Soil samples collected from four of the Rothamsted long-term field experiments over the last 100 yr were used to follow the effects of management on soil organic matter content. The experimental sites were: Broadbalk and Geescroft Wildernesses, both on old arable land that had been fenced off in the early 1880s and allowed to revert naturally to deciduous woodland; the unmanured plot in the Rothamsted Park Grass Continuous Hay Experiment, started in 1856; the unmanured and one of the NPK plots in the Broadbalk Continuous Wheat Experiment started in 1843. Total C, radiocarbon and (in some cases) soil microbial biomass C were measured in stored and contemporary soil samples. The current Rothamsted model for the turnover of organic C in soil was then used to calculate how the organic C content of the topsoils from the four sites would change with time for a range of annual inputs. The inputs that generated the best fit to the measured values were: Broadbalk Wilderness 3.5 t C ha-1 yr-1; Geescroft Wilderness 2.5 t C ha-1 yr-1; unmanured plot on Park Grass 3.0 t C ha-1 yr-1; unmanured plot on Broadbalk Wheat 1.3 t C ha-1 yr-1; NPK plot on Broadbalk Wheat 1.7 t C ha-1 yr-1. The model also generated the radiocarbon content of soil organic C for these inputs of C, knowing the C-14 content of the atmosphere over the period 1840-1985. The marked increase in the C-14 content of soil organic C brought about by thermonuclear testing during the 1960's was accurately represented by the model. The quantities of soil microbial biomass (measured by fumigation-incubation) in the top 23 cm of soil from the four sites were: Broadbalk Wilderness, 1.57 t C ha-1; Geescroft Wilderness, 0.58 t C ha-1; unmanured plot on Park Grass, 1.62 t C ha-1; unmanured plot on Broadbalk Wheat, 0.47 t C ha-1; NPK plot on Broadbalk Wheat, 0.76 t C ha-1. The values for soil microbial biomass C generated by the model for the above annual inputs agreed closely (with one exception) with the measured values. For a site under steady-state conditions, the annual input of organic matter to the soil plus the annual removal of organic matter from the site (if any) gives the Net Primary Production (NPP). NPP thus calculated was 4.0 t C ha-1 yr-1 for the unmanured plot on Park Grass, 2.2 for the unmanured plot on Broadbalk and 5.2 for the NPK plot on Broadbalk. The two Wilderness sites are still accumulating C in both soil and vegetation and here NPP is given by annual input to the soil, plus the annual increment of C in the trees. NPP calculated in this way was 4.8 t C ha-1 yr-1 for Broadbalk Wilderness and 3.3 for Geescroft Wilderness. This new way of estimating NPP, from measurements made on soil organic matter, needs to be tested on a wider range of soils, climates and vegetation types before it can be generally recommended. However, it has many potential advantages, not least that it can give a value for NPP that is integrated over many years from a single sampling. For sites under steady-state conditions it is not essential to have stored soil samples-the necessary measurements can be made on contemporary samples alone.
