NO. 5 SOLAR VARIABILITY — ABBOT, FOWLE, ALDRICH 1 7 



It is to be regretted that at this very time, when all these changes 

 were occurring, we made a radical change in the observing appara- 

 tus, for on September 23 we substituted for the bolometer which 

 had been in use on Mt. Wilson since 1905 the bolometer which 

 was employed in the Mt. Whitney and Algerian expeditions. We 

 made this change because the Algerian bolometer was less subject to 

 prejudicial influence from wind. But had we known how many other 

 changes were occurring at the same time, it is certain that we would 

 not have made a change of apparatus too. We have thought it 

 necessary, on account of this, to investigate very thoroughly the 

 merit of the solar-constant determinations succeeding September 23, 

 and we find the following facts to verify their accuracy. 



First, empirical determinations of the solar constant from pyr- 

 heliometry and psychrometry at Arequipa in Peru ^ indicate that the 

 values subsequent to September 23 were lower than those prior to 

 that date, and by about the same amount as do the Mt. Wilson 

 observations. Second, we have determined an empirical formula for 

 the solar constant from Mt. Wilson pyrheliometry and psychrometry. 

 This formula has been worked out solely by the use of observations 

 of 1910, 191 1, and 1912, and does not depend in any way on bolo- 

 metric observations of 191 3, except that we chose days of the earlier 

 years for which the precipitable water had values comparable with 

 those of 1 91 3. This formula gives the same change in solar-constant 

 values at about September 23 that is indicated by the bolometric work 

 itself. Third, we felt a suspicion that the determinations of the 

 water-vapor absorption in the holographic work might be interfered 

 with by some change in the excellence of the definition of the spectro- 

 scope. Instead of employing with the usual constants Mr. Fowle's 

 method of determining these absorption effects in the holographs, 

 from measurements on the band par, the areas of the absorption 

 bands were actually measured on many plates, just as we did formerly 

 in all our holographic reductions, and we thus determined new con- 

 stants applicable to the last part of 1913. But no substantial change 

 occurred in our conclusion regarding the fall of the solar-constant 

 values on and after September 24. 



We therefore see no reason to doubt that the days from September 

 9 to the end of 191 3 are as homogeneous as any other series of our 

 measurements, and we believe that the great drop of the solar- 

 constant which is indicated to have occurred just after September 23 

 is a real one. 



^ See " Arequipa Pyrheliometry," Smith. Misc. Coll., Vol. 65, No. g. 



