( 284 ) 
From the mean error of 2 pointings + 1.55 follows for that of 
the result of a star with 5 pairs of pointings, apart from the systema- 
tical errors, = 0.69, and for that of the mean result from a northern 
and a southern star + 0.49. Hence we see that there are conside- 
rable systematical differences between the results from the northern 
and the southern stars, which points towards considerable values for 
the systematical division errors, but at the same time we find con- 
firmed that their influence is eliminated already for a great part in 
the mean value 3(N +8) for each position of the circle. 
If once more we take the mean of the results of the different 
positions, we finally arrive at: 
gp = — 5° 12'4".01. 
If we had kept the two series separated until the end, we had 
found in the same way : 
15 series —— D42' 37,85 
Drek ne 05 
or if we had used in the first series only the 4 positions 0°, 90°, 
180° and 270° 
—5° 12'3".54, 
Hence it appears that after all it does not matter much which 
relative weights we assign to the two series and as final result for 
the latitude of my place of observation I consider the value: 
gy = — 6° 12'4'.0 
a.vaiue which probabiy will be not more than 1” in error. 
Although we have thus obtained a result as free as possible from 
the influence of division errors and flexure, I wanted also to know the 
value of these errors themseives, especially in order to be able to 
determine their influence on the chronometer corrections derived 
from my observations. 
Let p be the zenith-point and z the zenith distance of a star; 
we can derive from each observation an equation in which occur 
the difference between the division errors for the points p+e and 
p—z and the flexure for the zenith distance z. But much simpler 
and with a sufficient degree of accuracy we may proceed as follows. 
