302 Mr. S. P. Langley on the Amount 



thought, and that the coefficients of absorption and trans- 

 mission, which have been obtained by various writers on acti- 

 nometiy and photometry, are really in general the mean coeffi- 

 cients obtained from the average of hundreds of these alterna- 

 tions (and necessarily too small) ; and yet, more than this, that 

 the smallest part included in the field of the experiment, 

 whether the telluric lines are separately visible or whether 

 they are only latent there, is filled with alternations of trans- 

 mission and absorption, and therefore, according to our pre- 

 vious demonstration, the mean result, even when obtained by 

 a linear thermopile or bolometer, must still indicate too feeble 

 an absorption. I speak now only of the strictly selective 

 absorption ; but I again remind the reader that there is also 

 the partly selective and comparatively non-selective absorption 

 already mentioned, and that practically between the telluric 

 line and the general absorption we have every intermediate 

 coefficient of transmission from unity to zero. The previous 

 criticism applies then, though in a less degree, to the few 

 investigations where two or three coefficients have been used, 

 and even to investigations with the linear thermopile or 

 bolometer, using numerous coefficients, among which I mean 

 to include my own. I have done all I could to minimise the 

 error by measuring on rays as nearly homogeneous as possible 

 — that is, by measuring on parts of the spectrum so narrow 

 that they may be called, without exaggeration, linear, a name 

 which we may certainly apply to a hair-like line i of a 

 millim. in width, forming the working face of ihe linear 

 bolometer ; but even this strip, when laid down in the spec- 

 trum, covers more than the distance between the D lines ; 

 and if we fix our attention on that well-known region as a 

 type, we see that this hair-like line itself covers in this narrow 

 interval alone at least a dozen alternations between brightness 

 and almost total extinction, so that though in respect to 

 wave-lengths we may be said to measure approximately 

 homogeneous rays with a linear instrument, in respect to this 

 local absorption we do not. I am convinced that no one 

 knows what this absorption really is in amount ; but I think 

 we can now begin to see somewhat of what it is in kind, and 

 we may be prepared to agree that the data in the annexed 

 table (Table II.) may represent numerically the proportions 

 of nature with a certain approximation. In this table (II.) 

 we have certain numerical results consequent on the approxi- 

 mative hypothesis that the total heat in sun or star is divided 

 into a certain finite number of parts, each one of which has 

 its own rate of absorption. 



Here the radiant energy before absorption is supposed to 



