A facility has been developed at the University of Nottingham at which natural gas can be injected into soil to investigate the effects on the soil ecology and on the growth and development of plants. The facility involves 18 plots, 12 of which are equipped with a regulated and metered gas supply. The gas is released from a diffusive point source 1 m below the centre of each plot. Permanent pasture grass, wheat and bean were gassed from gas was injected into additional plots of each species to determine the effects of a new leak on a fully established crop. Spatial and temporal variability of soil gas concentrations were determined by two complementary methods. First, by daily extraction of soil gas samples from permanently buried sample pipes that come to diffusive equilibrium with the adjacent soil gas concentration. Second, by intermittent extraction of soil gas samples from bar holes inserted on a 50-cm grid. Contour plots of spatial variation were then constructed from the bar hole data.Methane (CH 4 ) concentrations near the centres of the plots were variable. They could reach values of up to 80% gas, but decreased rapidly away from the area of peak concentration. No relationship was found between gas flow rate or soil gas concentration and atmospheric pressure or other meteorological parameters. Soil structure affected gas dispersion, and an inverse relationship was found between CH 4 and O 2 concentrations.The grass showed visible symptoms within 44 days, whereas wheat and bean developed symptoms after four months. These symptoms included a circle of chlorosis in grass and poor development and leaf chlorosis for the wheat and bean. When the same species were gassed from mid-May; the wheat and bean showed no visible symptoms but the late-gassed grass also developed a further circle of yellowing.