This dissertation concerns solid spheres, with diameters ~ 6 - 10 mm, densities between ~700 - 1500 kg m-3 and emitting gas at various peripheral velocities, Ud, and their tendency to float or sink when introduced into gas-fluidised beds of Geldart Group B particles. This is relevant, for example, to the fluidised bed combustion of biomass, and the apparent tendency of the fuel to devolatilise predominantly near the upper surface of the bed with the attendant undesirable complications of unconverted volatile matter (VM) entering the freeboard. Inert spheres (viz. where Ud = 0) in a bubbling fluidised bed can sink, even if less dense than the fluidised medium, owing to the additional weight of bed particles which tend to settle on top of them forming a defluidised hood. A 2-D fluidised bed, at room temperature, was used to investigate the structure of the fluidised bed in the vicinity of a cylinder emitting gas, as a mimic of a 3-D system. It was found that if Ud is more than 0.7, then the gas emitted can fluidise the bed particles in the entire defluidised hood. Consequently, it was inferred for a 3-D system that gas emitting spheres are not burdened by a defluidised hood and will rise to the surface more rapidly than inert spheres, which are burdened. The hypothesis that a gas-emitting sphere forms a pocket of high pressure around its underside sufficient to enable it to hover above the surface of the fluidised bed, was investigated, in a mechanism akin to the Leidenfrost effect exhibited by liquid drops on a hot plate. Experimentation showed that this hypothesis could be rejected. In fact, by observing the structure of the bed and measuring the pressure around a gas-emitting cylinder close to the surface of a 2-D fluidised bed, it was found that the emission of gas from a freely-floating sphere decreases the net upthrust of the bed on its underside thereby causing the sphere to sink lower into the bed than buoyancy alone would suggest. However, it was also discovered that the emission of gas from a sphere sunk deep within a fluidised bed caused the net upthrust from the bed to increase, causing the sphere to rise more rapidly to the surface than an inert sphere. This suggests that there exists a stable depth at which gas-emitting spheres reach dynamic equilibrium just beneath the surface of the bed where the bed’s upthrust matches the weight of the sphere. An interesting aside of investigating the Leidenfrost mechanism was that, as far as Geldart Group B solids are concerned, experiments showed the two-phase theory of fluidisation holds exactly. To simulate spheres of devolatilising biomass, spheres of dry ice, sublimating in a hot fluidised bed were used, because dry ice emits a single, readily detectable gas. The spheres of dry ice were, however, much denser than any biomass fuel and so only segregated once the rate of sublimation was very high. The external heat transfer coefficient for the spheres of dry ice was measured at a variety of bed temperatures and bed particle sizes. Unlike inert particles, gas emitted by the dry ice particles caused the heat transfer coefficient to a) decrease as the bed material size was decreased and b) decrease as the bed temperature increased. For the first time, a heat transfer model, which accounted for the change in structure of the bed material near the gas-emitting particle, was developed to predict the rate of gas emitted from the dry ice particles and gave good agreement with the experimental results. A novel method for finding the peripheral velocity of VM, emitted by spheres of biomass during devolatilisation in a fluidised bed, was developed and validated experimentally. The mean molar mass and composition of the VM was measured, with the result that measuring the concentration of the combustion products of the VM alone could be used to find the molar flowrate of the VM. Using this method, values of Ud, for spheres of beech, devolatilising in a hot fluidised bed, were measured and, simultaneously, the depth of the spheres in the bed was determined using X-radiography. The simultaneous measurements of gas velocity and depth allowed the behaviour of freely floating, devolatilising spheres to be compared with the calculations obtained with the 2-D fluidised bed. The spheres of beech remained just beneath the surface of the bed throughout devolatilisation and were less influenced by the mixing motions of the bed than inert spheres, even when the fluidisation velocity was increased. The devolatilising beech behaved much as anticipated by the results of the 2-D bed experiments. Tentatively, a dimensionless plot was made which, brings the variables Ud, the incipient fluidisation velocity Umf, the densities of the gas-emitting particles and the bed material, and the depth at which a particle will neither rise nor sink in the bed, together. The plot shows under what conditions a gas-emitting particle is likely to have a sinking or rising tendency in a fluidised bed. The plot is a tool for predicting if segregation of a particular fuel particle is likely to occur in any bubbling fluidised bed. Overall, this dissertation concludes that the emission of VM from a devolatilising particle of biomass not only draws the particle to the surface of the bed but acts to keep it there, even at low rates of gas emission. To eliminate the segregation of biomass during combustion in a bubbling fluidised bed, the biomass must be denser than the emulsion phase of the fluidised bed and the velocity of VM leaving the biomass must be as low as possible. An impracticable degree of pre-processing of the biomass would be required to achieve these conditions.