A dry combustion (Dumas) reference method for measuring total organic C (T), is compared with two other methods: the wet acidified dichromate oxidation, Walkley-Black procedure (an estimator of organic C) and loss on ignition (L) (an estimator of organic matter) for 40 samples from various depths from five New Zealand mineral soils, spanning a range of textures and cultivation treatments. Regression analyses reveal the following:(1) an excellent linear relation exists for all samples between Walkley-Black procedure and reference method results, with a mean conversion factor 1.25, which compares with the original value 1.32 recommended by Walkley & Black (1934).(2) L and Tare less well-related by an equation of the form L = a+ bT. In contrast to the traditional assumption that b N 1.72 and a N 0, a strong 'intercept effect' (a > 0) is revealed,

No abstract

The X-ray attenuation (Sedigraph) method for particle-size analysis is known to consistently estimate a finer size distribution than the pipette method. The objectives of this study were to compare the two methods, and to explore the reasons for their divergence. The methods are compared using two data sets from measurements made independently in two New Zealand laboratories, on two different sets of New Zealand soils, covering a range of textures and parent materials. The Sedigraph method gave systematically greater mass percentages at the four measurement diameters (20, 10, 5 and 2 �m). For one data set, the difference between clay (<2 �m) percentages from the two methods is shown to be positively correlated (R2 = 0.625) with total iron content of the sample, for all but one of the soils. This supports a novel hypothesis that the typically greater concentration of Fe (a strong X-ray absorber) in smaller size fractions is the major factor causing the difference. Regression equations are presented for converting the Sedigraph data to their pipette equivalents.

Leaching in a salt-affected, permeable, sandy loam soil was evaluated under continuous and intermittent ponding conditions in 2 x 2 m (Sj), 4 x 4 m (S 2 ) and 6 x 6 m (S 3 ) plots. The soil contained large amounts of soluble salts throughout the profile to the water table, chiefly chlorides and sulphates of sodium, calcium and magnesium. The leaching curves did not differ significantly between S lf S a and S 3 plots under continuous ponding but did under intermittent ponding. The leaching efficiency decreased sharply with increased plot size. The leaching efficiency in Sj plots was significantly greater with intermittent than with continuous ponding, but the reverse was true in S 3 plots. The displacement of the resident soil solution in S! plots under intermittent ponding was nearly piston-like. With increased plot size, it tended to deviate from this behaviour. The leaching curves from S 3 plots (this size being reasonable in farmers' fields) were compared with those obtained from numerical solution of a simplified steady-state salt transport model. The model also included a source term, solubility rate constant, for the slightly soluble salts present in the experimental soil. The pore water velocity was estimated from field capacity and time-averaged infiltration rate. The effective dispersion coefficient and solubility rate constant were estimated by a least-squares minimization technique. A reasonably good agreement was obtained between simulated and experimental leaching curves. For practical purposes, this simple model may be adequate to predict leaching in salt-affected soils similar to the one under consideration.

The minimum number of parameters required to model the unsaturated soil moisture characteristic, relating volumetric water content (8) and matric suction (w), is shown to be two. A third parameter, 8= 8, at saturation, is required to define its saturation limit. The popular power-function ty/w,= (8/OS)* is the most general three-parameter model, with (u normalized by a notional air-entry potential, we. When log-transformed, e.g. as In w = a + bln(6/8,), it gives a good fit to observations over varying ranges of ty. We show, using US, Australian, UK and NZ data, that a,b and 8, in this formulation are uncorrelated across a wide range of textures, thus providing a 'basis set' of independent parameters. Gregson et al. (1987) used the alternative formulation In ty=a"+bln(1008), with 8 rescaled to a percentage. Their reported correlation between a" and b, which led to their 'one-