IV APPENDIX. 



caused by a periodic change in the sun itself, is shown by the fact 

 that these mean monthly "outstanding errors" have on the whole 

 opposite signs at Greenwich and Washington ; hence it is inferred 

 that they are in great part due to atmospheric causes. To test 

 this the Washington observations from 1867 to 1870 have been 

 tabulated in Addendum "C," where I have placed the mean 

 "outstanding error" of each year under one of the numbers I to 

 V. These numbers are the observer's estimate of the goodness 

 of the image of the sun at the time of observation, I being a 

 poor image and V a perfect one. It is plain from this table that 

 the goodness of the image has a direct influence upon the value 

 of the diameter deduced, as Wagner first showed in his own case. 

 (VierteJjahrs. Ast7\ Gesell., 1873, Jan.) I treated Wagner's 

 own observations, which are given in the cited work, in the same 

 manner as I had previously treated the Washington observations, 

 and I found that Wagner's class 4 (on a scale from 1 to 6) gave 

 mean outstanding error of — 0^055, while the Washington class 

 II-III, which exactly corresponds to it, gave — 0^063 : also, 

 his class 3 gave -[-0^029, while the corresponding Washington 

 class ail-IV) gave -f 0M80. 



The agreement of signs shows that the influence of the goodness 

 of image on deduced horizontal diameters is in the same direction 

 at Washington and Poulkova : the discrepancy in amounts is 

 probably due to different habits in assigning the weights. 



To show how rough a division of observations on account of 

 state of the image will show an influence in the deduced diame- 

 ters, I divided the Washington observations into two classes : 

 1st, those made when the sky was perfectly clear ; and 2d, those 

 made when, according to the estimates of the watchmen of the 

 Observatory, the cloudiness was from 5 to 10 (10 = wholly 

 cloudy). The results are given below : — 



Table of Outstanding Errors in Sun's Diameter as affected by the 

 cloudiness of the sky at noon. 

 Horizontal Diameter, Vertical Diameter. 



Year. Cloud = 0. Cloud = 5 to 10. Cloud = 0. Cloud = 6 to 10. 



1867 —0^033 +0^ 



1868 —0.047 +0 



1869 0.000 4-0 



1870 —0.093 +0 



I applied the same test to Greenwich Obs. for 1866, and found: 

 Horizontal Diameter. Vertical Diameter. 



Cloud = Cloud = 0-1 0. Cloud = 0. Cloud = 5-10. 



1866 —01022 +0'.010 + 0".55 — 0".41 



which agrees in horizontal diameter with Washington Observa- 

 tions, but is contrary in vertical diameter. 



(4) 



