﻿404 SMITHSONIAN MISCELLANEOUS COLLECTIONS [VOL. 47 



The first column contains the date of the observation ; the second 

 the tangent of the slope of the curve, and is equal to the logarithm of 

 the value of "a" in Bouguer's formula; the third, this "a"; the 

 fourth, the range of air-masses of the actinometer values ; the fifth, 

 extrapolated value of the solar radiation via the actinometer; the 

 sixth, an estimation of the quality of the actinometric series ; the 

 seventh the solar radiation via the holographic process ; the eighth 

 an estimation of its quality, and finally in the ninth the percentage 

 correction necessary to bring the actinometer extrapolation in accord- 

 ance with the bolometric value. 



The table shows, as Mr. Langley demonstrated, that the values 

 from the actinometer are too low. It is surprising, however, that 

 the correction to be applied seems so nearly the same, averaging 

 about 14 per cent., nearly independently of the transparency of the 

 air or from what air-masses the actinometer extrapolation is made. 



There are but three notable exceptions to this in the table : of 

 these the one of October 14, 1904, is undoubtedly caused by a defect 

 in the latest actinometer value. The other two are extrapolated 

 from very small air-masses where, as will be shown, smaller correc- 

 tions are to be expected, but besides this the holographic observa- 

 tions were poor on these latter dates, and in the case of July 7, 1903, 

 too small a range of air-mass was available to suitably determine 

 the extrapolation. 



Referring again to the remarkable and unexpected uniformity of 

 the differences between the actinometric and bolometric extrapola- 

 tions, it seems desirable to see how far this uniformity accords with 

 what theory would lead us to expect. The analytical discussion, how- 

 ever, becomes very complex except in some comparatively simple 

 hypothetical cases, so that the following graphical method has been 

 employed. By the same process used in the bolometric extrapolations, 

 actinometric values may be computed for any desired air-masses. 

 In the accompanying figure 63 are shown two curves thus derived. 

 Curve (A) was computed for March 25, 1903, a day of compara- 

 tively great general absorption, and curve (B) for October 14, 1903, 

 a day of small general absorption, but both days of moderate water- 

 vapor absorption. The abscissae are of course air-masses, the or- 

 dinates are the logarithms of the computed actinometer readings. 

 It would have been better to have used more extreme cases, e. g., a 

 day of great general and great water-vapor absorptions, together 

 with one when the absorptions of both these kinds were small. Such 

 days, however, were not available. The values of " a " for these 

 days as varying with the wavelength are given in Table in. 



