96 Report S.A.A. Advancement of Science. 



the barometer between those years was essentially different from that 

 of latter years. A slight difference of exposure in a place so subject 

 to violent southerly winds, as is the Cape Observatory, would modify 

 the diurnal curve. (For some remarks on the buildings and 

 exposures, see Gill, Journal Royal Meteorological Society. Vol. 

 VIII.-, p. 238.) It has therefore been attempted to derive the 

 reductions to mean of day from the series of observations, 187 1 to 1905 

 themselves. Stone's figures for each month were first plotted, and 

 the deviations of the later series were inserted, a freehand curve was 

 then drawn, and the reductions to mean of day were taken from this 

 curve. Two solutions were essayed and the mean adopted. Table i 

 shows the whole scheme. 



One example will suffice to show how the observations have been 

 reduced to the mean of day. We take those for January, 1891 : — 



8 hours 

 Noon 

 20 hours 



Means 29.893 29.882 29.885 



The mean of A represents the unreduced mean, that of B is 

 derived by using Stone's reductions, that of C by using the mean 

 of the solutions Ii and L. Applying +o.oo9ins., we have the figure 

 given in the table of mean monthly readings. This example proves 

 that it is impossible to deduce the annual or even the monthly changes 

 in pressure unless there is an accurate reduction to " mean of day." 



It is believed that the same barometer has been in use since 1841. 

 It is Newman's No. 53, the interior diameter of the tube is o.5iins. 

 The scale is attached to an ivory point, which is adjusted to the 

 surface of the mercury at each observation. Although the barometer 

 has remained unchanged, the index errors applied, have varied from 

 time to time. 



During 1841-46, the readings published by Sabine were reduced 

 to 32° F. No statement about capillarity or index error is available, 

 but it is probable that they have been applied. It is the custom at 

 the Cape Observatory to include the correction for capillarity 

 + 0.003) i'^ the temperature reduction, and Stone has evidently 

 considered that Sabine did so as well. 



If the Cape barometer is corrected to 32° F., and for capillarity,, 

 we assume it still requires the following reductions : — 



To Sea-level 37 feet -f- 0.042 



To Standard Gravity 45° Lat. —0.030 



Provisional Index error -0.013 



Total -o.ooi 



