Abstract
The object of this work has been the improvement of the existing knowledge of the spectral distribution of power of daylight, i. e. the relative proportions of the various wavelengths present. The work differs from previous studies in that (i) measurements have been made at much smaller wavelength intervals, (ii) the sampling conditions have been chosen to more nearly represent those applying in actual buildings, (iii) and that measurements have been made in both urban and rural locations. With one exception, all previous studies have been made in other countries, mostly in the U. S. A. The method used has been that in which the spectral distribution of daylight is determined by comparison with that of a known source - a tungsten lamp running under accurately controlled conditions. The spectral distribution has now been observed on 434 occasions. The results have shown no startling difference between urban and rural locations, but there is clear evidence that the spectral distribution in London is affected by large amounts of diffusing material in the atmosphere. Attempts have been made to correlate the result obtained on any occasion with the prevailing weather conditions and time of day. These have indicated that the position of clouds relative to the observing station have a far larger effect than any other factor, save when the sky is fully overcast. The results have been compared with the less detailed ones of previous studies; there is general agreement over the visible range but there are wide differences in the near ultra-violet region. The results of this study are in remarkable agreement with a recent study made in South Africa, and indicate discrepancies in the internationally agreed values for typical spectral distributions. By reference to values of daylight illumination, a few results have been translated into terms of absolute, rather than relative spectral power distribution.