Published online by Cambridge University Press: 10 July 2009
In an attempt at some assessment of the errors to which fly-round data obtained for the study of populations of Glossina are subject, three fly-rounds at Shinyanga, Tanganyika, were worked every weekday for four weeks. The species studied was G. swynnertoni Aust. One of the rounds was a ‘transect fly-round ’, the others, the older type, divided into a few long sections according to the vegetation. The catches showed no obvious trend during the experiment, suggesting that the true population had not altered in the period, but in all three rounds there were considerable day-to-day variations in catch and, in this respect, the transect fly-round did not differ from the older method. An analysis of variance of the transect fly-round data showed that the variance between days was greater than could be accounted for by random errors
By considering separately the data collected on certain days of the week, it is concluded that a 7,500-yd. fly-round done once a week could not detect less than a five-fold change in the mean catch. To detect a two-fold change, a 15,000-yd. fly-round done twice a week would be necessary.
The relation between fly-round catch and true density is discussed. This relation is usually termed ‘ availability ’. Earlier workers have shown that its mean value, over periods of about one month, is not necessarily constant; the present work shows that it varies in one place from day to day.