216 MR. W. F. G. SWANN ON THE SPECIFIC HEATS OF AIR AND 



is the average number of times which a particle of gas strikes the wall of the tube in 

 travelling through a distance of 1 cm. parallel to the axis, the effect on the heat loss 

 is to introduce a term which to the first order may be written )t 2 /k(n%l) JSQ or 

 A 2 /10&JSQ, where n and I have the significance and values accorded to them on p. 212. 

 The correction thus tends in the opposite direction to that discussed on pp. 211-214. 

 Now, in view of the fact that there was only a small space between the wall of the 

 tube and the bundle of glass rods which contained the leads of the heating coil and 

 thermometers, and further, since the heating coil wire and the pieces of wire gauze, 

 &c., offered a large amount of obstruction to the flow of the gas, it is highly probable 

 that k was at least equal to 1. If we take k equal to unity, the correction for the 

 heat-loss term amounts to h 2 / LO JSQ, and produces an effect amounting to about 

 5 parts in 10,000 on the value of the specific heat at 20 C. and 1 part per 1000 on 

 the value of the specific heat at 100 C. The effect tends in the opposite direction to 

 that discussed on pp. 211-214, but the conduction of heat to the walls of the tube 

 by the discs of copper gauze tends to make the value of the correction much smaller 

 than that above deduced, so that it is safe to assume that entire neglect of it does 

 not produce an error of more than 3 or 4 parts in 10,000 on the values of the specific 

 heats. 



In the foregoing discussions no account has been taken of the difference in 

 temperature between the external and internal surfaces of the wall of the calorimeter 

 proper. Assuming 0'002 for the thermal conductivity of glass, and taking O'Ol calorie 

 for the average heat loss per second per degree rise in temperature during the 

 experiments, it is easy to calculate from the dimensions of the calorimeter proper 

 that the average difference in temperature between the surfaces of the calorimeter 

 wall for a rise in temperature of 5 C. was about 0'05 C. Since the heat loss 

 remained constant in the two flows to the first order of its value, the difference 

 in temperature between the surfaces of the wall of the calorimeter proper would 

 remain constant to the first order ; in fact, it is easy to see, making use of the 

 investigation given on pp. 211-214, that it would remain constant even in the 

 case of the experiments at steam temperature to about 0'00l C., so that the effect 

 due to it would eliminate on the two flows certainly to 1 part in 10,000. In the case 

 of the experiments at air temperature the effect would be even less than for steam 

 temperature. 



(14) The Final Experiments. The final experiments were divisible into two 

 main sections : the calibration of the fine metal tubes, and the main calorimetric 

 experiments. 



(15) Calibration of the Fine Metal Tubes. The method first tried was a 

 modification of that already referred to in connection with the preliminary experiments ; 

 several circumstances, which it is unnecessary to enumerate, however, rendered this 

 method somewhat untrustworthy. It was desirable that a method should be employed 

 in which the whole 16 tubes could be calibrated at once; I consequently devised the 



