406 S. P. Langley — Invisible Solar and Lunar Spectrum. 



there is always a feeble heat to be observed throughout this 

 extent. This, however, the use of the sifting train shows to 

 be largely, at any rate, factitious, but we admit the possibility 

 that subsequent research may prove that it is not all so. 



Let us now recur to iig. 2b, where we shall find below 10^ 

 the same dependence of the effects upon the season and the 

 hour, as in the part above 5^. At 10^2 observations made 

 during the autumn showed scarcely the feeblest suspicions of 

 heat, and the same has held good iu the very mild weather of 

 the past winter (of 1887) ; but on a few days, when the temper- 

 ature had fallen below the freezing point, a notable maximum 



was found at this point, followed by a minimum at 10^*7. The 

 height of this maximum relatively to the principal one in this 



region at about 13^ appears to be correlated with the composi- 

 tion of the air as affected by the temperature. On the coldest 

 day (temperature at noon — 6*7° C.) the deflection at midday for 



/ = 10^*2 was nearly one-half that at 13*", but on other days, 

 when the temperature was near 0° C, the deflection at 



10^-2 did not exceed one-fourth that at 13", while at tempera- 

 tures above +10° C. it was not noticeable. 



It is in the region near 13^ to 14^, or over twenty times the 

 length of the visible spectrum below it, that we have found 

 the maximum of the lunar heat spectrum, and it is here that 

 we first obtained indications of solar radiation corresponding in 

 its great wave-length to this special lunar radiation, but of 

 amounts non-estimable by the means till now employed. I 

 have already spoken of its almost unrecognizably small amount, 

 and a perhaps more vivid apprehension of its extreme minute- 

 ness will be gained from the statement that on this graphical 

 construction, on the scale of ordinates used in delineating the 



curve from to 5^, no heat appears below 5*" anywhere, not 



even at the maximum near 13^ and 14/*; because, though heat 

 exists to the bolometer, the highest ordinate which would 

 represent it on our drawing is not so great as the thickness of 

 the thin black line, which denotes the axis of abscissae. T have 

 accordingly here been obliged to exhibit it separately by a 

 dotted curve whose ordinates are one hundred times those in 

 the remainder of the spectrum. The points of minima in it 

 are identifiable with absorption bands, which we have directly 

 observed, and which we have independently found to exist in 

 our own atmosphere, by studying the radiations from a copper 

 surface, one meter square, at the temperature of boiling water 

 placed in the open air at a distance of one hundred meters 

 from the bolometer. (The description of this will be found in 



