Practice curves and errors of measurement in estimating underwater weight by hydrostatic weighing

Abstract

The practice curves and errors of measurement in estimating body volume by the underwater weighing method were determined in 135 men and women during 10 consecutive measurements. The practice curves were of an exponential nature, the weight increase being 7.2% in the men and 10.3% in the women during the 10 trials. For the entire sample the last trial brought the underwater weight to within .014 density units of the asymptote. However, in approximately 40% of the subjects, 10 trials was insufficient for the upward trend to level off. In this situation more practice trials must be given in order to get the underwater weight to a point where it approaches the individual's asymptote. There was also a large and progressive decrease in the intra-variability — 93% in the men and 63% in the women. Little change occurred after the seventh trial. Selecting the heaviest weight in 10 trials (even though it occurs early or late during practice) may be invalid because it simply may reflect a statistical fluctuation in the random instrumental or observational errors, or the within-individual variation, Estimating the weight from the trend line based on a series of measurements averages out the variable errors, and is therefore more dependable.