482 



SCIENCE. 



[Vol. I., No. IT. 



how much of the radiation from a gas-burner 

 was visible, and how much was not, but to 

 give a map of its distribution on the normal 

 or wave-length scale, which would enable 

 any one to see the qualitj' and amount of 

 the energy in each part of the light and heat 

 region. 



The ordinar}^ argand burner, burning com- 

 mon house-gas within a glass chimnej', was 

 first placed at the centre of curvature of a 

 large Rowland concave grating ; and, by means 

 of the bolometer, the heat was measured at 

 successive points in the spectrum down to a 

 wave-length of about .001 mm., where the 

 overlapping second spectrum began to be sen- 

 sible. Even in the prehminar}^ determination, 

 it was interesting to observe that the distribu- 

 tion of the heat was totally different from that in 

 the sun, and that, instead of growing smaller, 

 it grew greater, as the bolometer passed from 

 the visible to the invisible end. As it was 

 evident that the heat was still increasing at 

 the point where the evidence from the grating 

 failed, all the measures were next repeated 

 with a prism of a special glass known to be 

 transparent to the invisible rays. (It was 

 first attempted to use the linear thermopile ; 

 but the heat was insufficient, and the linear 

 bolometer was substituted.) With this, as 

 many as thirteen ordinates were measured 

 (representing the proportionate heat at as 

 many points), their respective indices of re- 

 fraction being determined by the known re- 

 fracting angle of the prism and the observed 

 deviations on the circle of the spectro-bolom- 

 eter. 



In a late communication to the academ}^ I 

 gave the results of a recent research upon the 

 connection between indices of refraction and 

 wave-lengths, which enable us to deduce the 

 normal spectrum (invisible as well as visible) 

 from the prismatic one. It appeared to me 

 when I was engaged in the first investigation, 

 which to all but students of the subject must 

 seem abstruse, that its results were of a kind 

 which could never have much other than a 

 theoretical interest : but it happens that this 

 their first application is of a utilitarian char- 

 acter ; for, having thus converted the prismatic 

 values into corresponding ones on the wave- 

 length scale, I was able to represent the con- 

 clusions from both b}' the normal maps which 

 I now have here, and which exhibit the results 

 of the analysis of the radiant heat which has 

 come through the chimiiej'. Let us remember 

 that this radiar.t energ}' differs whollj' in its 

 qualities in difi'erent parts, and that the quality 

 is shown by the wave-length numlier on the 



horizontal scale ; the amount, by the height of 

 the curve at that point. Near the part with 

 wave-length .5 it gives the eye the sensation 

 we call blueness, and near .7 it appears as 

 a dull red, bringing verj- little light ; at the 

 point .9 or .10 it makes on the most sensitive 

 ej-e no impression whatever, but has the 

 power of passing freely through the glass 

 chimney ; near .3 the glass, so transparent 

 to the light, is almost wholl3f opaque to the 

 energy : so that each part has some quality 

 peculiar to itself. By far the most important 

 of these qualities, for our present purpose, is 

 that of giving light. If we then analyze the 

 radiant euergj' which comes through the chim- 

 nej', the result is shown in our lower curve. 

 The energ}', which is what the gas supj^lies 

 at the cost of the production from the coal, is 

 for our i^resent purpose regarded as saved 

 or wasted, according as it is visible (hght) or 

 invisible (dark heat) . The energj' first be- 

 comes measurable in the blue, where there is 

 very little of it, but where all there is, is effec- 

 tive as light ; it increases steadilj' to the 

 extreme red ordinaril3- visible where there is 

 a great deal of it, but of a quality which is 

 onl3- interpreted bj' the eye as a dull reddish 

 glow of little value for lighting ; and then goes 

 on increasing where it passes into complete 

 invisibilitj', and still continues to increase as 

 (for the present purpose) pure waste, till its 

 maximum is reached at a wave-length of 1.5 or 

 1.6, — something like three times the length of 

 the visible spectrum below the lowest visible 

 raj-. The energy at any point being propor- 

 tional to the height, the entire radiant heat is 

 proportional to the area of the curve. If we 

 draw it on such a scale that this whole area 

 equals 100, we can see the percentage expend- 

 ed in anj' kind of radiation at a glance. 

 The small, nearly triangular area to the left of 

 the line' at .7, for instance, represents all the 

 radiant energy useful as light ; and this area 

 being by measurement 2.4, while that of the 

 whole curve is 100, we see that 2.4% are 

 employed as light, and the remainder, 97.6%, 

 are wasted. But this refers to the radiant 

 heat alone, and takes no account of that 

 expended in heating the air hy convection cur- 

 rents. I have heard this latter estimated .at 

 three-quarters of the whole, but have not 

 mj'self measured it. Admitting that this is 

 approximately' correct, however, it follows, 

 that, since only one-quarter is radiated, it is 

 2.4% of this one-quarter onty, which is light, 

 and that finally less than 1 % of the whole is 

 used, and more than 99% wasted. 



It is instructive to take an amount of solar 



