Gravity in Soittli Africa. 
89 
To summarise, we may say that the chief sources of error in the 
experiment, and kinds of error they give rise to, are : — 
Irregularity in rate of clock Diurnal. 
Variation of temperature Diurnal. 
Unsteadiness of support Local. 
Change in Length Secular. 
Change in strength of current Irregular. 
Eesults. 
Johannesburg. 
Cape of 
Vereenig- 
Oood 
Hope 
Lourengo 
Pola. 
School of 
Observa- 
ing. 
Observa- 
Marques. 
Mines. 
tory. 
tory. 
0-50 
Period of 23 
85315 
82409 
83977 
79970 
44 
88974 
86109 
87627 
83694 
„ 62 
81833 
81870 
81522 
79038 
80536 
76642 
Diff. 44—23 
3659 
3700 
3650 
3724 
g2 
7104 
7071 
7091 
7052 
Diff. from Pola 23 
5345 
2449 
4007 
44 
5280 
2306 
3933 
62 
5191 
5228 
4880 
2385 
3894 
Mean 
5284 
2413 
3945 
Corresponding difference 
in " g " 
2004 
2040 
1884 
931 
1523 
"5^ " observed 
9-78638 
9-78602 
9-78758 
9-79711 
9-79119 
9-80642 
9-79011 
9-79011 
9-79046 
9-79618 
9-79001 
9-80584 
Correction for altitude 
- 538 
- 556 
- 442 
- 5 
- 17 
- 9 
+ 193 
+ 195 
+ 160 
- 2 
+ 4 
+ 3 
"5f" calc 
9-78666 
9-78650 
9-78764 
9-79611 
9-78988 
9-80578 
- 28 
- 48 
- 6 
+ 100 
+ 131 
+ 64 
The above table gives a summary of the observations taken at the 
stations described. The concordance of the measurements is not as good 
as might be wished, and it seems probable that some secular change in 
the pendulums is the cause of this. Accordingly the old observations at 
Pola may not be properly comparable with those in South Africa, and it 
would be well to swing the pendulums again at Pola. The absolute values 
of the anomaly may therefore be wrong ; at the same time the general 
character of it is unmistakable, and there seems no reason to doubt the 
values found for the difference between sea-level and the plateau. Prob- 
ably the most trustworthy of all the observations are those at Louren9o 
Marques and Johannesburg Observatories. These were both done under 
favourable conditions, and within a month of one another ; and as the 
difference in latitude is trifling, the effect of level stands out with special 
clearness. The difference in anomaly is between + 131 and - 36 or 
177 units, pointing, as far as one can see, to a greater density of rock 
in the lower strata near sea-level than on the plateau. 
