Reaction before Complete Equilibrium. 73 



remains suspended in the air ; Mr. Harker keeps his man- 

 ganin coils and the platinum-thermometer suspended in the 

 air. 



Care was taken to avoid corrections for thermoelectric 

 effects altogether, as these may always influence the results and 

 constitute in our case a serious source of error. Since no 

 manganin blocks could be procured, advantage was taken 

 of the observations made b) r the Berlin Reichsanstalt, that 

 copper and manganin give no thermoelectric effects, at airy 

 rate at ordinary temperatures. The whole circuit consisted 

 of copper and manganin only. In the resistance-box the 

 manganin coils were soldered to rods and thick blocks of 

 copper. The manganin leads were fastened to large flat 

 pieces of copper galvanized with tin (to secure good contact). 

 The blocks to which the manganin wires of the bridge were 

 fastened and the two contact-makers were also of copper. As 

 to the thermoelectric effects of platinum-manganin junctions 

 in the platinum coil, these were nullified by the compensating- 

 leads, and by taking great care to secure the equality of the 

 temperature of the junctions (see below). 



The Bridge. — My bridge was constructed differently from 

 those of Callendar, Griffiths, and Harker. It consists of 

 copper and manganin only, so that the thermoelectric effects 

 due to the bridge are avoided. In Callendar and Griffiths' 

 arrangements 1 or 2 cm. indicate 1° C, and for more 

 accurate readings a microscope is used. In my arrangements 

 two thick manganin wires, each 100 cm. long, are stretched 

 in the air, and 98 cm. of their length equal 0'0909 ohm and 

 0*0955 ohm. One tenth of a millimetre of my bridge indicates 

 C, 0()02 ; the readings can be made with the eye simply and 

 with certainty. The expansion of the wire under the un- 

 avoidable changes of the temperature of the room makes 

 readings with the microscope to 0*01 mm. futile, and such 

 readings seem not to be of much use where greater accuracy 

 is required, as long as the scale itself does not expand to the 

 same extent as the wire. A scale cut on a flat strip of 

 manganin and gilt could serve very well the purpose of more 

 accurate reading. For our purpose it sufficed to have a 

 wooden scale. A wooden millimetre scale instead of a metal 

 one was fixed to the wooden box in order to diminish 

 the variation of its length under the influence of the 

 temperature of the room. Two metal strips were fixed to the 

 two narrow sides of the wooden scale, and were connected 

 inside of the box to the terminal /;. the carriage with two 

 contact-makers sliding on them. The temperature-coefficient 

 of the thick wires after repeated annealing was found by 



