48 Mr. S. R. Milner and Prof. A. P. Chattock on 



respectively through the two pairs of heating-coils and sinks 

 arranged in series, and the temperatures 6 and 6' of the two 

 heating-coils brought to equality by adjusting the thickness 

 of the water layers. On now altering one of the water layers 

 by y, so that 6 (say) becomes l} the difference d x — 6\ which 

 is independent of small permanent changes in the flow of 

 either the water or the electricity, may be used to replace the 

 6 1 — 6 of equation (1) . 



The differential arrangement has the further advantage of 

 readily lending itself to the measurement of 0i — 6'. Experi- 

 ment had already shown that the thermopiles we originally 

 intended to use were not altogether suitable for this purpose ; 

 but by constructing the two heating-coils of platinum wire and 

 connecting them to form the two arms of a Wheatstone bridge 

 they constitute a differential thermometer which is the more 

 sensitive as the heating current flowing through them is 

 necessarily considerable. 



Heating- Coils. 



It has been pointed out that the flow of heat is linear only 

 at the centres of the coils : the temperatures must therefore be 

 measured in these regions. Fig. 2 shows the winding of the 

 heating-coils adopted for this purpose. Each coil consists of 

 three parts connected in series, the inner circle « ft (4 cm. in 

 diam.) forming the temperature-measuring part, and the two 

 rings surrounding it, fty and y 8 (total diam. 11*5 cm.), acting 

 as guard-rings, flexible connexions being soldered on at 

 a, ft, 7, 6\_ 



The division of the guard-ring at 7 allows of more accurate 

 measurement of the average resistance per sq. cm. of the 

 central portion of the coil than can be got from the smaller 

 area « ft. It also enabled us to test whether the area aft was 

 sufficiently small for all the lines of heat-flow from it to be 

 parallel, by observing if the temperature was the same at the 

 centre whether the current flowed through the whole coil a 8, 

 or only through xy. Most important of all, it helped to 

 prevent flow of heat in the plane of the mica, a function it 

 shared with the gap round a ft. 



To ensure that the two heating-coils were the same size, 

 holes were pricked through two sheets of mica laid one on the 

 top of the other, and pure platinum wire 0'2 mm. diam. 

 threaded through those on each sheet, so that it all appeared 

 on the same side of the mica except at the short cross- 

 connexions between the grids. These cross connexions were 

 so arranged that the amount of wire per sq. cm. of the mica 



