126 KEPORT— 1894. 



It is clear that in the above series the chief variations are duo to 

 changes consequent on readjustment of the standard coil, and the re- 

 moval and replacement of the resistance coils in their mercury cups. 

 Counting as independent only those of the observations before which 

 there was readjustment of the standard coil or removal of the resistance 

 coils from the mercury cups, the general mean is 



•9997G. 



The maximum variation from the mean is -000106, or about one part 

 in 10,000. 



Assuming that the international ohm is the resistance of a column 

 of mercury at 0^ of 1 sq. mm. sectional area, and 106"30 cm. long, we 

 have as a result of the above measurement that the true ohm is the 

 resistance of a column of mercury of the same sectional area and 

 106-326 cm. long. 



The figure I arrived at in 1890, working direct on mercury, was 

 106-307, with a probable eri'or of ^f^-Oll. The new i-esult is therefore 

 a little larger than I was prepared for. The accuracy of the result 

 depends primarily on — 



(i) The accuracy with which the resistance coils arc known in terms 

 of the international ohm. 



(ii) The accuracy with which their temperatures are known at the 

 times of observation. 



(iii) The accuracy with which the coefficient of mutual induction of 

 the coil and disc has been determined. 



Upon the first point I can say little. Mr. Glazebrook knows better 

 than anyone to what figure the values of the resistances may be 

 relied on. 



The effect of error in estimation of the temperatures of the coils can 

 be but slight. The observations have been made in two ways, viz., with 

 one-minute tapes, the current being put on only during the time of ob- 

 servation, and with three-minute tapes, the current being kept on con- 

 tinuously, whether observations were being made or not. During the 

 last few days of the observations the current was kept passing through 

 the coils night and day. I ha\e calculated the eflect that would be 

 produced on the result olitained with one-minute tapes if all the heat 

 generated by tlie current were to remain in the coils — an extreme case, 

 obviously less favourable than the actual conditions. It is something 

 less than two parts in 100,000. The smallness of the eflfect is due to the 

 fact that if y is the main current, a current equal to |I1',' y passes through 

 the tenth-ohm manganine coil with its small temperature coefficient, and 

 only -5-^1 y through the platinum-.silver coils ; while the effect of under- 

 estimating the temperature of the manganine tenth-ohm coil is to produce 

 an error in the result opposite in sign to that produced by underesti- 

 mating the temperature of the platinum-silver coils. 



There cannot, then, in the case of the one-minute tape oljservations be 

 an appreciable error due to underestimation of the temperature. But 

 the first four sets of observations show that the results of the one-minute 

 tape observations and the three-minute tape observations are practically 

 the same. Hence it follows that to the degree of accuracy aimed at 

 our results are unaffected by error due to underestimation of the 

 temperature. 



