504 



SCIENCE 



[N. S. Vol. XXVII. No. 691 



California by the Smithsonian Astrophysieal 

 Observatory indicate that in the two periods 

 June-October, inclusive of 1905 and 1906 

 there were variations of the " solar constant " 

 between 1.95 and 2.14 calories. Some of these 

 apparent variations were of short duration, 

 as ten days for example, others of several 

 months. 



All the determinations, 130 in number, are 

 by the method of high and low sun measure- 

 ment with homogeneous rays. The probable 

 error of a single good determination (and 

 nearly all were excellent) is about one per 

 cent. But it is urged that the method 

 though apparently sound in theory may in 

 fact fail to estimate properly the transparency 

 of the air. Supporting evidence of the sound- 

 ness of the work is as follows: 



The mean value of the " solar-constant " 

 observed on Mt. Wilson in 1905-6 is 2.03. 

 The mean value observed in Washington 

 1902-6 is 2.06. There is one mile difference in 

 altitude, 3,000 miles in longitude, and in the 

 one case the average maximum solar intensity 

 at the earth's surface is about 1.60, and in 

 the other only 1.20. This close agreement 

 outside the atmosphere seems strongly con- 

 firmatory of the accuracy of the means of 

 estimating the transparency of the air. 



On seven days in 1905 and 1906 practically 

 simultaneous measurements were made at 

 Washington and Mt. Wilson. The Washing- 

 ton results were systematically a little higher, 

 about 3 per cent, in fact. Allowing for this 

 small systematic difference the remaining 

 average deviation is only 1.5 per cent. 



Considering Mt. Wilson work again, there 

 is no relation either direct or inverse which 

 holds consistently between the variations of 

 the " solar-constant " and the variable ele- 

 ments of the air, such as its transparency at 

 different wave-lengths and its humidity. The 

 fluctuations of the " solar constant " observed 

 are not hap-hazard in character, but several 

 days or weeks of observation lead gradually up 

 to a maximum or down to a minimum. 



If it were in fact true that " solar-constant " 

 measurements on Mt. Wilson are competent 

 to indicate variations of 1 per cent, in solar 



radiation the real change of several per cent, 

 due to the elipticity of the earth's orbit ought 

 to be plainly shown. This is the case in both 

 1905 and 1906. 



Confirmation from the earth's temperature. 



With the aid of numerous measurements 

 made on Mt. Wilson of the reflection of clouds 

 at different angles and of the brightness and 

 transparency of the sky, the value of the 

 earth's albedo has been determined to be 37 

 per cent. Combining this with the mean 

 value of the " solar constant " (2.1 calories in 

 round numbers) there remains 1.32 ir B' calo- 

 ries available to warm the earth as a planet. 

 This would maintain a perfect radiator or 

 " absolutely black body " at 256° absolute 

 temperature. The radiating surface of the 

 earth as a planet is not its solid and liquid sur- 

 face, for clouds during half the time cut off 

 terrestrial rays entirely. During the other 

 half, as the measurements of Rubens and 

 Aschkinass, Langley, Keeler, Very and 

 Nichols suffice to indicate, the water vapor is 

 sufficient to cut off nine tenths of the ter- 

 restrial rays. Accordingly the earth's radia- 

 ting layer may be set at about 4,000 meters 

 above sea level, where the mean temperature is 

 about 263° absolute. If this layer were a 

 perfect radiator, its maximum emission would 

 be 1.45 TT R' calories, so that the solar constant 

 can not exceed 2.33 calories unless the albedo 

 exceeds 37 per cent. This maximum possible 

 value is in reasonable accord with 2.1, the 

 adopted value, and differs no more than would 

 be expected in view of the uncertain tempera- 

 ture and radiating power of the earth's radia- 

 ting layer. 



A statistical comparison of the monthly 

 temperatures for the last quarter century of 

 47 inland stations distributed as well as pos- 

 sible over the world indicates variations, often 

 more than four times the probable error, in 

 the mean temperature of the world. These 

 changes of temperature, amounting to several 

 degrees centigrade, may be caused by varia- 

 tions of the sun. 



The magnitude of temperature changes 

 caused by solar changes of short period is 

 much greater for inland than for coast or 



