Multiple regression analysis applied to ventilation records of stopes on the Witwatersrand showed that it was possible to estimate the temperature increase per 100 ft for a new stope with a standard deviation of 0' 5F. The analysis also threw light on the relative effect on this estimate of five different factors.Application of the analysis of variance technique to a factorialexperiment on the production of dust during rock drilling showed that the effects of air and water pressure were quite small but also showed that the between-machine variation was unexpectedly large.Analysis of variance also proved a valuable means for assessing the bias and random error variation of aneroid barometers used in pressure surveys.INTRODUCTIONThree examples are discussed to illustrate the use of the techniques of "analysis of variance" and "multiple regression" in research on the ventilation of gold mines. Although the techniques are basically similar their application to each example has a slightly different emphasis and the examples were chosen for this reason. The variation ofemphasis is as follows.1. Temperature increases in stapes. This work was undertaken as a preliminary to possible extensive controlled experimentation (Lambrechts, 1959). Data collected in routine work by the mine ventilation personnel were used for analysis and no specific experiments were carried out. The main object was to see whether such data could provide any strong pointers to factors which might be worthy ofmore careful study and to factors which could initially be ignored. Of course, there was hope that the study might provide an empirical formula for rough forecasts.