Existence of an Atmosphere surrounding the Sun. 451 



The period 25'^"240 for the sun's sidereal rotation satisfies 

 best the above longitudes, and has been used, as in the first 

 example, in computing the dates t. In the first column of the 

 next Table are given the quantities (p) ; in the second the cor- 

 rections to / on hypothesis m= 1-010, computed by a fresh 

 reduction with the same elements used before; in the third, 

 the uncorrected discordances of t ; in the remainder as indicated. 



The discordances of the first series are larger than those of 

 the second, as was to be expected fi'om the spot being of greater 

 size, and ofi"ering a less definite centre for observation. It will 

 be necessaiy to have regard to this circumstance in considei'ing 

 the uncorrected and corrected sums. The first hypothesis on 

 the whole improves the result ; the second, while it diminishes 

 the total, sensibly increases the sum of the second series ; and 

 the third, judging principally by the second series, shows 

 decided signs of over-correction. If we now treat the discord- 

 ances as before by equations of condition, the method of least 

 squares gives by the first series ot=1 '008176, and by the 

 second, which is to be preferred, m = 1*001998; by the whole 

 m= 1 '005684. If we hold by the evidence of the second series, 

 we have 1'0020 as the most probable value of m, precisely the 

 same which resulted from our former discussion. If it be 

 thought better to include the first series, we find a somewhat 

 larger value. As a general conclusion we may say that the 

 motions of the solar spots in longitude will be more nearly pro- 



