of Cyclically Varying Temperature. 287 



balance was then found by putting a suitable resistance in 

 the box and adjusting the position of the slider on the wire. 

 If the temperature had been the same at the same point in 

 the stroke in every working stroke, an accurate balance could 

 have been obtained by shifting the slider till the deflexion of the 

 galvanometer was reduced to zero,-but this, however, was not 

 the case, and no attempt in the later work was made to obtain 

 an accurate balance on the bridge-wire. The slider was set 

 in such a position that the galvanometer threw about six scale- 

 divisions to one side of zero, and ten successive throws 

 observed. The slider was then moved until the galvanometer 

 threw about six divisions on the other side, and ten throws 

 observed. By taking the mean of these and interpolating, 

 the true reading on the bridge-wire was obtained. 



The gas-pressure was rising towards the end of the experi- 

 ment from 1*6 to 1*9 inches of water. 



Below I give a specimen page of my note-book : — 



Resistance in box 2*2 ohms. 

 Bridge- wire. 



Galvanometer-throws at 500 200 300 



6 left 



3 right 



2 right 



5 



14 







7 



5 



10 



8 



16 



3 



7 



7 



7 



3 



15 



12 



4 



24 



5 



10 



10 



16 



4 



15 



17 



6 



25 



3 



Mean 6-0 left 13 '4 right 7 "5 right 



Taking the throws at 500 and 300 we obtain, by inter- 

 polation from the throws, a bridge-wire reading of 407, and 

 taking those at 500 and 300 we obtain a reading of 411. 



One division on the galvanometer-scale corresponds to 

 from 15 to 50 divisions of the bridge-wire scale, the former 

 figure being for points towards the end of the stroke, and the 

 latter for points towards the commencement of the stroke of 

 the engine. As a check on this method, I have taken the 

 bridge-wire reading from three points or more instead of two, 

 and have found the results obtained agreed within a few 

 scale-divisions of the bridge-wire. To be accurate, this 

 method presupposes constancy in the testing battery, and in 

 the earlier work considerable trouble was found from the 

 variation of the battery, which was two Leclanche cells. In 

 the later work, both in the determination of the fixed points 

 of the thermometers and for the experiments in the gas- 

 engine, a single Daniell cell with an added resistance of 



