Reply by the author to G. D. Garland.
Although the Tokaanu experiment does contradict the proposal that the gravitational constant G increases with scale, the result is not significant. The standard error in the least‐squares adjustment is at least 1 percent, which exceeds the predicted variation in G. The uncertainty in mean density is nearer 5 percent. Gravity data with sufficient precision to test for a scale effect in G are obtainable; the main problem appears to be the uncertainty in density determinations. Stacey et al (1981) made a least‐squares determination of G using gravity and density measurements from a mine. However, the pattern of residuals obtained indicated the presence of anomalous masses not adequately accounted for by their density averaging. The method I have used which models the spatial variation in density offers the possibility of obtaining a least‐squares fit for G with a satisfactory residual distribution. However, the problem of the effect on bulk density of joints and voids not sampled in hand specimens remains.