164 S. P. Langley on the Selective Absorption of Solar Energy. 



The middle curve (I) is that at high sun. Except for the 

 heat below wave-length (1^0), the area of the curve may be 

 considered to represent the heat actually observed by the acti- 

 nometers, at noon, as presently given. 



The lower curve (II) is that at low sun. Its area is pro- 

 portional to the heat received when the sun shone through 

 double the absorbing air-mass that it did at noon. 



The upper dotted curve is "the curve outside the atmo- 

 sphere." Its area will give the heat which would be observed 

 if our apparatus were taken wholly above the absorbing air, 

 and the distribution of this heat (energy) before absorption. 

 Knowing the values in calories corresponding to the middle 

 curve, we readily obtain the absolute heat before absorption 

 (the solar constant). 



It should be noticed that if we had attempted to deduce this 

 latter value by applying our logarithmic formulae directly to 

 ordinary actinometric observations (i. e. to observations where 

 only the indiscriminate effect of all heat- rays is noted by the 

 thermometer) made at high and low sun, we should have 

 obtained a quite different result. This has been the usual 

 process, but it can never be a correct one; for these expo- 

 nential formulae are in theory only applicable to homogeneous 

 rays, and the departure from theory here involves an error 

 which is demonstrably large. 



The above values in Table VII. are relative only. To 

 obtain absolute ones we have now to combine this result with 

 the actual measurements of solar radiation in calories, or other 

 units furnished by actinometers under approximately the same 

 conditions. We shall at the same time thus obtain a prelimi- 

 nary value for the solar constant. Taking the mean of our 

 observations with the Violle and Crova actinometers on clearest 

 days, we have 1*81 calory observed at Allegheny in March 

 1881. This is the absolute amount of heat represented by the 

 area of a completed " high-sun " curve. 



To this result the energy distributed through the whole 

 spectrum has contributed, while our bolometer-measurements 

 in the diffraction-spectrum end at wave-length 1 M, 00. Never- 

 theless, since we do in fact know from subsequent measures 

 (to be given later) where the effective spectrum ends, we can 

 by the aid of these measures prolong the curves and obtain 

 their relative areas with close approximation. In this way 

 we determine, by measuring the charted areas, and making 

 allowance for the (here) uncharted area below X = 1 M, 0: — 



