larly over the latter, for in it the water absorptions play such an important rôle that the law of scattering could hardly be taken into account also. It will be noticed that observations have been made at all times of the year, and that the winter observations, when the surface of the ground at Grindelwald has been covered with snow and the air has been intensely cold (the thermometer sometimes showing as much as 40° F. of frost), give the coefficients of least absorption. The writer has held that most of the scattering particles are water in an extremely fine state of division, and that there is a probability that such is the case is shown by the fact that the most transparent atmosphere is that in which there is most warm aqueous vapour present. A change in the summer to what artists call an "atmospheric landscape" invariably shows a higher coefficient of absorption. Whatever these particles may be, there is no doubt that they are in the same volume fewer in number at high altitudes than they are at lower ones; as the coefficient of absorption is so much less in the former case than in the latter. Any one who has observed the sky at these higher altitudes will have observed that when the sun is brightest and highest, the sky is blackest, that is, that there are fewer scattering particles.
XXXIII.—Standard of Light.
It is somewhat difficult to know as to what standard of light to refer the light of the sun. The Author has made many comparisons between it and an amyl acetate lamp, and has come to the conclusion that at midday at sea level, and in the clearest atmosphere, the brightness of the* sun exactly overhead would be close upon 7000 such standard lights at 1 foot distant from the screen, which would be equivalent to 5600 standard candles at the same distance. For the purpose of this paper, it is, however, unnecessary to refer the light to any particular standard, since all that was sought for was to obtain a comparison of the losses suffered on any one day by the light after passing through various thicknesses of atmosphere. It will be noticed in the various tables that the calculated brightness at the zenith varies very considerably even at the same altitudes. This must be laid down to one of three causes: (1) either an error in calculating the coefficients; (2) a slight haze of coarse particles intervening; or (3) owing to a varying sensitiveness in the paper used. Where the actual observed intensities throughout a day do not vary much more than those calculated, it may be presumed that the coefficient is not very far from the truth, and, consequently, the variation will in all probability be due to the second cause. As stated before, there was a change made in the paper which occurred in 1889, and it will be found that the intensities are practically the same before and after the above year.
XXXIV.—Description, Tabulation, and Discussion of Results.
The results which are tabulated do not show by any means all the observations made. They have been carried on for four years, and only those days are recorded in this paper when the sky has been practically cloudless whilst the observations have
