abbot] studies of solar constant of radiation 79 



calendar year, the means of the general absorption coefificients have 

 been taken for the observations of 1901-02, and for those of 1903 

 separately. There is an average difference of ten percent in favor 

 of the earlier years, and this cannot, so far as I know, be accounted 

 for in any other way than by recognizing an actual decrease in the 

 transparency of the air, beginning somewhere between November 

 15, 1902, and February 19, 1903. It might be urged that the 

 change is perhaps an annual one, as most of the results of 

 1901-02 are in the autumn and those of 1903 in the spring. 

 But in contradiction to this view we find the observations of 

 March and May, 1902, generally above the mean of that year, 

 so that I incline to think the change rather extraordinary than 

 annual in character. Such a change would imply a correspond- 

 ing reduction in the amount of direct solar radiation at the 

 earth's surface, and if general over a wide area would seem to 

 be likely to occasion some alteration of climate. Recent actino- 

 metric observations reported by several observers in this coun- 

 try and in Europe^ seem to strengthen the probability that the 

 change in transparency of the air is widespread, for their measures 

 of solar radiation at the earth's surface have been appreciably lower 

 of late than for the same months of former years. Several writers 

 have suggested the possibility of the wide dissemination of fine 

 dust clouds from the volcanic eruptions of 1902, in explanation of the 

 lower values. It will be noted from Table I that the differences 

 between the means of 1901-02 and 1903 are largest for short wave- 

 lengths and diminish nearly uniformly toward the infra-red as far 

 as a wave-length of 1.2 //, which would probably be in harmony with 

 this hypothesis ; for such small dust particles might be expected to 

 scatter and absorb the shorter wave-lengths most, not being large 

 enough to act like an opaque screen diminishing all wave-lengths 

 proportionally. 



computations of the solar constant of radiation 



The coefficients of general atmospheric transmission resting upon 

 measures at twenty-four different wave-lengths from 0.37 //. to 2.3 ,a 

 on series of holographic curves have been employed at the Astro- 

 physical Observatory in connection with holographs and actinometric 

 data to compute the solar constant of radiation outside the atmos- 

 phere. Referring to plate xxi, the area included underneath a 

 spectral energy curve is directly proportional to the total radiation 

 absorbed by the bolometer over the range of wave-lengths included 



^ See note by H. H. Kimball, Monthly Weather Rcvieiv, May, 1903. 



