Measurement of Gas-Engine Temperatures. 93 



the latter parts of the stroke where the possible error is very 

 great. The wire temperature (curve B) reaches a maximum 

 at crank-angle 210° C. (30° after in-centre). The wire tem- 

 perature must here be ecpal to that of the gas since there is 

 no transfer of heat; it is 263° C. The pressure at this point 

 was found to be 104'5 lbs. per sq. in. absolute, and the volume 

 is 0*333 cubic feet. The mean temperature is therefore 



^77° x °i^ X 310 = 506° Absolute, or 233° 0. 



The temperature calculated on the assumption of adiabatic 

 compression from 14*7 lbs. to 104*5 lbs. is 542° abs. or 260° C. 

 The temperature given by the wire again falls, as it should 

 do, between the two limits. 



The exhaust- valve opens at crank-angle 310° (50° before 

 out-centre). The pressure falls to near atmospheric at the 

 out-centre, and then rises again during the exhaust stroke. 

 At the out-centre the temperature shown by the wire is 40°, 

 but is subject to a considerable possible error since the wire 

 is still hot from the compression and is cooling rapidlv. 

 During the e>diaust stroke the pressure rises, and there is a 

 corresponding rise of temperature which continues till near 

 the end of the stroke. The inlet valve opens at crank-angle 

 515° (25° before the in-centre) and provides an additional 

 exit for the air. At the same time, the piston is slowing 

 down and there results a rapid drop in the pressure, and a 

 corresponding fall in temperature due to the expansion. 



It appears therefore, that when corrected for lag in the 

 manner described, the temperature of the wire gives that of 

 the gas in its neighbourhood at least as accuratelv as the 

 latter can be obtained from the indicator diagram. The 

 corrections are, however, very large at certain points, and 

 the temperature correspondingly uncertain : the possible 

 error amounting sometimes to as much as 20° C. This large 

 error is mainly due to uncertainty in the shape of the wire- 

 temperature curve, which gives rise to errors in the final gas- 

 temperature greatly exceeding the errors of measurement in 

 the wire-temperature. It is d oubtf ul whether the possible error 

 can be reduced except by using a much smaller wire, with a cor- 

 respondingly smaller correction for time-lag*. There was no 



* Even with the somewhat rough method? of measurement that I 

 used, accidental fluctuations in the wire-temperature were quite obvious — 

 due, no doubt, to the eddying motion of the air in the cylinder. Such 

 fluctuations will, of course, appear whatever the size of wire, but will 

 introduce greater uncertainty, the larger the wire. More accurate 

 resistance measurement would not eliminate them. 



