CARBON DIOXIDE AT DIFFERENT TEMPERATURES. 225 



arrangement Qp was immediately followed by another with arrangement Pq, in which 

 the flow of gas and rise in temperature were almost identically the same (see p. 222), 



we have 



CjEj/80! (1 a) = JSQ, + /i for the arrangement Qp, 

 and . 



'i (1-6) = JSQ'j+A for the arrangement Pq. 



Hence, since a is nearly eq\ial to b, and CiEj/S^i is nearly equal to C' l E\/80' ) , we have, 

 to the second order of approximation 



= JS 



A similar equation represented the combination of the two experiments for some 

 other value of the flow, the proper value of (a + b) being substituted, and from this 

 equation and the above equation, h could be eliminated, and JS determined in terms 

 of measured quantities. This was the usual method of applying the correction for 

 the leads. In certain cases, notably in the case of the experiments on carbon dioxide, 

 the runs were only taken with the electric current passing through the leads Qp, but 

 the correction in these cases could easily be calculated and applied.* 



The temperature of the gas was not always exactly the same in the slow and quick 

 flows, and it was necessary to consider whether such small differences as occurred 

 were of appreciable importance. If the temperature varies, the specific heat varies, 

 and also the value of the quantity h. It is easy to see that with such small 

 variations as occurred, the first correction was entirely negligible in the case of air. 

 With regard to the second, it can be easily shown that if the heat loss were 

 entirely due to radiation, h would, on the assumption of STEFAN'S law, vary 

 approximately as the cube of the absolute temperature. The actual measurements 

 of h in the neighbourhood of 20 C. and 100 C. are more accurately represented 

 by h = l'44^ 2 ' 18 x 10~ 7 , showing h to be proportional to a lower power of the absolute 

 temperature than the cube, which is what we should expect since part of the heat 

 loss was due to conduction and convection. The above expression enables the 

 corrections for the variation of h in the two flows to be calculated ; they are all small, 

 only amounting to one or two parts in 10,000 on the specific heat. 



The complete tables for the calorimetric experiments are preserved in the archives, 

 but the tables which follow, and which are considerably abridged, show the main 

 quantities upon which the values of the specific heats depend. 



The second column gives the rate of flow of the gas in grammes per second. The third 

 gives the rise in temperature 80 corrected to the absolute scale by multiplying the 

 rise $0 pt as measured on the platinum scale by 80/80^, this quantity being obtained by 

 differentiating the difference formula 08 pt = d.0(dWO)/iO\ The fourth column 

 gives the values of CE/(l a) or CE/(l 6) according as the. run in question was 

 performed with the arrangement of leads Qp or Pq. These quantities represent the 



* The method of calculating a and b is described in the records preserved in the archives. 

 VOL. OCX, A. 2 G 



