NO. 4 SOLAR RADIATION ABBOT, FOWLE, AND ALDRICH 9 



intensity outside the atmosphere ; m, the air-mass, and a the coeffi- 

 cient of atmospheric transmission, assumed as constant. If log a 

 only apparently, not really, is constant, our results are wrong. Both 

 Mr. Very and Mr. Kron indicate pointedly that they believe log a is 

 not constant, but that in fact the transparency of the atmosphere 

 continually diminishes during the forenoon periods we have chosen 

 for our observations, so that our transmission coefficients are too 

 high, and our value of the solar constant too low on account of this 

 source of error. Mr. Kron indicates possible errors of the solar 

 constant values of not more than 5 per cent as due to this cause. 



It appears, however, that Mr. Very attaches great weight to this 

 second objection, for he says of the work of Abbot, Fowle, and 

 Aldrich : 



The neglect of diurnal variation of atmospheric quality, and the erroneous 

 supposition that the same coefficients of transmission can be used at all hours 

 of"the day, completely vitiate these reductions. 



Again he says : 



The Smithsonian observations, for example, usually stop when the air-mass 

 becomes as large as 3 or 4 atmospheres. Some do not even extend to 2 

 atmospheres. Reduced by Bouguer's formula these mid-day readings agree 

 among themselves, but solely because they have stopped before reaching the 

 point where disagreement begins. This is- equivalent to shirking the diffi- 

 culties, and the seeming extraordinary agreement of the measures is mis- 

 leading. If the missing readings had been supplied the discrepancies would 

 have been obvious. Such incomplete observations are incapable of elucidating 

 the laws of atmospheric absorption except through the aid of more perfect 

 measures. By supplying deficiencies under guidance of a criterion we may 

 in some cases rescue observations which are, otherwise, useless. 



Again he says : 



The portion of the diurnal curve between the limits of 4 and 10 atmospheres 

 conforms tolerably well to the conditions needed for a determination of its 

 slope and general form, and, as a rule, it would seem to be the best part of 

 the curve to select for computation. 



3. Mr. Very states that we adopt too high a value of the absorption 

 of terrestrial radiation by water vapor and too low a value of its 

 absorption of solar radiation. 



excellence of the observing conditions. In such applications to pyrheliometry 

 we recognize, however, that A would not be the solar constant. In this 

 connection see figure 3, in which, although for a range of 20 air-masses there is 

 a steady and well-marked curvature in the plot of pyrheliometry, any range of 

 only two air-masses shows this but little. We, therefore, fail to see how Mr. 

 Very's emphatic criticism of our procedure in this respect, which he gives in 

 the French article above cited, is justified. 



