( 553 ) 
this part has been subtracted from them. Finally a constant cor- 
rection for personal equation, different for each observer has been 
applied to the o—c, in order to reduce the mean value of the cor- 
rected o—e for each observer to 0. 
8. The quantities o—e corrected as explained in 7 consist of the 
accidental errors of observation and the influences of systematical 
errors in the adopted values for the inclination of the axis, the 
constant of collimation and the azimuth. The azimuth-corrections 
might have been determined with greater precision if the observations 
had been or could be made free from the influence of the first men- 
tioned systematical errors. The observations at my disposal do not 
contain any data from which to determine the systematical cor- 
rections of the inclinations adopted; some reflection-observations of 
Poraris from 1882—84, made for this purpose, have not been 
considered as their number was too small for the determination 
of a satisfactory correction of the inclination. Somewhat different 
from this is the opportunity for correcting the constants of collima- 
tion from the observations themselves as they have been made in 
the two positions of the instrument. During the period of the obser- 
vations the transit circle has been reversed 24 times which divide 
the whole interval into 12 periods in which the clamp was east and 
11 during which it was west. 
The longest of these periods lasted 28 months, the shortest nearly 
2 months. 
The influence of an error in the assumed amount of the constant 
of collimation on the azimuth-correction, calculated from the value 
of o—c, changes in sign by the reversal of the instrument; it can 
therefore be found when we compare the azimuth-corrections from 
observations of Polaris, immediately before and after the reversal, 
or, if that error proves to be constant during a longer time, when 
we compare the mean azimuth-corrections deduced from observations 
during a longer period before and after the reversal. 
It now appeared that the observations of Polaris immediately before 
and after the reversals were not numerous and that moreover one 
single observation is not accurate enough to betray a small syste- 
matic error. The graphical representation of the mean azimuth-cor- 
rections during long periods, on the contrary showed clearly the 
influence of a small error in the assumed constant of collimation. 
These mean values during the same position of the instrument showed 
generally a regular variation; if however we would combine the means 
of all the periods in a regular chronological order, then the curve 
40 
Proceedings Royal Acad. Amsterdam, Vol II. 
