OF SUNLIGHT THROUGH THE EARTH’S ATMOSPHERE. 
259 
present is not obtainable than the figure in the 4th place of decimals. The adopted 
minimum loss of light from scattering is therefore represented by T = J e - 0013jA-t . 
Taking the mean of a large number of days, I find that the average value of 
k is '0017. 
§ XL Langley s Coefficients of Absorption. 
Turning to the work which Langley has published in the volume already alluded to, 
I find no single instance of coincidence with the above law. For instance, at page 151 
he gives a Table of coefficients of transmission, which are as follows in the visible 
spectrum:— 
X . 3580 3830 4160 4400 4680 5500 6150 
Coefficient of transmission for one atmosphere '449 '531 '600 ‘636 '677 '734 '781 
Taking X 6150, which agrees nearly with my scale number 47, and X 4400, which 
agrees nearly with my scale number 57, it will be seen that the coefficient of absorp¬ 
tion increases nearly by X -3 instead of by X -4 . How this wide discrepancy arises, I am 
at a loss to understand. There is one point, however, to be remembered, which is in 
favour of the observations I have recorded, that within certain limits I was quite inde¬ 
pendent of any small haze which might have passed between the slit of the collimator 
and the sun, as the spectrum was compared with one portion of the light coming 
through the slit, whilst another portion formed the spectrum itself. Now any small 
amount of white haze would probably give a general absorption without in any large 
degree altering the relative proportions of the component rays, whilst the bolometer 
readings would be affected considerably; but this would not be sufficient to account 
for the systematic and great differences between our two results. A glance at the 
curve of illumination of the solar spectrum will show that, as far as X 6500 at least, 
there can be no difficulty in making visual measures. With the width of slits used, 
the light coming through scale number 47, a candle had to be placed but 5 inches 
off the screen on a bright summer day at noon to balance the illuminated shadows ; 
and, as the colour of the canclledight and that of the D-light is very nearly of the 
same tint, no difficulty could be found in judging of the value of the latter according 
to the old method, # and this gave identical results with the new method adopted, and 
described at the beginning of this paper. For that reason I have selected Langley’s 
coefficients of transmission of X 6150 to compare with mine, as no physiological 
objections can be brought against such comparison. 
The following in Column II. are the coefficients he tabulates for a stratum of air 
equivalent to a column of 1 decim. of mercury for this wave-length :— 
* “Colour Photometry,” ‘Phil. Trans.,’ 1886. 
2 L 2 
