ON GASEOUS EXPLOSIONS, 837 



.accuracy of the absolute values of the specific heats obtained, but consider 

 tliat the ratios or relative values, and the rates of increase with tempera- 

 ture, are more likely to be correct tJian the absolute values, because the 

 various sources of error which they discuss are more likely to be elimi- 

 nated in the relative values. 



The value found for the mean specific heat of air over the range 115" 

 to 270° C. by Holborn and Henning was -2315, which is about 5 per 

 cent, smaller than the probable value over this range. For the rate of 

 increase of the mean specific heat they gave the formula : 



B„,,=S<,(l + -00004i!) (Holborn and Austin). 



but considered that the rate of increase shown by their experiments was 

 within the limits of probable accuracy of their work, and that it could not 

 be regarded as certainly established that there was any increase over the 

 range of their experiments. 



.Later experiments by Holborn and Henning^ with a platinum 

 heating tube, extending to 1400° C, were made by a similar method, 

 except that the gain of heat by the calorimeter from the heating tube 

 was partly compensated by surrounding the calorimeter at 115° C. with a 

 jacket maintained at a much lower temperature. This compensation was 

 found necessary at high temperatures in order to prevent an excessively 

 rapid rise of temperature of the caloiimeter ; but although it reduces the 

 apparent magnitude of the correction for gain of heat by the calorimeter, 

 it does not diminish the actual amount of heat transferred and does not 

 reduce the uncertainty of the correction. The magnitude of the eflfect at 

 high temperatures may be judged from the fact that it was found neces- 

 sary in the experiments at 1400° C. to maintain the jacket at as low a 

 temperature as 40" C. by passing a stream of cooling water through it in 

 order to prevent the calorimeter rising above 115° C. when no gas was 

 passing. Under such conditions the calorimetric corrections become so 

 uncertain that the probability of systematic errors must increase con- 

 siderably with rise of temperature. If the method gives a probable error 

 of 5 per cent, in defect over the range 115° to 270° C. it does not seem 

 at all impossible that the error may amount to 10 per cent, over the 

 range 116° to 1400° C. 



The rate of increase of the mean specific heat of nitrogen at atmo- 

 spheric pressure between 840° and 1340° C, shown by the later experi- 

 ments, was about double that found in the earlier series. Both series of 

 experiments could be represented within the limits of probable error by 

 the linear formula 



S,,(= -2350(1 -F -000080 (Holborn and Henuing). 



It appears probable, however, that the value of the specific heat at 0' C. 

 given by the formula is too low, and that the rate of increase is not 

 uniform, but increases with rise of temperature to some extent in the case 

 of nitrogen. 



8'p^ific Beat of CO^. 



4. Similar remarks apply to the determination of the specific heat of 

 CO2 by the same methods, but tlie case of CO2 is of special interest on 

 account of the rapid variation observed at ordinary temperatures. The 



' Vi'xed. Ann., 23, 1907, p. 809, 

 1908, • z 



