ON GASEOUS EXPLOSIONS. 215 



persists, there is necessarily some loss of heat by radiation to the walls. 

 In order to estimate this loss, I made a series of direct measurements 

 of the actual proportion of the heat of combustion radiated from various 

 flames, luminous and non-luminous, some of which were quoted by 

 Hopkinson in his paper. I found that the heat radiated from an 

 ordinary non-luminous Bunsen flame might amount to 15 per cent, or 

 20 per cent., but that it depended on the duration of the incandescence 

 and was much smaller, corresponding with a reduction in the size of 

 tbe flame, in explosive mixtures. It is not possible to estimate 

 separately the exact amount of this loss in the cylinder of a gas-engine, 

 but I think it belongs chiefly to losses of the type A, being proportional 

 to the wall surface exposed, and practically independent of the time, 

 since the duration of the flame is short in the most efficient mixtures. 

 It is probable, however, that part of the radiation loss taking place 

 during the propagation of the flame and throughout its mass is propor- 

 tional to the volume and not to the surface, in which case it would 

 be represented by a constant term in the expression for the loss of 

 thermal efficiency.' 



The only account which I have been able to find of previous syste- 

 matic experiments on the proportion of the heat of combustion radiated 

 from a flame is in a thesis for doctorate by Robert (the son of Hermann) 

 von Helmholtz. For the majority of non-luminous hydrocarbon flames 

 mixed with air E. Helmholtz finds approximately the same result, 

 namely 5 per cent, of the heat of combustion radiated. According to 

 my experiments this low value is to be explained by the fact that he 

 employed in these measurements small flames, 6 mm. diameter ty 

 60 mm. high, which were probably burning at a comparatively low 

 temperature, and which do, as a matter of fact, give a percentage of 

 this order. In one case he finds 8 - 7 per cent, of the heat of combustion 

 radiated by a flame 11*8 mm. diameter. 



In my own experiments the heat radiated from flames of various 

 sizes, and burning under different conditions, was measured, in calories 

 per square cm. per minute, at a measured distance, by means of an 

 Angstrom pyrheliometer in a special mounting. Tbe constant of the 

 pyrheliometer, which had shown signs of change, was checked by 

 means of a radio-calorimeter and also by an absolute measuring bolo- 

 meter. An ordinary wet-meter was employed for measuring the gas 

 supply to the flames, and the same meter was employed in the measure- 

 ment of the calorific value of the gas with a Boys calorimeter. In 

 some experiments the air supplied to the flame before ignition was 

 measured with the apparatus subsequently employed by Swann 1 

 in his experiments on the specific heat of air and CO,. This was 

 useful for estimating the strength of mixture in relation to the 

 appearance of the flame, and for varying the temperature, but could 

 not give quite exact results because the flames were necessarily burn- 

 ing in free air. With the air and gas adjusted as nearly as could 

 be estimated in the proportions required for complete combustion, the 

 proportion of heat radiated varied from 10 to 15 per cent, for burners 



; Phil. Trans. A. 210, p. 208. 



