ON rRACTICAl, STANDARDS FOR ELECTRICAL MEASUREMENTS. -Jt 



coriuspuml with about 0"-02 C. All of the coils are surrounded by paratiin 

 wax, and it is only by maintaining a constant temperature for many hours 

 that very accurate observations can be made. The scale of temperature 

 employed for the 1908 measurements is the hydrogen scale ; that used for 

 previous observations is almost certainly the Kew glass scale. Dr. J. A. 

 Marker has recently shown ^ that the difference between these two scales 

 is negligibly small ; hence we may assume that the same scale of tempera- 

 ture has been used throughout. 



The present method of comparing the coils is by substitution in one 

 arm of a Wheatstone shunt bridge, of which the other three arms 

 consist of manganin resistances. The high-temperature coefficient coils 

 are kept in a room remaining constant in temperature to 0°'01 C. over 

 several days, and the temperature does not diifer from lG°-0 C. by more 

 than 0°-5 C. About 16 measurements, spread over several days, are 

 made of each coil, and the value at 16°-0 C. is deduced from these 

 measurements. During 1908 approximate values for the temperature 

 coefficients of resistance of the coils have been obtained by varying the 

 temperature from 14^ to 17° C. These values are given in Tables III. 

 and IX. 



In 1867 the temperatures are given at which the coils were 1 B.A. 

 unit, and this procedure was in part followed in 1876, 1879, and 1888. 

 The unit of 1867 Avas, however, probably different to those of 1876, 

 1879, and 1888. Messrs. Chrystal and Saunder (1876) assumed one of 

 the coils (B) to have remained constant between 1867 and 1876, and 

 expressed the values of the other coils in terms of it. The unit, in terms 

 of which the measurements of 1879-81 were made, is the mean B.A. 

 unit as indicated by Fleming on his chart ; it is supposed to represent 

 the mean of the resistances of the six coils A, B, C, D, E, G at the tem- 

 peratures at which they were originally correct. It is this unit which 

 was used by Lord Rayleigh in his work on the ohm, and by Dr. 

 Glazebrook since about 1880, and it has been closely adhered to in all 

 measurements made by the Committee since that date. 



A close examination of the chart at the present day shows that 

 tiie mean of the values of the six coils is really about 0"99985 unit ; 

 hence if this interpretation be accepted, the mean B.A. unit is really 

 15 parts in 100,000 less than the unit which has been taken since 

 1880 ; but it has not been thought wise to attempt any correction on 

 this score, except in the compilation of Table III. At times a sudden 

 change in a coil has been recorded, as in 1888, when Dr. Glazebrook 

 reported that F had suddenly risen in value by 0-00048 B.A. unit,- and 

 that Flat had fallen by 1 part in 10,000.^ Similar changes may have been 

 observed when the coils wei-e comparatively new, and it is possible that a 

 slightly variable coil was disregarded, or a correction applied because 

 of it, when the chart summarising the observations for 1879-81 was 

 constructed. 



The chart gives the values of the coils from 0° C. to 25° C, and the 

 graphs are such that the value of a coil can be read with an error not 

 greater than 3 parts in 100,000, which is equivalent to a change in tem- 

 perature of 0°-l C. of a platinum-silver coil. The resistances of the, coils 

 at various temperatures as given by the chart are given in Table II. 



' Proe. U.S., A, voL 78. • B.A. Report, 1888. 



» Phil. Tram., A, 1888, p. 364. 



