80 REPOKT OF THE SECRETAKY. 



Table I.— Coefficients of atmospheric transmission for radiation from zenith sun. 



Wave length 



0.40 0.45 0.50 0.60 0.70 0.80 0.90 1.00 1.20 1.60 2.00 



Transmission coefficients lor unit air mass. 



October '2.5 . . 

 November 2 



March 21 



Mays 



September 11 



October 9 



October 15 . .. 

 October 16... 

 October 22 . . . 

 November 15 



iy03. 



February li> 67 



February 25 \ . 4H 



March 3 40 



Marcli 25 47 



March 26 



April 17 



A pril 28 39 



April 29 46 



•Mily 7 .' 42 



(ieneral mean 



Mean ol 1902 



Mean of 1903 



Percentage difference between 

 mean of 1903 and that ol 1901-2. 



0.64 

 .60 

 .48 

 .50 

 ..58 

 .60 

 .52 

 .49 

 .60 



.80 



.77 

 .78 

 .78 

 .78 

 .58 

 .82 

 .79 



.72 

 .68 

 .73 

 .66 

 .68 

 .77 

 . 64 

 .66 

 .69 



.84 

 .90 

 .87 

 .84 

 .86 

 .79 

 .88 

 .83 



.700 

 .765 

 .627 



.730 

 .769 

 .692 



.808 

 .8.57 

 .753 



0.95 

 94 



.91 

 .93 

 .93 

 .96 

 .91 

 .95 

 .95 



.90 

 .93 

 .96 



.88 

 .89 

 .97 

 .88 

 .88 

 .89 



0.94 



.920 

 .930 

 .909 



.919 

 .950 

 .912 



It is to be regretted that the earlier work did not reach up so far in the violet as 

 we now observe, but the trend of the observations makes it appear that the trans- 

 parency of the air for the extreme visible violet rays may be 30 per cent less than 

 last year, and that the transparency for the visible and infra-red spectrum as a whole 

 has diminished by 10 per cent. If this change is widespread it should be likely to 

 influence climate, and that it is widespread the falling off of actinometric observa- 

 tions both in this country and Europe, as reported by several observers, would seem 

 to indicate. 



Selective atmospheric absorption. — Referring to fig. 3, page 89, of my last year's report, 

 the depression at a wave length of 0.58// in the curve there shown would a])pear less 

 marked in the more recent results, owing to the great decrease in transparency for 

 the blue and violet rays; but nearlj- all recent work gives evidence of bands of dimin- 

 ished attiiospheric transmission at wave lengths 0.43/i, 0.48/^ and 0.58//. 



Interesting results have been obtained in respect to the atmospheric transmission 

 within the great bands of water vapor and oxygen, and it has been found, in con- 

 firmation of the accuracy of the observations, and especially of the exponential 

 formula employed in reducing them, that the values of transmission coefficients 

 within the water-vapor bands of the infra-red are such as would very nearly obliter- 

 ate these bands from an energy-sisectrum curve corrected to represent the distril)ution 

 of solar radiation in the spectrum outside the earth's atmosphere. 



Actinometric and ijyrlieliometric observations. — In the autumn of 1902 an alcohol acti- 



