212 REPORTS ON THE STATE OP SCIENCE. 



but not in proportion to the volume of the flame. The radiation from 

 very large flames would tend to become proportional to the surface, but 

 no certain inference as to the diameter of flame for which this would be 

 substantially true can be drawn from Callendar's experiments, because 

 he was looking along a thin row of flames in which there was but little 

 lateral extension. 



The flames met with in a gas-engine cylinder or in explosion vessels 

 differ from open flames such as can readily be produced in the labora- 

 tory, both in respect of the lateral extension which has just been 

 mentioned, and also in respect of density. In both these particulars the 

 difference is rather great, the least dimension of the mass of flame in 

 a gas-engine cylinder being only in the smallest sizes comparable with 

 the diameter of the Meker burner flame, while the density of the gas 

 just after firing in the gas-engine is from twenty to thirty times that of 

 the burner flame gases. It does not seem possible from theoretical 

 considerations to determine the effect of these two factors with suffi- 

 cient accuracy to enable any quantitative inference as to radiation in 

 the gas-engine to be drawn from laboratory experiments on flames, but 

 it is useful to discuss their probable qualitative effects. 



In fig. 1, p is the point of observation at which the pyrometer is 

 placed, as in Callendar's experiments, and the portion of the flame 

 from which the radiation is measured is that intercepted by the small 

 cone. If a second similar flame b is placed behind a at a considerable 

 distance but so that it is intersected by the cone, then the radiation 

 recorded by the pyrometer will be increased, say, by 50 per cent., 

 showing that of the radiation emitted by b and falling on a 50 per cent, 

 is absorbed and the remainder is transmitted to the pyrometer. The 

 absorbed energy is of course not lost, but must result in slightly in- 

 creased radiation from a in all directions. The flame a appears to be 

 a little hotter because of the proximity of B. Thus the increase of radia- 

 tion absorbed at the pyrometer is due not only to the radiation trans- 

 mitted from b but also to an increase in the intrinsic radiance of a. If 

 the two flames ai'e a considerable distance apart, the latter part is negli- 

 gibly small, since the flame a does not then receive much radiation 

 from b, and what it does receive is dissipated in every direction. But 

 when flame b is pushed close up to a in to the position of b' (fig. 2) this 

 effect may be* considerable, and it is obvious that it will be greatly 

 enhanced if the two flames are extended laterally as in fig. 3. For in 

 such case flame a must get rid of the energy which it is receiving by 

 radiation from b' mainly by an enhanced radiation in the direction of p. 

 It may, therefore, be expected that the effect of lateral extension will 

 be to make the flame apparently more transparent. 



To a first approximation it may be expected that the radiating and 

 absorptive powers of a gas at a given temperature will be proportional 

 to its density. That is to say, two geometrically similar masses of 

 flame, in which the temperatures at corresponding points are the same, 

 and the densities in inverse proportion to the volumes (so that the total 

 masses are the same), will radiate in the same way and to the same 

 total amount. It would seem that this must be so, so long as the 

 vibrations of the radiating molecules are the same in character and 



