478 
Whereas the deviation from the previous results of those obtained 
in Cabinda might be explained by their smaller accuracy, the great 
deviation of the more accurate series Matuba II is very striking, 
and it becomes much more so, if the flexure is represented by 
«sin z, and the coefficients c are compared. This cannot be attributed 
to accidental errors, (cf. the m.e. + 0'82 found above as against 
+ (66) and there seem to be only two explanations possible, either 
we must assume a change with the time, or suppose that the flexure 
may even change its sign with the zenith distance. The latter was 
namely in the last series noticeably smaller than in all the previous ones. 
At any rate it is necessary for each series to use the flexure 
deduced from it. 
— 
5. Determination of the longitude of Matuba and Cabinda. 
The difference of longitude between Matuba and Chiloango, and 
later that between Cabinda and Matuba, was determined geodetically, 
which in the given circumstances seemed preferable. 
a. Determination of the geographical position of 
Matuba with respect to Chiloango. 
As early as 1901, 1 bad connected my observation-pier at Chiloango 
with the flag staff of the Residency at Landana and in 1910 1 
connected my new observation place at Matuba with it also, by a 
tacheometric measurement of the road between the two places. I 
determined their relative magnetic coordinates, and, in order to be 
able to reduce these to astronomical ones, I made at the same time 
some determinations of the magnetic declination in Chiloango, in 
which the already determined azimuth of the harbour light provided 
me with the absolute orientation (once the sun was observed directly). 
For the magnetic declination I obtained, with a needle of 86 mm. length 
1910 Det” 3, perm: 14°. 40) 35" West 
er Aran 39 31 
Nov..2. p. m. 45 53 
et tekenen SN 
eee 52) OD 
on rede EN: 35-38 
4 vetep. mn. 50 22 
Mean 14° 42'7 West 
