106 Transactions of the Royal Society of South Africa. 
cases. Moreover, there is this important fact to be borne in mind : that 
whereas the monthly range, or the range of greatest disturbance, is almost 
independent of the number of atmospheric depressions and anticyclones 
which pass over any station, an abnormal number of such depressions 
or anticyclones must necessarily raise the value of the monthly and 
annual mean barometric variability accordingly. That is to say, the 
variability gives the better measure of cyclonic activity. 
In Table 1 will be found the method of computation for a specimen 
month, showing — 
1. The atmospheric pressure at 8 a.m. at Kenilworth (Kimberley) for 
each day of October, 1904 ; 
2. The differences of pressure between one day and the next, with the 
average of them all, that is, the monthly mean variability, equal to 
•102 inch in this case ; 
3. The range from the greatest pressure to the least, equal to 
•374 inch; and 
4. The average range from highs to lows reckoned on the two most 
conspicuous instances in the month, equal to "312 inch. 
If it had been possible, it would have been well to have made 20 
years the shortest period of observations which should entitle a station to 
have its constants of barometric variability determined and put on record 
here. For the monthly means of variability vary very much both from 
month to month, and for any one month from year to year. And so when 
the period is short the incidence of a month of exceptionally large or 
small variability may teud to distort the annual curve. To insist in every 
case, however, upon a long period would have shut out several stations 
whose geographical positions have a special importance. The results 
given, therefore, are to be regarded as provisional until better can be 
obtained for longer periods. Nevertheless it may be said that a com- 
parison of all the annual curves shows that those derived from the shorter 
series of observations are probably very much better than might have 
been expected a priori. 
Some stations of the first importance have had to be omitted. Not 
because their available records cover too short a period, but because they 
have been published in such a form as to be of little utility. Mauritius, 
for example, does not publish daily readings at a given hour, but only the 
average of the hourly readings for each day. Again, all the early results 
published in the Argentine Anales de la Oficina Meteorologica are shown in 
the same way. This circumstance unfortunately precludes their use for 
our special purpose. And for this reason, that the monthly mean 
barometric variability computed from the readings at a given hour 
differs considerably from the variability computed from daily means. 
Table 2 gives a comparison of the two sorts of variability at Cordoba. It 
