542 
So no improvement is found for all the observations together ; 
and although this is indeed the case for those at Jidda, the value 
of Il remains still considerably higher than the one found for III, 
which shows that even when mean values are used the mean error 
of the zenithpoint has not yet become really small. 
We shall now consider the values of IV and V, which, not taking 
into account the influences of flexure and division-errors, must be equal 
1 
to id Now these two errors must have been almost eliminated 
; 1 é br Bee: 
in the 5 (yy t+ fg) Owing to the nearly equal zenithdistance of North- 
_ 1 
and Southstar, but they may be considerable in the 5 (Px--Fs); 
and as on different nights couples of different zenithdistance were 
observed, the value of IV must also have been increased by that 
influence. 
1 
We now find, adding for comparison the values of ae III 
Jidda Mecca Together 
IV. (43°36)? = 11.26 (4 4".85)? = 293.51 (4 444)? = 17.15 
Vo» Gr 4.14" =17.12,, G5 ..06)* = 25:60, (Fed 56)? 20 ae 
1 
B UI (43.74 =1400 (4 4.99) = 2486 (-- 4.31)? ian 
So we see that the values found for IV are not only not higher 
but on the contrary somewhat lower than those of V and that both 
are almost equal to —. II, on which flexure and division-errors 
—_ 
must have had some influence too. From this we may conclude 
that the two influences cannot have been great. 
Coming now to a consideration of the mean results for y in the 
different positions, we shall first compare those with the telescope 
left and right. | 
Denominating the correction of the employed zenithpoint AZ then 
we see that 
Northstar 7. L;. .A@= an 2 
TB Alps Pr PR=d2AZ 
Southstar 7. Lp Ag=+AZ 
T. Bo Ap Ay fy, —Pp=—IAAZ 
Thus oh? See Oi ak Psst 
