Dr. Herschel &c. 365 
per opportunities. The only way then left, was to take the 
fituations of thefe fateliites, in any place where I could afcer- 
tain them with fome degree of precilion, and to reduce them 
afterwards by computation to fuch other fituations as were 
required for my purpofe. 
In January, February, and March, 1787, the pofitions 
were determined by cauhng the planet to pafs along a wire, and 
eftimating the angle a fatellite made with this wire, by a high 
magnifying power ; but then I could only ufe fuch of thefe fitua- 
tions where the fatellite happened to be either dire&ly in the pa- 
rallel of declination, or in the meridian of the planet ; or 
■where, at leaft, it did not deviate above a few degrees from 
either of them ; as it would not have been fafe to truft to more 
diftant efiimations. I11 October I had improved my apparatus 
fo far as to meafure the pofitions by the fame angular micro- 
meter with which I have formerly determined the relative po- 
fitions of double flars *. 
I11 computing the periods of the fateliites I have contented 
myfelf with fynodical appearances, as the pofition of their 
orbits, at the time when the fituations were taken from which 
thefe periods are deduced, was not fufficiently known to attempt 
a very accurate fidereal calculation. By fix combinations of 
pofitions at a dift'ance of 7, 8, and 9 months of time, it 
appears that the firft fatellite performs a fynodical revolution 
round its primary planet in 8 days 17 hours 1 minute and 
19,3 feconds. The period of the fecond fatellite deduced 
likewife from four fuch combinations, at the fame difiance of 
time, is 13 days 11 hours 5 minutes and 1,5 feconds. The 
* For a defcription of this inilrument, fee Phil. Tranf. Vol. LXXI. 
p, 500. arul Vol. LX XV, p. 46. 
C C C 2 
com- 
