Described in this thesis are the results of an experimental and theoretical study of the effects of turbulence on combustion in single and dual chamber spark ignition engines. The techniques adopted in the experimental study included the use of high speed cine photography, and the collection of simultaneous cylinder pressure records using an on-line computer. The experimental results confirmed the potential of the dual chamber design for increasing burning rate, and for controlling the level of turbulence within an engine cylinder. High speed photographs were filmed through a perspex window in the engine cylinder head. These showed that flame propagation was much faster when the engine was fitted with a divided chamber cylinder head than when equipped with a disc shaped single chamber head. The acceleration of combustion rate has been shown to be a function of flow velocity through the interconnecting orifice during the compression stroke. At very high flow velocities the nozzle became choked, and engine performance was impaired. In the theoretical work, a computer model for the thermodynamic cycle of an engine was developed. The use in this model of empirical laws to describe combustion rate was shown to be inadequate; this was primarily because of uncertainty in the length of the combustion period, which one needs to specify when using this method. When burning velocity data (derived from work by colleagues using a turbulent combustion bomb) were incorporated into the model, good qualitative results were possible. The use of an empirical law to describe the effect of turbulence on the burning velocity of a developing flame was, however, shown to be inaccurate. The turbulent flame front in an engine is a thick reaction zone containing pockets of unbumt charge. Analysis o f data for flame projected area (derived from high speed photographs) and simultaneous cylinder pressure data, revealed that a considerable quantity of unburnt charge was present behind the visible flame front. There was some evidence that a greater proportion of unburnt charge was present behind the flame when the mixture was lean than when it was stoichiometric. Modelling of this effect by assuming that mass, once entrained, would burn at an exponential rate, was shown to produce reasonable results.