ÖFVERSIGT AF K. VETENSK.-AKAD. FÖRHANDLINGAR 1899, N:0 2. 93 



than tlie true one, and an observation on the west limb an 

 excessive value if the observations are made before noon, and 

 the contrary if the are made after noon, and thus if we take 

 the mean of observations on both limbs the result would be 

 free from error, if only both observations, the means of which 

 are taken, were made on the same time, that is however im- 

 possible. I will now show the amount of such an error. For 

 convenience sake I assume the sun's declination to be zero and 

 that the observation is made on a geografic latitude like ip, 

 tlien the velocity of the spöt of observation is 0.46 cos \p km. 

 per second; I further assume that the observation on the east 

 limb is made on the meridian and thus without errors of that 

 kind, and that the observation of the corresponding point on 

 the west limb is made half one hour afterwards; then an error 

 like 0.06 cos ip affects the observation on the west limb and 

 thus an error of 0.03 cos i/^ the result. Now the mean af Duner's 

 observations on the heliographic latitude q) = 75° is 0.34 and 

 the geografic latitude like 60°, thus the occurring error here, 

 under the assumed circumstances, would he h % of the whole 

 velocity, an error which is constant, if the observation always 

 begin on the same limb of the sun. Very probably here the 

 time past between the two corresponding observations on both 

 the limbs is generally less than Vo ^^our and do not always 

 begin on the same limb of the sun. In any case, on niaking 

 such observations for determining the constants of the sun, it 

 would be good to mark the time of observation and correct 

 for the motion of the earth, especially as the velocity of the 

 sun in high latitudes is very small and even a little error will 

 seriously effect the accurate determination of i and i3. 



The only determination of i and i2 from observations of 

 the sunspots which is published in extenso, is so far as I am 

 aware that of WiLSiNG in A. N. Bd. 107 pag. 282 and he has 

 determined not i and Ü. but di and dQ with differential for- 

 mules. There is a question whether not a determination of 

 these quantities i and ß themselves with exact formules would 



