280 BRIDGMAN. 



as a function of the resistance may be readily calculated by Kirchhoff's 

 laws, and such corrections were applied in all the computations. 



It is not worth while to go into greater detail about this electrical 

 installation, except to say that in designing it and setting it up I 

 profited much by suggestions in various papers of White. ^ All leads 

 were of copper from the same spool. All connections were either 

 soldered (with pitch) or were the clothes pin type of spring clip, with 

 the in-leading wire thermally protected by wrapping with copper 

 strip, as advocated by White. All switches, both single throw and 

 reversing, were of the jack knife type, and of all-copper construction. 

 All the connections and switches, except the pressure connections 

 already described and the connections to the galvanometer binding 

 posts, were inclosed in a single covered box, to secure temperature 

 equality, and all switches were operated through the sides of the box 

 with rods. The reversing switches were so connected together with 

 a sliding rod as to be operated by a single push or pull, and were so 

 related in phase than an unbalanced e.m.f. was never thrown on the 

 galvanometer at any stage of the reversal. The copper resistance 

 coils of 10, 1, and 0.1 ohms were immersed in an oil bath of Bureau 

 of Standards resistance oil inside the same box as that housing the 

 switches, and the oil was stirred and the temperature read with a 

 thermometer reaching through the walls of the box after every read- 

 ing of e.m.f. The resistance of the copper coils was measured at 22^^ 

 on a Carey Foster bridge against coils calibrated at the Bureau of 

 Standards. The temperature coefficient was not measured, but was 

 assumed to be 0.00382 at 22°. This corresponds to copper of con- 

 (hictivity 98^. If the copper were actually anywhere from 96% to 

 100% conductivity, which is greater than the variation met with in 

 commercial copper, the error so introduced would remain less than 

 0.1% o^■er a temperature range of 10°, which was the maximum varia- 

 tion in room temperature during this work. 



In addition to all these, several coils not already mentioned were 

 introduced either in parallel or in series with the galvanometer so 

 that its sensitiveness could be appropriately varied. 



The galvanometer itself was protected with a shield of heavy sheet 

 iron heavily covered on the outside with cotton. Internal thermal 

 e.m.f. in the galvanometer was never entirely absent, rising sometimes 

 to as much as | X 10"^ volts, but the method of measurement, reversing 



5 W. P. White, Jour. Amer. Chem. Soc. 36, 1856-1868, 1868-1885, 2011- 

 2020, 2292-2313, 231.3-2333, 1914. 



