The Coefficient of linear extensibility (COLE(std)) is often used by s
oil survey agencies to characterize the shrink-swell behaviour of soil
. Undisturbed clods or cores are required but these are rarely collect
ed during routine survey in Australia. The standard linear shrinkage t
est (LS(std)) uses sieved soil and is often used as an alternative. Ho
wever, the natural soil fabric is destroyed and the results can be dif
ficult to relate to field behaviour. We have developed a modified line
ar shrinkage test (LS(mod)) in which disruption to the natural soil fa
bric is minimized. The LS(mod) was found to be a good predictor of COL
E(std) (r2 = 0-88) with the slope of the regression line close to unit
y. LS.td was a weaker predictor of COLE(std). LS(mod) and LS(std) were
more precise measures than COLE(std). The modified method has been te
sted on a limited range of soils but a more comprehensive evaluation i
s required. This can be performed wherever COLE(std) measurements have
been made and sieved material is still available.