An experimental study into the behaviour of premixed turbulent flames of hydrocarbon/air and hydrocarbon/oxygen mixtures, stabilized on burners, is reported. The state of the subject which led to the work is reviewed and the limitations of earlier investigations are discussed. New optical methods are introduced to obtain relevant information. Particle tracks are combined with instantaneous schlieren contours to give various, but reproducible, definitions of local and average burning velocity. Ray deflection methods are developed to assess the comparative effectiveness of turbulence inducing devices and for measuring the time average distribution of products and reactants. An electrical method for determining flame area, based on extracting all the ions generated in the flame and measuring the resultant current, is developed and tested. It is shown to be particularly suited to irregularly shaped and fluctuating flames where more usual methods of area measurement are difficult to apply. The "wrinkled laminar" concept of turbulent flame propagation is checked by appropriate burning velocity results, and independently, by the above electrical method. Both tests show the model to fail, the failure being greater than can be explained by small scale turbulence in the burner flow. The movement of the turbulent flame front is shown to be significant, compared with its thickness, during the transit of a pocket of gas. Part of this fluctuating motion is caused by the continuity of the perturbed flame front and part is explained by the interaction of flame generated momentum with the confined product flow. The turbulence 'seen' by the moving flame front accounts for the measured increase in burning velocity. It is shown that the electrical measurements can be related to those of burning velocity if changes in flame structure are assumed to occur.