Of the three primary particle sizes, clay is the most important factor of aggregate stability (Payne, 1988). Although aggregate stability generally increases with increasing clay content (Fauck, 1972; Pieri, 1989), clay rarely works alone. Interacting especially with organic or inorganic substances, clay will affect virtually all soil properties. Many of the important soil properties are directly related to the colloidal nature of clay (Hayes, 1990). For something so very small (< 2 m m), clay particles exert their dominance from their very large specific surface area, and their permanent negative charge.

Soils only need at least 20% clay to begin to express clayey properties such as cohesion, shrink-swell ability, decreasing permeability, and an aggregate structure in which clay particles can bind the coarser soil constituents and organic matter (Hayes, 1990). Likewise, Horn et al. (1994) remarked for soils to form aggregates, they need at least 15% clay.

However, the often neglected silt particles are often found to be quite important as well (Soong, 1977, 1978; Voronin and Sereda, 1976). Perhaps this is due to the substantial accumulation of organic matter in the silt fractions as discovered by Ahmed (1981), Ahmed and Oades (1984), Anderson and Paul (1984), and McGill and Paul (1976). Although these workers worked with silt-sized fractions rather than the primary silt particles, these fractions would have considerably amount of silt particles.

Working with Emerson's (1959) aggregate model, Mokhtaruddin and Norhayati (1995) hypothesized that certain amounts of very fine sand and silt particles are needed with clay to form and to stabilize aggregates. Mbagwu et al. (1993), however, discovered quite the opposite. Working with four different soil types (Alfisol, Entisol, Inceptisol and Ultisol), they discovered that soils with high amounts of silt and fine sand dispersed easily, while stable soils were generally related to the sum of coarse sand and clay. Similarly, Soong (1977) found that fine sand was negatively correlated with aggregate stability of nine Malaysian soils. However, such relationships may be fortuitous¾ relationships produced by chance due to other factors instead of particle size distribution affecting aggregate stability.

Perhaps the issue here is not the combination of which primary particles, but rather the compound particles of which size that is most important to aggregate stability. Dexter (1988) remarked that many aspects of soil behaviour are strongly influenced by these compound particles (as illustrated in Figure 2), rather than the primary particles of which they are composed.

Studies on clay and organic matter frequently overshadow the importance of cations. From the C-P-OM model (Edwards and Bremner, 1967), polyvalent cations are crucial to bridge clay and organic matter. Otherwise, clay and organic matter ought to repel each other because both carry negative surface charge (Hsu, 1989). This is probably why cations are related more to soil dispersibility than to slaking. Without this bridging, clay and organic matter complexes would repel each other, causing dispersion. Giovannini and Sequi (1978) hypothesize that these cations serve as junctions in a net or mesh composed of polymeric chains of organic matter. When these cations are removed, this mesh is weakened at the junctions; thus, stability decreases.

Of the four major cations (Ca, Mg, Na and K) in a soil, Ca and Na most consistently affects dispersion. Frequently observed, Ca promotes flocculation, but Na disperses the soil (Levy and Torrento, 1995; Little, 1989). However, Little (1989) remarked that if the concentration of Na is high enough, flocculation may instead occur. The effects of other cations, however, are less clear. While Mg is often associated with greater dispersibility (Emerson, 1984; Rengasamy et al., 1986), K is associated to good structure. For example, K-saturated soils have been found to have larger aggregates and greater aggregate stability than Ca-saturated soils (Cecconi et al., 1963; Ravina, 1973). However, the effects of Mg and K are variable. Conflicting results, especially relating to K (Levy and Torrento, 1995), conceal the behaviour of these two cations.

However, as a general rule, the dispersibility decreases in the following order (Dexter and Chan, 1991; Rengasamy et al., 1984): Na > K > Mg > Ca. This order indicates that for monovalents to cause dispersion is easier than of the divalents. This may be the reason Quirk and Schofield (1955) found soil dispersibility was related to the ratio of monovalent to divalent cations in the soil solution.

However, cations that disperse clay in water tend to indirectly produce stronger aggregates when the soil is dried (Shanmuganathan and Oades, 1982; Chan, 1989). Particles which repel in water remain free to adjust themselves as the soil dries and therefore pack into a higher density and strength (Gerard, 1965). On the other hand, particles which flocculate are immobilized, and are unable to reorientate themselves during drying.

Interestingly, the distribution of cations varies with the aggregates sizes. For example, in an Oxisol, greater amount of Ca was found in the smaller aggregate fractions, whereas more K was found in the larger aggregate fractions (Cruvinel et al., 1993). The larger amount of Ca in the smaller aggregates may be due to their higher adsorption surface area than the larger aggregates at a given bulk density. The higher concentration of K in the larger aggregates may indicate endogenous and exogenous absorption (Cruvinel et al., 1993).