432 THE BRITISH ASSOCIATION 



that the temperature had remained constant to o-oi° C. for several hours 

 before each measurement. The mercury thermometer used was graduated 

 in hundredths of a degree, and was caHbrated on the hydrogen scale. 



A factor of some importance is the heating effect of the measuring current. 

 This has, of course, long been known, but as the magnitude of the effect is 

 surprisingly large for some of the coils, and as no record of it appears to 

 exist, it was measured in each case. The procedure was as follows : The 

 galvanometer circuit was kept permanently closed, and when balancing the 

 bridge the current was reversed, but was allowed to flow for as short a time 

 as possible. Readings were obtained, first with as small a current as would 

 give the required sensitivity, then with rather larger currents. The rela- 

 tion between bridge reading and the square of the current strength was 

 always approximately linear, and thus the resistance corresponding to ' zero 

 current ' and to any other values of current was readily obtained. The 

 actual value of current used in the older comparisons is not known, but the 

 usual practice was to connect two quart size Leclanche cells in series with 

 the bridge, which had an over-all resistance of a little more than i ohm, 

 when I -ohm coils were being compared. A trial experiment has shown that 

 the total current obtained in this way is about 0-5 ampere, which means 

 that the current in each coil was about 250 mA. This probably represents 

 the maximum current used in the older comparisons. Its heating effect 

 has been found to cause an error of as much as 5 parts in 10,000 for certain 

 of the coils, but there is no doubt that in the more recent comparisons, par- 

 ticularly those of 1908, there is no question of an uncertainty of this magni- 

 tude. An examination of the results suggests that a current of the order 

 of 120 mA. was used on this occasion, and for purposes of comparison this 

 will be assumed to represent the standard condition. The results of the 

 1932 measurements given in Table V show the values for ' zero current,' 

 120 mA.,and the maximum correction, i.e. the difference between the values 

 for ' zero current ' and 250 mA. The values obtained in 1908 are given 

 alongside of those for 1932 with 120 mA. for the purpose of comparison. 



A glance at Table V is sufficient to show that there have been no very 

 large changes in the values of the coils in the last twenty-four years. In all 

 cases the change is less than 2 parts in 10,000, which must be considered very 

 satisfactory behaviour, the more remarkable since, so far as is known, the 

 coils are connected to the terminal rods by means of soft soldered joints. 

 In spite of this fact and also of the presence of the paraffin wax in which 

 the coil is embedded, the coils are probably as satisfactory as standards of 

 resistance as when they were first constructed sixty-seven years ago. The 

 paraffin wax has become discoloured, being now yellow with age, and in 

 some cases having acquired a greenish tinge from its chemical action on 

 the copper terminal rods, and on this account the insulation was suspected 

 as early as 1886. However, at that time the leakage resistance between 

 coil and case was measured and found to be of the order of 8,000 to 

 10,000 megohms. The insulation was again tested in 1890 and found to 

 be several thousand megohms. Measurements made in 1932 gave the 

 following values : 



Insulation Resistance, 1932. 



