The Statistical Design of an Experiment to Test the Stimulation of Rain

Abstract

The problem of analysing statistically the results of a C.S.I.R.O. rainfall experiment over the catchment area of the Kiewa river is considered. It is shown that in order to obtain a significant result with any reasonable supposed increase in the rainfall within a few years it is necessary to use a control variable strongly correlated with the test variate. The powers of various tests using either streamflow or rainfall as test variates are calculated when the control variable is either streamflow in the Murray or rainfall in other areas. The general problem of designing such tests and the relationship between rainfall and streamflow are also considered.