Estimates of the size of populations of the red locust, Nomadacris septemfasciata (Serv.), in their outbreak areas have been made in Tanganyika from counts of adults caused to fly up by the approach of a slowly moving vehicle. For research purposes it was desirable both to improve and estimate the accuracy of such assessments. For control purposes it was desirable to devise an objective aerial assessment technique which could be used in areas where ground methods were either impracticable or expensive. The errors involved in both ground and air estimates can be divided into those concerned with the ratio of locusts counted in the sample area to those actually present, and those concerned with the distribution of the locusts.

It was found that a Land-Rover moving at five miles an hour, with the wind blowing from a rear quarter, flushed a high and virtually constant proportion of locusts during most of the day and in the most important grasses of the Rukwa outbreak area, throughout the dry season. Variation in locust density and in the state of the grass had little effect, if any, on the value of the equivalent strip (e), i.e., a strip whose width is such that the number of locusts in it is equal to the number flushed by the vehicle in a strip equal in width to its own. The change with time, and thus presumably with locust behaviour, in the value of e, estimated as recommended, was small but significant.

Counts by various observers in one vehicle gave similar results and it seems likely that, even with high numbers, there was no systematic mis-estimation. This is almost certainly true for counts of under a hundred locusts flushed per 0·1 mile. With upwards of one thousand 0·1-mile counts in each assessment, it is believed that the random errors due to mis-counting of the locusts flushed will be negligible.

The locusts are always distributed in an over-dispersed manner, consequently normal techniques for ascertaining the standard error of estimate cannot be used. Three assessments were made by the methods recommended, using a grid of two-mile squares; in one case, where the distribution was very gregarious, the most densely populated sector was more intensively sampled. The standard error of assessments laid out in this way was estimated. The method of analysis overestimates the standard error to some degree. Alternatively, the standard error resulting from the nature of the locust distribution can be computed for assessments where the 0·1-mile counts can be shown to be distributed log-normally.

Provided the population assessment is carried out under the conditions stipulated above, the additional errors will be minor. The effect of such errors will be likely to be so small that the estimated standard error, derived from the lognormal or assessment-grid analysis, may be regarded as a valid estimate of the standard error of the population estimate.

An attempt was made to estimate populations of the red locust in part of an outbreak area from locusts flushed by a ‘ Prospector ’ aircraft. The trials suggested that, given a clear rear view, red locusts which were flushed could be identified as such. Further, the number of locusts flushed could be counted at low densities, and it is thought likely that a reasonably objective technique could be developed for estimation at higher densities.

Flushing appeared to be most effective at an aircraft height of 20 ft. or less and when the direction of flight was with the wind. However, the majority of locusts present failed to take off and the proportion which did rise varied very greatly. The spraying of concentrated ammonia under pressure increased the efficiency of flushing.

A large-scale comparison of air and ground assessments suggested that, under the prevailing conditions, a considerable proportion of the locusts could be flushed only from the dense concentrations. At all densities the proportion of locusts which the aircraft caused to rise was a small fraction of those present. With flushing of such a low level of efficiency, little idea could be obtained of the size of the locust population at a moderate or low density.