( 288 ) 
The mean of these values is 
—=/5°-12'4",8 = 21 
in sufficient agreement with the result obtained above. On the 
other hand we find for the difference of the two: 
p Northern star — p Southern star — + 24".2 + 4.2 
which result points to the existence of considerable systematical 
errors. 
As the errors of my sextant had been investigated before (comp. 
above) and as the adjustment of the mirrors and the telescope 
was always tested, while moreover the readings for the two stars 
differed only 20°, I was led to conclude that the difference between 
the two values results from a personal error in making the two 
images of the star coincide. The value of this personal error would 
then. be 12”. 
If I conclude that this is the case and assume that such a personal 
error already existed in my observations in 1894 and 1895, then 
I have to apply corrections to a few results obtained from the 
zenith distances of w Crucis for the latitude of Ambriz and of 
San Salvador!). By doing so the agreement between these results 
and those derived from observations of the sun would much improve. 
Therefore I hold this supposition not improbable and I will 
indicate how, if it is accepted, the values formerly adopted for the 
latitude of the two places must be changed. 
For Ambriz we find (see lc. pag. (405) 8) as the result of the 
meridian altitude of @ Crucis: — 7° 49'57".5, while 6 observations 
of the meridian altitude of the sun give 57’.2 and the mean result 
becomes 57’.3 instead of 59.0. 
The mean result from the 3 series becomes then: 
— 7° 49'59".6 
instead of 60".2, or it remains: 
— 7° 50'0". 
— 
1) C. SANDERS, Determination of the latitude of Ambriz and of San Salvador 
(Portuguese West-Africa). Proc. Acad. Amsterdam, vol. IL, p. 898. 
lll pe 
