohms, or by comparison with, the steady potential 

 difference of the Clark standard cell, in a way that does not involve 

 any measurements of resistance. A full description is given of the 

 apparatus employed, and the methods of using it. It is probable 

 that the measurements made in this way of potential differences of 

 about 2000 volts are correct to about y 1 ^ per cent, in absolute units. 



In Part II the authors discuss the measurement of high potential 

 difference in terms of the absolute electrostatic unit by means of the 

 Kelvin absolute electrometer. With the object of determining the 

 degree of accuracy in the indications of this instrument, a sensitive 

 Ayrton-Mather electrostatic voltmeter was calibrated both in electro- 

 static and in electromagnetic units, in the former unit by the absolute 

 electrometer, and in the latter unit by the method described in Part I. 

 A comparison of these two calibration curves yielded values for the 

 ratio of the electrostatic unit to the electromagnetic unit of potential 

 difference; and, by comparing these values with the mean of recent 

 independent researches on the ratio of the units (2"998 X 10 10 cm. per 

 second), an approximate idea could be formed as to the accuracy of 

 the measurements made by the absolute electrometer. By experi- 

 ments such as this it was found that the results ^iven by the absolute 

 electrometer were always too large by an average error of 1^ per cent. 



An experiment made on the coach-springs, with all electrical effects 

 eliminated, showed that the cause of this error layin a gradual alteration 

 in the strength of the springs during the electrometer measurements. 

 The main part of this alteration in strength was due to the gradual 

 yield on loading the springs, and, in a long series of experiments 

 made subsequently, this difficulty was overcome by keeping the 

 springs constantly loaded (except during actual use) with the weights 

 required for the measurements. 



The remaining part of the alteration in strength was due to change 

 in the temperature of the coach -springs, and the error thus caused 

 is of importance if maximum accuracy is essential. 



Three methods are enumerated by which this temperature error 

 can be overcome, the first of which was not, at that time, tried 

 because it would have involved considerable structural alteration to 

 the instrument. The second method, which consists in enclosing the 

 whole instrument in a non-conducting case to minimise the rate of 

 change of temperature, was tried, but was not found to be sufficiently 



