90 REPOET OF THE SECRETARY. 



It will be noted that though the agreement of these values would not be held 

 very close in some lines of physical investigation, yet when it is taken into account 

 that an invisibly thin cloud or an unaccountable disturbance of the galvanometer is 

 sometimes suflicient to make an energy curve erroneous, the results pronuse well. 

 It is yet too early to do more than call attention to the depression in the curve in 

 the yellow and green at about 0.55//. This may be due to the absorption of water 

 vapor, but experiments here given are preliminary. Results of interest appear 

 probable in relation to the applicability of Bouguer's formula to wave lengths where 

 selective absorption is active, and relating to the variation of atmospheric absorption 

 through the day and through the year. It will be observed from the preceding table 

 that the general absorption is least at wave lengths 1.05 /< and 1.30/< in the infra red, 

 where the transmission coefficient rises to 95 per cent. 



As an illustration of the results obtained from measures of areas, and representing 

 total radiation for considera))le ranges of wave lengths, I call attention to the diagrams 

 of Plate IX. In these diagrams the horizontal distances represent the time of the year, 

 the vertical distances the corrected areas or total radiation received, corresponding to 

 certain selected portions of the curve. The observations on which these selected areas 

 are based were all made at or near noon. In order that the several spectrum energy 

 curves might be comparable as regards instrumental conditions, all the directly meas- 

 ured areas were nmltiplied by a factor chosen so that, as corrected, the curves w'onld 

 all be of a uniform height of 10 centimeters at 1.05 /<, where the transmission, as I 

 have just pointed out, is greatest. The upper line in Plate IX represents the total area 

 included under the energy curve between wave lengths 0.76 /< and 1.98 /<. The lower 

 full line includes thesummation of several portionsof this area known to be least subject 

 to diminution by the absorption of water vapor, namely, the regions 0.97 i-i to 1.10 /<, 

 1.16/< to 1.33 /1, 1.49 /< to 1.82/1. Crossing itat manypointsis a dotted line represent- 

 ing the summation of the remaining areas, including the great water vapor bands of the 

 infra red to a wave length of 1.98 /<. These curves (prolonged through July and August 

 for additional interest) show how important are the effects of water vapor on the direct 

 radiations we receive from the sun. In order to compare the several parts of the 

 curves justly, times equidistant from the June solstice should be .selected so that 

 equal air masses are traversed. The differences in air mass are not sufficient to pro- 

 duce much effect upon the general absorption in this region, l)ut affect the water- 

 vapor absorption appreciably. It will be seen that the total solar radiation in the 

 region between 0.76 // and 1.98 // was on an average fully 15 per cent greater in March 

 of this present year than in August, on account of the greater absorption of water 

 vapor. Comparing the separate points, it may be seen that the variation in water- 

 vapor absorption, while, as Ave have just seen, on the whole seasonal, is j'et very fluc- 

 tuating from day to day. Thus, for example, it could be inferred from a discussion of 

 the curve in March that in two api)arently equally clear days of the same week there 

 was a difference of 15 per cent in the solar radiation received, owing to the difference 

 in water-vapor al)sor))tion. In sharp contrast to the variability of the areas of the 

 water-va])iir absdrption regions is the In'hux ior df llic remaining i)ortion-of the spec- 

 trum. This contains some secondary hut still considerable water-vapor bands, 

 so that some fluctuation still remains. I>ut it seems possible as a result of this study 

 that certain large regions, notably that between wave lengths 0.97 // and 1.10 /<, are 

 so nearly unaffected in area by any terrestrial atmospheric absorptimi that they can 

 serve to indicate if the total solar radiation fluctuates from year to year. 



Absorption of tlie ml ar envelope. — The experiments on the absorption of the solar 

 envelope briefly mentioned in last year's report were continued as far as i>racticable 

 with the apparatus available. Fig. 4 shows the results thus far reached. Hori- 

 zontal distances are proportional to wave lengths, vertical distances to transmission 

 coefficients. These transmission coefficients are not exactly similar to the ones just 

 given f(jr the earth's atniosphere. If we represent the intensity of the radiation we 



