93 
of gravity at as many points of the earth’s surface as possible, but 
in no wise very great accuracy in the determination of their abso- 
lute value. We may change the value of the acceleration (g) of 
gravity by 1oo units of the fifth decimal, without thereby percep- 
tibly affecting the results of the comparison on which it depends. 
Although, therefore, we may regard the absolute value of gravity 
in general as already known, yet we must not on this account con- 
sider its determination as definitely set at rest, especially as the re- 
sults hitherto obtained still differ considerably from each other. 
This is clearly shown by the following investigation, carried out 
quite recently. Starting from Vienna, very accurate relative deter- 
minations of gravity were carried out at many stations at which 
the absolute value had been already previously determined. 
The thorough trustworthiness of the results of these relative de- 
terminations was proved on the occasion of their repetition with 
different apparatus, at different times, and by different observers, 
which led to the same result. 
If the various absolute determinations had been perfectly correct, 
the results deduced from them by means of the differences deter- 
mined for Vienna Geographical Institute must all be the same. 
The results are as follows, expressed in the lengths of the seconds 
pendulum L.W., for Vienna Geographical Institute, as deduced 
from absolute determinations by 
MM. 
I. Peters in Berlin, 1870, L.W. = 993-745 
2. Lorenzoni “ Padua, 1885, coc = 750 
3. Anton « Berlin, 1878, 6. Fh 2.60 
4. Peters « Altona, 1869, 668 | 9 = .763 
5: Mahlke « Hamburg, 1891, CONE MR ko 
6. Peirce « Berlin, 1876, a -791 
7. Bessel « Berlin, 1835, a .804 
8. Biot « Padua, 1820, 5c 805, 
g. Sabine « Altona, 1828, a .810 
Io. Oppolzer < Vienna, 1884, ip OBA: 
(Tiirkenschanze) 
11. Defforges < Paris, 1884, te a 835 
12. Orff «¢ Munich, 1877, rete Rey 
As we see, the results exhibit some important differences, which 
appear to be attributable to systematic errors. We may put down 
some of them to imperfection of the comparisons of the scales used 
for the absolute determinations. 
