[ 166 ] 
The mean of 43 refults which differ 
not more than 1 5" from the mean 
of the whole is 
The mean of 19 refults which differ 
lefs than 1 5 ', and more than S' * 
from the mean of the whole, is — . 
The mean of 24 refults which differ 1 
lefs than 8" from the mean of the > 
whole is — J 
/ // 
9 16 
= 9 r 4> 
= 9 r 7> 5 
The mean of the above 5 means is = 9 15,8 
And even the mean of thofe 20 refults which dif- 
fer more than 1 5" from the mean of the whole, 
and which are rejected, gives the faid difference 
— 9' 1 2 // -i, which differing only from the 43 
refults, is a proof of the great accuracy in the deter- 
mination of the differences of longitudes by obferva- 
tions of the tranfit of Mercury over the Sun. 
Let us now examine the limit of the errors in thefe 
10 feveral fets of determinations, and we (hall find 
that the limit of the errors in the year 
1723 is = 27 by the internal contadt at ingrefs. 
1736 is — 30 by the external contact at egrefs. 
1743 is — 40 * by the internal contact at egrefs. 
1743 is — 26 by the external contact at egrefs. 
I753 isr=z25by the internal contadt at egrefs. 
1753 is = 33 by the external contadt at egrefs. 
* If we reje<ft the obfervations of M. le Monnier, in which 
there feems to be fome miftake, becaufe it differs confiderably 
from the reft, the limit of the error will be rz 29", agreeing 
nearly with the other limits. 
From 
