A note on terminology: what is 'normal'?
In this paper we shall be concerned with the swelling and shrinking of soil. Inevitably we shall come across the term 'normal', and this will cause difficulty. 'Normal' shrinkage refers to volume change that is equal to the change in water content. Mitchell (1992) suggested that the term should be abandoned because it is imprecise and confusing. We shall here be dealing with 1-dimensional measurements of height change and water content change. A soil might shrink 3-dimensionally in a clod, and reduce in volume by 1 cubic millimetre for every cubic millimetre of water lost. However, the soil may crack between the clods, and 1-dimensional measurements of subsidence at the surface and water content change might show that 1 millimetre is lost in surface elevation for every 3 millimetres depth of water lost. Thus, normal, 3-dimensional shrinkage may be interpreted as 1-dimensional, less-than-normal shrinkage. Theory suggests that normal, 3-dimensional shrinkage will lead to a ratio of height change to water lost of 1/3 (e,g. Bronswijk 1991; Mitchell 1991). One might conclude that the converse is also true: the 1-dimensional result indicates normal, 3-dimensional shrinkage. But, as McIntyre (1984) suggested, the 1-dimensional result might also be a composite of different soil layers, each with a different ratio of height change to water lost (see also Aitchison and Holmes 1953): we cannot tell which is the case. Since we are here dealing with 1-dimensional behaviour, explanations about 3-dimensional processes are in any case irrelevant.
We report 1-dimensional observations in terms of height change, water content change, and their ratio. We shall resist inferences of normality or otherwise, and dimensionality of the shrinkage. However, when discussing our results in relation to the literature, we shall report where other workers have discussed shrinkage in terms of normality and dimensionality, in order that we may properly connect with that literature (and to avoid inventing new terms and creating the attendant problems of translation).
Heavy clay soils can swell and shrink appreciably during wetting and drying. The swelling and shrinking complicates the estimation of the amount of water stored in the soil. This can in turn lead to incorrect estimates of terms in the water balance, in particular the drainage term, which is often found by difference from the other terms.
Swelling soils are considered to shrink and swell in structural, normal (or basic, the term preferred by Mitchell 1992), and residual phases (Fig. 1 (Stirk 1954). In the structural phase, change in water content occurs through filling or emptying of macropores or cracks and there is little volume change. In the normal phase, change in water content is accompanied by change in the volume of the soil between macropores and cracks, and results in height change in the soil profile. In the residual phase, the soil matrix is strong enough to resist further volume change, and any change in water content results in little volume change, and hence little height change. In the normal phase, if the volume change is 1-dimensional, the ratio between height change and water content change is 1, but if the volume change is 3-dimensional, and equal in all directions, the ratio between height change and water content change ([alpha]) is 0.33 (Bronswijk 1991).
[FIGURE 1 OMITTED]
Field studies of water content change and height change (or bulk density and water content change) have generally found that the ratio between them varies from 0 to 0.33, which has led to the conclusion that shrinking and swelling is often 3-dimensional, and either normal or less-than-normal. Furthermore, the ratio between height change and water content change can vary from one soil layer to another. Woodruff (1936), Aitchison and Holmes (1953), Bronswijk (1991), Cabidoche and Ozier-Lafontaine (1995), Baer and Anderson (1997), and Coquet (1998) all measured behaviour consistent with these broad conclusions using depth gauges anchored into the soil at different depths and measurements of soil water content. McIntyre et al (1982) also measured behaviour consistent with these broad conclusions, but using movement sensors based on a manometer principle (Barrow et al. 1975). Mitchell (1991) measured changes in elevation of plates on the surface of a moderately shrinking soil, and found in several treatments [alpha] to be <0.33. Berndt and Coughlan (1976) and Jayawardane and Greacen (1987) also found [alpha] <0.33 from measurements of bulk density and water content.
In contrast, Fox (1964) concluded that a heavy clay soil from Queensland exhibited 3-dimensional expansion when the gravimetric water content was <46%, and 1-dimensional expansion in wetter soil. Chan (1981) obtained a similar result for a subsoil with large structural units. Chan (1981), Berndt and Coughlan (1976), and Jayawardane and Greacen (1987) suggested that these results might have been an artefact of sampling, in particular sampling between the cracks. Mitchell (1991) found that in one treatment the ratio between surface elevation change and water content change was initially >0.33 and approached that of 1-dimensional …