POOR, THE FIGURE OF THE SUN 389 



phere, or the astronomical "seeing," was also found by Wagner/ and was 

 corroborated by Gylden at Pulkowa, by Becker at Neuch^tel and by New- 

 comb and Holden^ at Washington. The Washington observations show 

 clearly that both the vertical and horizontal diameters vary with the amount 

 of cloudiness in the sky. In bright weather, w-hen the image was clear and 

 sharp, the diameters as measured were smaller than when the sky was over- 

 cast, or was more or less covered by clouds. 



In the paper above referred to, Auwers also investigated two long series 

 of Greenwich observations; one made by Bradley and Maskelyne between 

 the years 1750 and 1786, and the other made during the years 1851-70. 

 The intervening years were partly covered by two shorter series; one made 

 by Bessel at Konigsberg in the years 1820-28, and the second by W. Struve 

 at Dorpat in 1823-38. These series all showed small variations in the 

 measured diameter; but when these variations were compared with the 

 fluctuations in the number of sun-spots, no dependence of the one phenom- 

 enon upon the other could be traced. As to the minute variations shown 

 by the observations, Auwers concluded that there was "no indication what- 

 ever of the reality of these fluctuations." 



Newcomb and Holden. — New comb and Holden^ investigated the 

 subject of possible variations in the diameter in an entirely different man- 

 ner. If there be two independent series of simultaneous observations, and 

 the difference between each measure and the mean of the whole series to 

 which it belongs be taken, then, if the differences are due entirely to acci- 

 dental errors, there will be no relation between the differences of the two 

 series. If, on the other hand, a portion of the difference is due to ac- 

 tual change in the sun, then, as a general rule, positive differences in one 

 series will correspond to positive differences in the other, and negative 

 differences to negative differences. When the corresponding differences of 

 one series are multiplied by those of the other, and the sum of all the prod- 

 ucts taken, then, if there be a real change in the sun, this sum should be a 

 large positive quantity. 



Unfortunately no such simultaneous series were available. The two 

 series which most nearly satisfied the conditions were those made at Green- 

 wich and Washington during the years 1862-70. Instead of being simul- 

 taneous, the observations were thus separated by an interval of about five 

 hours. These series contained a total of 3639 observations, of which num- 



» Vierteljahrschrift, January, 1873. 



2 On the Possible Periodic Changes in the Sun's Apparent Diameter (American Journal of 

 Sciences and Arts, October, 1874). 



3 On the Possible Periodic Changes in the Sun's Apparent Diameter (American Journal of 

 Sciences and Arts, October, 1874). 



