The Barometer in South Africa. 71 



temperature correction of Kew barometers does not depend on the 

 mercury and scale alone, but also on the size and material of 

 construction of the cistern, for which no allowance has been made. 

 In countries of high latitude where the fluctuations of pressure are 

 considerable, Kew barometers may be useful, but in South Africa,, 

 where the range is small, we must rely on the Fortin pattern and 

 educate observers in its use. With the observations available there 

 are many breaks and very often very large errors. Practically all the 

 observations published by the Cape Meteorological Commission have 

 been extracted and tabulated by me, and the best of the material 

 (where a choice exists) used. In some cases where there are very few 

 stations and very few years' records, we have perforce had to use 

 them. The first result of an inspection of these sets of observations 

 is to show that the annual variation of the barometer over the whole of 

 South Africa and the adjoining regions consists of a well-marked 

 single oscillation, pressure being greatest in winter and least in 

 summer. The only possible exception to this rule is with Swakopmund 

 in German S.W. Africa, in which case there is a slight divergence 

 which would perhaps disappear if a longer period of observations 

 than 4 years were available. The annual variation is, in the main, 

 an almost pure sine-curve, that is, it can be represented by an 

 expression of this form 



A sin {t + b) 

 in which A is the difference of the highest or lowest reading from the 

 mean of the year. The time of the year is represented by t as 

 follows: — January i5th = o°, February 14th 30°, and so on ; d 

 represents the difference of phase. A few examples will make the 

 meaning of this formula quite clear. If for a given place we have 



Bar = mean height -I- o. 1 1 2 inches sin (t-l-27o° 

 it tells us that in the middle of January when t = o° and sin 270°- i 

 that the barometer will be 0.112 inches under its mean height, 

 and as sin (t+0) cannot be greater than + or - i, we know that it 

 is then at a minimum. In the middle of March when t = 90° 

 sin (t-2 7o°) will be zero, so that the barometer is then at its mean 

 height. If, however, 9 is not 270° but say 255° it would mean that 

 the minimum does not take place when t = o°, but when it is=i5° 

 or about 15 days after the middle of January. 



Table i gives an analysis of the annual variation for a large 

 number of places in and near South Africa. The figures for the 

 heights above sea-level and the mean-barometer are to be considered 

 approximate. An inspection of the table shows that the first term 

 having a period of a year is by far the most important. A very natural 

 question in regard to such an analysis as this, is what are the value 

 of the figures, to what extent are they uncertain owing to fluctuations 

 from the mean, etc. ? To answer this question, the analysis has been 

 made for 6 decades for the Cape Observatory, — in this case the 

 extreme divergencies are : — of maximum phase 3 days, of 

 amplitude 3/ioooths of an inch. The differences amongst the second 

 and third terms are larger, but the substantial reality of these terms 

 cannot be doubted. 



