A STUDY OF CORRELATIONS AMONG TERRESTRIAL TEMPERATURES. 333 



spots. This periodic change in the amount of solar spottedness indicates that a change 

 of some sort is going on in the sun ; and if the radiation of the hitter is subject to any 

 periodic change, we must expect this to be one of the principal periods. Two methods 

 of investigation are open to us, which would be identical if the variation in the spotted- 

 ness were a rigorously harmonic fluctuation in a fixed period. One is to take the 

 degree of spottedness from time to time as the term of comparison ; the other is to 

 assume a period in the general terrestrial temperature exactly equal to the mean period 

 of the spots, and determine the coefficients of the fluctuation so as to best satisfy the 

 observations. The second method seems preferable because we have some reason to 

 suppose that the degree of spottedness is a secondary rather than a primary phe- 

 nomenon. The writer showed in his paper on the period of the solar spots that the 

 irregularities in the period of the observed phenomenon tended to compensate them- 

 selves, in the course of time returning to an original primordial period. This was 

 especially shown by the fiict that about 1760-90 the epochs of maximum and mini- 

 mum were accelerated for several years, but afterward returned to the original places 

 in the period. That is to say we have in the spots a fairly exact period subject to 

 fluctuations on one side and on the other. Now the change in radiation is as likely 

 to follow the rigorous period as to follow the apparent phenomena of spots. 



The irregular and fragmentary character of our data affords another reason for 

 taking as the basis of our work the hypothesis of a period of 11.13 years simply. If 

 we had at our disposal a uniform and homogeneous system of observations in various 

 widely separated regions, extending through a long period, either method could be 

 applied with equal facility. But the fragmentary character of the actual data would 

 render weak a comparison of the temperature during a period of such great bespotted- 

 ness as that of 1870-71 with that of the year 1900, during which there were very few 

 spots. 



The most exhaustive attempt with which I am acquainted to discover the relation 

 between the solar spottedness and the terrestrial temperature is that of Koppen.* The 

 material made use of comprises mean fluctuations of temperature in various regions 

 of the globe, from 1767 to 1877. The regions were classified according to their lati- 

 tude as tropical, sub-tropical, warm, temperate, etc. The general conclusion was that 

 the temperature of the tropical regions was lower by about 0°.73 C. near the time of 

 maximum sun spots than near the time of minimum. It is known that the spots 

 radiate less heat in proportion to their surface than does the photosphere, and the 

 general nature of this result is the same as if the temperature per unit area of the 

 non-spotted photosphere were invariable, so that the total radiation was diminished 



* Zeiischrift der Oesterreichen Gesellschaft fiir Meteorologie, VIII Band. 

 A. P. S.— XXI. TT. 13, 1, '08. 



