142 



REPORT 1892. 



Table IV. 



On the other hand, from comparisons of the sum of No. 22 and 

 No. 23 with four different mercury standards (I., II., III., and IV., each 

 of about 1 ohm), I proved that the absolute values of the two standards 

 had remained constant within the errors of observation, as the following 

 numbers show : — 



The ' patent-nickel ' would therefore be a material well fitted for 

 resistance-coils. A large number of alloys were also examined, consisting 

 of nothing but nickel and copper. An alloy containing about equal 

 amounts of each metal was found to have an extremely small tempera- 

 ture-coefficient, the latter amounting to about 0-003 per cent, per 

 Centigrade degree as against 002 per cent, for patent-nickel. Unfortu- 

 nately, however, the thermo-electric effect of these alloys against copper 

 is very high. For the alloj^ just mentioned (consisting of 50 per cent. 

 Ca and 50 per cent. Ni, called ' constantan ') it amounts to nearly forty 

 microvolts per degree Centigrade, considerably surjjassing the fchermo- 

 electromotive force of most of the usual thermo -junctions, like iron — 

 German- silver, for instance. This high thermo-electric effect evidently 

 constitutes a considerable drawback, as the connecting pieces have to be 

 made of copper. 



On the whole oar experience has led us to the conclusion that for 

 standards such alloys do best which, besides copper and nickel, also 

 contain manganese. A few years ago Mr. Weston, of Newark, U.S., 

 discovered that alloys containing manganese possess a very small tem- 

 perature-coefficient, and that it is even possible to obtain metals with 

 negative temperature-coefficient in this way. I am not aware how far 

 this discovery has been practically taken advantage of in the United 

 States. After hearing of Weston's observation the further investigation 

 of manganese alloys was taken up at the Reichsanstalt, and we obtained 

 very interesting results. 



