INLAND WATER SURVEY 381 
of action by frost. Experience shows, however, that if daily readings are 
available from a number of gauges in a given area, the monthly total measured 
in a mountain gauge can be apportioned with fair accuracy among the wet 
periods occurring in a given month. The loss of precision due to the 
absence of daily readings from particular localities would therefore not 
be very serious. 
Given a sufficient number of stations with daily readings, supplemented 
by monthly totals at stations inaccessible for daily readings, the evaluation 
of general rainfall for rain periods presents no serious difficulty. ‘The most 
accurate method of making such evaluations is to draw large-scale rainfall 
maps, measure the areas with falls between given limits, and thus arrive 
at the space-average or general fall. A quicker and only slightly less 
accurate method is to make a selection of stations, well spaced within the 
area, and determine by a preliminary investigation the relation between 
the arithmetical average of the falls of these stations and the general 
rainfall determined from maps. When this has been done the selected 
stations can be used for the routine evaluations and the maps can be dis- 
pensed with.!* The number of stations need not be very great provided 
they are well distributed and the gauges are well exposed. It is found, 
for instance, that the arithmetical average of the falls measured at twenty- 
four selected stations in the Thames Basin gives a very close approximation 
to the monthly general rainfall deduced by a planimetric method from a 
rainfall map based on readings from over two hundred stations in the basin 
and its vicinity. Six stations suffice to yield an equally close approximation 
in the Lea Basin. When dealing with shorter periods than a month rather 
more gauges would be desirable in order to ensure that localised heavy 
rains were not missed, but it would appear unnecessary, in routine work 
on run-off problems, to deal with daily readings from more than a dozen 
or so gauges in each catchment area, the number depending of course on 
the size of the catchment. A close preliminary study of the rainfall in 
the area would, however, be necessary before the gauges whose records 
were suitable for such routine use could be selected. Such a survey would 
necessitate reference to the records from many more gauges than would ° 
suffice for routine evaluations of general rainfall, and it should be made 
clear that the British Rainfall Organization would always regard it as essential 
to collect records from all available gauges for the purpose of rainfall study 
in general. 
Returning to the question of the time interval, my considered view is 
that while the study of rain periods may possess substantial advantages 
for dealing in detail with the run-off data for single catchments (such as 
might form the subjects of special memoirs), it would be necessary to adhere 
to definite calendar periods such as months, or groups of months, for the 
purpose of any regular periodical publication containing results from all 
the catchments participating in the scheme. 
In regard to the use of recording gauges, I do not regard it as necessary 
to have many in each catchment area. For detailed study of particular 
falls, especially falls associated with floods, the records from one or more 
continuously recording gauges with daily charts would undoubtedly prove 
useful if not essential. Any such gauge should be in the charge of a com- 
petent employee, who could be relied upon to give it the necessary attention, 
*and it would be necessary, in each case, to take daily readings for check 
purposes from an ordinary 5-in. gauge close to the recording gauge. 
18 The routine evaluations could, if necessary, be revised at a later date by the 
British Rainfall Organization, using the cartographic method. 
