( 276 ) 
all parts of the heavens may be observed without much difficulty. 
That in this way I have succeeded in rendering the stability of 
the instrument very satisfactory, may be seen from the results of 
the transit observations in the meridian which | have lately made 
for finding the time. 
The observations of 6 October may serve as an instance. Within 
a space of 1 hour and 15 min. 9 complete levelings of the horizontal 
axis were made and the greatest difference between the results 
obtained amounts to 3”.0; as, however, no regular variation can 
be seen, it is probable that these differences are due for the grea- 
ter part to errors of observation. 
To facilitate the mounting of the instrument in the required 
position, I determined the azimuth of the harbour-light situated 4 
kilometers to the south and found therefor 1° 23' 7". 
Before the arrival of the universal instrument I used a sextant and 
an artificial horizon for finding the time and the latitude. It is 
obvious that these earlier observations are of no value for determining 
the latitude which has been found much more accurately by means 
of the universal instrument. Yet from another point of view 
they deserve notice. The sextant used was a different instrument 
from the one with which I formerly observed at Ambriz and San 
Salvador. Its errors have been investigated in 1899 by Mr. 
L. RooseNBurg, director of the Amsterdam branch of the Netherland 
Meteorological Institute; they were found to be lying between 
+ 7" and — 6". 
The chronometer used was one of THomas Hewirr; its rate was 
tested at regular intervals by determinations of time. 
Except of late these determinations were made by means of 
observations of zenith distances at great hour angles, first from 
March till August 1900 with the sextant, after that time with 
the universal instrument. With the latter instrument I always 
observed the zenith-distances of stars. The way in which I made 
these observations will be deseribed in the next section. Here I 
only give the formulae borrowed from “CASPARI Astronomie pratique’, 
by means of which the mean hour angle was derived from each pair 
of observations made in the two positions of the instrument, using 
only an approximate value of the zenith-point. 
Let z and z’ be the two zenith distances, t and ¢’ the corresponding 
- hour angles; let 
cos Z = cos 4 (ze — 2) cos 3 (2 + 2’) 
