E. J. Smith’s research while affiliated with The Commonwealth Scientific and Industrial Research Organisation and other places

Isolated cumulus clouds with supercooled tops were seeded from an aircraft. Either a large (20 gm), small (0.2 gm) or zero quantity of silver iodide was used, with random choice of treatment. The amount of rain which fell from the cloud was measured at cloud base by means of an impactor on the seeding aircraft.Clouds with tops 10C or colder which were treated with the larger quantity of silver iodide yielded significantly more rain than similar, untreated clouds. The mean rainfall from clouds seeded with the reduced treatment was also higher than that from the unseeded clouds, but the margin in this case was insufficient to demonstrate the statistical significance of the result.

The performance of a silver iodide smoke generator, mounted on an aircraft, was measured in flight. The ice-nucleus output was 1014 per gram of silver iodide active at −15C and 1012 at −10C. Considerable variations in the burner configuration and the solution which was burnt had little effect on the performance.

From 1958 to 1963 an experiment was conducted in the New England region of Australia, in which clouds were seeded with silver iodide smoke released from an aircraft. Clouds over two areas were seeded, with random choice of area. Rainfall measurements in the two areas suggest that seeding increased the rainfall during the first year, but no net changes in rainfall could be detected in subsequent years. The seeding appears to have increased the variability of rainfall.

A cloud-seeding experiment was conducted in the Snowy Mountains of Australia from 1955-1959 inclusive. The objective was to determine if silver-iodide smoke released from an aircraft into clouds could increase the precipitation over a selected area. The method involved a comparison of the precipitation in a target area and that in a control area during randomized periods of seeding and no seeding. Over the five years, the ratio of the precipitation in the target to that in the control area was higher in seeded than in unseeded periods. Three statistical tests are presented which show that the seeded periods are different from the unseeded periods at significance levels of 0.03, 0.09 and 0.03 (one sided). This supports a positive seeding effect. Other analyses both detract from and support this contention. The net result is that the experiment in inconclusive. Further, improved experiments are proposed.