•368 Transactions of the Boyal Society of South Africa. 
adhered to. But in areas not well supplied with gauges records of shorter 
period have had to be admitted, since it has seemed better to get an 
approximation to the average rainfall at such places than to ignore them 
altogether. Naturally the means given in the table are not all of the same 
value ; short-period records cannot be expected to give such good mean 
values as long-period records do, and all observers are not equally faithful. 
A common fault in every country is the omission to register small falls 
of rain, because, forsooth, the observer thinks them too contemptible to 
bother about. A scrutiny of the returned number of days of rain is 
sufficient in most cases to reveal this fault ; and when there are two or three 
gauges in one town the record containing the greatest number of rainy days 
is usually taken as the standard. When there is only one gauge, even when 
the number of days of rain reported seem to be rather too few, the record of 
quantity has been used for what it is worth. It is a pity that some observers 
do not realise that the trouble they take in measuring the heavy falls of rain 
is largely labour in vain, in a scientific sense, unless they take pains over the 
smaller falls. Generally speaking the longer records are the most 
trustworthy, especially when they are the work of one observer. This is 
because the observer who has enough intelligent energy to go on recording 
the rainfall year after year does not shirk the measurement of the smaller 
quantities. Nevertheless there is at least one long-period record in the Cape 
Province of little scientific value simply because only the heavy falls of rain 
have been regarded as worth bothering about. It cannot be insisted upon 
too often that the lighter rains must be as carefully looked after as the 
heavier ones. For one thing, if there be a secular change of climate 
anywhere it must proceed very slowly, and the facts can only be discovered 
by a uniform system of observation which does not neglect small things. 
The latitudes, longitudes and altitudes stand, as far as possible, for the 
latest determinations. In a few cases they have been measured off 
roughly from maps. The numbers under the heading " years " do not 
necessarily stand for an unbroken period, although they mostly do. 
Sometimes they represent two or more periods of observation alternating 
with longer or shorter intervals during which no records were kept. In such 
cases the number of years is really the quotient of the number of months 
divided by twelve. Thus at Lilyfontein (III A, 44) from 1885 to 1908 only 
ten full years of observation are known ; but there were further odds 
and ends running into fifty-eight months, making 178 months in all, which 
we call fifteen years. 
Occasionally a rain record has been started at one spot, and continued at 
another not far away. If the aspects and altitudes of the two places are 
not materially different such a record has been regarded as continuous. 
For instance a part of the Dumbiedykes rainfall was really observed at 
Geelbeks Yley a few miles off ; and the means for Sea Point have been made 
