An experiment with two blocks containing phased sequences of continuous spring barley after beans or fallow was located on sandy soil over Lower Greensand on a gentle north to south (N–S) slope at Woburn Farm. Season had the greatest effect on yield with a 135% difference between the worst (1975, 1·73 t/ha) and the best (1974, 4·06 t/ha). years. N–S position was the next most important factor with average differences of 65 and 52% between the plots at the top and bottom of the site in blocks I and II respectively. The third most important factor was E–W position which gave a maximum difference of 35% in 1975.
A fertility trend with a strong linear component, which was most conspicuous in drier years, followed the main slope of the experiment and was attributed to erosion (fieldwash). After 1972 as different cropping sequences were progressively introduced, yield variation due to crop sequence differences was confounded with this positional effect.
Crops in the eastern block were taller by l·5–12·3% and, after adjustment for the linear trend, yield was on average 15·6% greater than in the western block. The site is astride a N–S soil boundary with Stackyard series to the east and Cottenham series to the west. The Stackyard soil has a greater available water capacity, and is subject to drought less frequently than the Cottenham soil. Using Penman's (1971) data for the Cottenham series at Woburn and estimates of profile available water for the two series elsewhere on the farm, theoretical yields were derived, which were generally greater than actual yields adjusted for the N–S linear trend (block means 1·47–4·32 t/ha), but which showed similar trends in the between-block differences. Explanations for discrepancies between theoretical and actual yields are discussed. The incidence and severity of take-all disease and differences in soil pH were always small and unlikely to have caused significant yield variations.