90 Transactions of the Boyal Society of South Africa. 
The result may be put in this form : — 
" g " Louren^o Marques - " Johannesburg = + 517 
Difference due to latitude = - 10 
.-. Difference due to altitude (1805 m.) = + 527 
or 3*4 metres per unit (10~^ m./sec.^). The comparison between Lourengo 
Marques and Vereeniging gives practically the same. This quantity has 
usually been stated as 5 metres. 
An explanation of this, suggested to the writer by Prof. Schwarz, is 
that the South African plateau is a sort of floating island of lighter rock, 
the roots of which go down to a depth at which the rock in most regions 
is denser (of basic material) — something like an ice-floe on water. The 
suggestion is quite a possible one physically, and it is easy to adjust the 
arithmetic to it. Thus, supposing the defect in density for a depth of 
10 kilometres below sea-level were about 0-5, or if the deficiency were 
less, but extended to a greater depth, the anomaly in " ^ " would be 
accounted for. 
Calibkation of Theemometee 53. 
Temperature Centigrade. 
1894. 
1895. 
1910. 
8 
8-85 
8-60 
8-61 
9 
10-49 
10-25 
10-18 
10 
12-10 
11-92 
11-75 
11 
13-73 
13-55 
13-32 
12 
15-31 
15-20 
14-89 
13 
16-94 
16-82 
16-46 
14 
18-51 
18-40 
18-03 
15 
20-10 
20-00 
19-60 
16 
21-71 
21-62 
21-17 
17 
23-27 
23-20 
22-74 
18 
24-84 
24-80 
24-31 
The last column is equivalent to the equation — 
t = - 3-95 + 1-57 X. 
