On the basis of detailed rill surveys carried out on bare plots of different lengths at slopes of 12 per cent, basic rill parameters were derived. Rill width and maximum depth increased with plot length, whereas rill amount and cross-sectional area, expressed per unit length, remained similar. On smaller plots, all rills were connected in a continuous transport system reaching the plot outlet, whilst on larger plots (10 and 20 m long) part of the rills ended with a deposition areas inside the plots. Amounts of erosion, calculated from rill volume and soil bulk density, were compared with soil loss measured at the plot outlets. On plots 10 and 20 m long, erosion estimated from volume of all rills was larger than measured soil loss. The latter was larger than erosion estimated from volume of contributing rills. To identify contributing soil loss area on these plots, two methods were applied: (i) ratio of total soil loss to maximum soil loss per unit area, and (ii) partition of plot area according to the ratio of contributing to total rill volume. Both methods resulted in similar areas of 21.8-23.5 m(2) for the plot 10 m long and 31.2 m(2) for the plot 20 m long. Identification of contributing areas enabled rill (5.9 kg m(-2)) and interrill (2.6 kg m(-2)) erosion rate to be calculated, the latter being very close to the value predicted from the Universal Soil Loss Equation. Although rill and interrill rates seemed to be similar on all plots, their ratio increased slightly with plot length. Application of this ratio to compute slope length factor of the Revised Universal Soil Loss Equation resulted in similar values to those predicted with the model. The achieved balance of soil loss suggested that all the sediment measured at the plot outlet originated from contributing rills and associated contributing rill areas. The results confirmed the utility of different plot lengths as a research too[ for analysing the dynamic response of soil to rainfall-runoff. Copyright (C) 2005 John Wiley Sons, Ltd.
