NO. 5 



-SOLAR VARI.\BILITY ABBOT, FOWLE, ALDRICII 



13 



years, and they differ somewhat from each other.' The weights for 

 the two years were not smoothed by wave lengths. They are as 

 follows : 



Table 8. — Adopted luetghts 



For the purpose in view, the differences between the two sets of 

 weights are of little significance. The numbers are, on the whole, 

 very like those which might be derived from figure 5. 



We have made two dift'erent reductions of the data. In one, we 

 multiplied the tangents for each day by the appropriate multiplier 

 taken from the preceding table, and having taken the mean of the 

 weighted tangents, we thus obtained for the given day a value repre- 

 senting the average weighted departure from the standard condition 

 of distribution, as determined by observations at seven different wave 

 lengths. In another reduction, we omitted the wave lengths 0.3737 

 and 1.008 yu, and took the simple mean of the tangents for the other 

 wave lengths. We omitted these two wave lengths in consideration 

 of the fact that the accidental error at 0.3737/x, is unduly large, owing 

 to the smallness of galvanometer deflections there ; and also that the 

 changes of contrast at wave length at i.ooSju, are relatively small 

 compared with the accidental error there. 



The following tables (9 and 10) give the solar-constant and solar- 

 contrast numbers for the years 1913 and 1914. The values given 

 must be regarded as preliminary and subject to minor changes before 

 final publication. 



As so much depends upon the values for 1913, a great deal of 

 attention has been paid to the remarkable decrease of the solar- 

 constant values which occurred on and after September 24, 191 3. 

 Not only did the solar-constant values fall oft' at this time, but also 

 a change in the contrast values occurred. If we divide all the days 

 in which solar-contrast values were obtained in 191 3 into two groups. 



^It may be significant in view of what follows to note that the weights 

 obtained for 1913 depended on the relative amounts of the deviations of indi- 

 vidual wave lengths on different days, while the weights obtained for 1914 

 depended rather on the relative deviations for these wave lengths found from 

 the year as a whole, compared to 1913 as a whole. 



