Abstract
In some randomized cloud seeding experiments a single-area design was used. The results of seeding were estimated from the ratio S/U (S and U being the average rainfall amounts, respectively, for the time units seeded and unseeded in accordance with some random scheme). It is pointed out in this paper that the ratio S/U can be looked upon as the “result” of a two-area crossover experiment where the “rain” amounts of the “second” area are unity for all time units. This approach makes it possible to use results developed by Gabriel and Feder for the root-double-ratio pertinent to two-area crossover projects. It then follows that while the effect of seeding is estimated as (S/U)−1, the statistical significance level of that result is estimated from a curve representing (S/U)½. A consequence of this is that the significance level is very much lower than is the case for crossover designs for any given value of indicated increase (or decrease). The theoretical results are examined in light of data obtained in the 1961–67 Israeli randomized cloud seeding experiment.
