80 SMITHSONIAN MISCELLANEOUS COLLECTIONS [vOL. 45 



in the curve. But this area is not strictly proportional to the total 

 solar radiation at the earth's surface, as determined by actinometer 

 observations, for the reason that the radiation has been unequally 

 reduced at different wave-lengths by losses at the siderostat mirror, 

 within the spectroscope, and by selective absorption at the bolometer 

 itself. It is necessary to correct the curve so that it shall as accu- 

 rately as possible represent the distribution of energy in the solar 

 beam prior to these losses. Inasmuch as the coefiftcient of total 

 absorption of the lampblacked bolometer strip is upward of 95 

 percent, it is believed that no considerable error is admitted by 

 neglecting its differences of absorption for different wave-lengths, 

 and no correction is applied for this. The relative absorption of the 

 spectroscope for different wave-lengths is frequently determined, 

 and that of the siderostat mirror still more frequently, for in both 

 these optical parts of the apparatus there is rapid deterioration of 

 the reflecting power of the silvered glass surfaces. At present this 

 indeed forms one of the main difficulties and sources of error of 

 the investigation, for a whole day of observing and several days of 

 computing are required for each determination of the absorption of 

 the apparatus, which would be determined once and for all if con- 

 stant reflecting surfaces could be employed. 



With the coefficients of absorption of the apparatus thus deter- 

 mined, each small area included under the holographic curve for a 

 very narrow range of wave-lengths is increased so that the total cor- 

 rected area is then proportional to the solar radiation at the earth's 

 surface as measured with the actinometer or pyrheliometer. Then 

 by the aid of formula i, given above, and employing the trans- 

 mission coefficient a determined from the series of holographs of the 

 day, each small area is again corrected till it becomes proportional 

 to the total radiation of that wave-length outside the atmosphere. 

 The ratio of the sum of these finally corrected areas to the total 

 corrected area at the earth's surface is the factor by which the re- 

 duced pyrheliometer reading is to be multiplied to give the " solar 

 constant " so-called. 



It is evident that these values depend directly upon the pyrhelio- 

 meter or actinometer readings for their accuracy, so that these in- 

 struments become here of major importance. In the work thus far 

 a mercury pyrheliometer has been used as the primary standard, and 

 the daily observations have been taken sometimes with it, sometimes 

 with a Crova alcohol actinometer (specially constructed for the 

 Institution under M. Crova's valued supervision), and sometimes 

 with both instruments simultaneously. It has been shown by re- 



