Preliminary experiments having shown that dimefox (bis(dimethylamino) fluorophosphine oxide) placed in holes drilled at ground-level in cacao trees gave encouraging results in controlling the most important of the mealybugs that transmit swollen-shoot virus (Pseudococcus njalensis (Laing), Planococcus citri (Risso) and Ferrisiana virgata (Ckll.)), two further experiments to investigate the technique and determine the optimum dosage were performed in Ghana in April-June and September-November 1953, respectively. A randomised-block layout was used, with four replicates (each comprising a plot of seven trees) of each treatment at Tafo and, in the first experiment, another two at Bunso. The first experiment disclosed no significant difference in effect when the implantation holes were spaced four inches as against two inches between centres and the former spacing alone was used in the second experiment. The dosage levels of dimefox, based on the weight of the aerial part of the tree estimated from the girth—weight ratio, were 75, 150, 225 and 300 parts per million w/w in each experiment, the lowest representing about one-twelfth of the soil-application rate, and also 375 p.p.m. in the second experiment. The results were assessed in terms of the total number of mealybugs, of all three species, counted on the various parts of the trees, which were felled and examined six weeks after treatment; counts made on a duplicate set of trees in the second experiment that were felled at nine weeks were significantly greater.

Statistical analyses of the logarithmic transformations of the population figures obtained at six weeks showed the regression of population on dosage to be a curve of quadratic form. There was thus a limit to the response obtained, the minimum populations of total mealybugs (adults + nymphs) being achieved at a dose rate of 281 p.p.m. in the first experiment and 317 p.p.m. in the second. In the former, the minimum value of the average population of total mealybugs per tree was below that required for efficient control (one per tree), but the variability of the response was such that the population on 35 per cent, of trees so treated would have been above that level; in the latter, the minimum average value attainable was only slightly less than one, and consequently the population on nearly half the trees would have exceeded the level for effective control. Analysis of the combined results from those replications of each experiment that were sited at Tafo showed that a given dosage rate appeared only about half as effective, in terms of the residual population (expressed logarithmically), in October-November as in April-May, but concurrent records from untreated control trees suggest that this may partly have been because populations were falling naturally at the earlier date. Slight leaf-shedding occurred in some of the treated trees.