174 Report S.A.A. Advancement of Science. 



a small mirror fixed to wall behind it, and a telescope arranged to 

 observe the seconds hand of the chronometer reflected in the mirror. 

 The pendulum wire (slightly out of focus, of course) swung in front 

 of the mirror, and could therefore be seen along with the chrono- 

 meter. A set of five thousand complete oscillations was observed, 

 the time being noted for every hundredth up to 2000, and every five 

 hundredth from there onwards. The period for the first two 

 thousand and for the whole set was calculated in the usual manner (o 

 to 2500, 500 to 3000, etc) and corrected for amplitude. 



Semiamplitude. at start 8 cm. 5^ cm. 



,, finish 3 cm. 2^ cm. 



Temperature 10° to 18° 7° to 13°. 



Chronometer loses i in 18000. 



Corrected period 2.68845 | 2.0490. 



g = 4 n-' 



179.19 



4 n^'°+°7 



2.0490^ 



.68845^ 



= 978.71 I = 978-59- 



The difference between numbers may be accidental or may be due 

 to an error in the position of the centre of inertia of the ball. I think 

 it is due to the former cause chiefly, and as the measurements cannot 

 claim an accuracy beyond the first place of decimals, I take the result 



to be : — 



978.7. 



The latitude of the physical laboratory is 26° 11' south, and its 

 altitude 1753 metres. Helmert's formula, 



g=98o.62 — 2.6 cos 2^~ gives 978.50. 



tor a table land theory would give a correction for altitude of only f 

 as great ; it is usually stated that the table land formula is inappli- 

 cable owing to displacements in the density of the underlying strata, 

 but this would seem not to be the case with the great South African 

 table land, as the formula 



g=Q8o.62 — 2.6 cos 20-^. gives g=978.7o 



* ^ 8 3300 



in agreement with the observations. Since there are only two 

 really large table lands in the world — South Africa and Tibet — 

 some geophysical interest attaches to the results, and it is hoped they 

 may be repeated with better apparatus. 



