b'Figure 1: Ballycanew grassland plot trial experimental designEach plot was initially analysed for bioavailable P (Morgan, 1941) to establish the baselinenutrient availability prior to application of treatments. Subsequently, at each site, plots were paired according to their closest soil P levels (14 pairs per site). From each pair, one plot was chosen to receive P in chemical form (Triple Superphosphate) and the other one in organic form (cattle slurry).Thereafter, 6 plots with low soil P levels (Morgan, 1941) per site were selected (3 receivingchemical P and 3 receiving cattle slurry), as well as 6 plots with high P soil P levels (3 receiving chemical P and 3 receiving cattle slurry). The selected plots were each equipped with 4 soil pore water probes (Figure 2). The probes were inserted below the first soil horizon (approximately 30 cm) directly under each plot. A tube protected by a sewer pipe made the link between the probe and the surface to maintain easy access on the side of the plots. Control probes were placedat 3 different positions around the area of the plots.Grass samples were harvested on a monthly basisfrom end of February to November each year.Herbage was harvested across the entire width ofeach plot with a ride-on mower (Etesia Hydro 124)to a height of 4 cm and was weighed on site using a collection box and a field balance. Total weight was recorded on site and a sub-sample was taken andreturned to the lab for analysis. Dry matter (DM %) of the grass was determined by drying and weighinga sub-sample (approx. 100g) at 70C for one week. Total herbage yield (kg/DM/ha) was calculated bymultiplying fresh yield by DM %. The dry sample was then ground and analysed for total N, P, K and Scontent.Figure 2: Soil pore water samplers at thepoorly-drained grassland site. Grass harvests88'