A method of stabilizing a pure acetylene decomposition flame on a burner, over a range of temperatures and pressures, has been developed. It depends for its feasibility and safety on a novel flow and compressor system, details of which are described. By its use, the variation of burning velocity with temperature and pressure has been determined not only for acetylene but also for a series of hydrocarbon + air mixtures, in order to test both apparatus and method of interpretation. The latter involves an attempt to deduce rate-controlling reaction steps from effective activation energies and orders derived from the pressure and temperature dependence of burning velocity. It is shown that, while these `global' parameters are in good agreement with values determined for the hydrocarbon + air mixtures, for instance, by `well-stirred reactor' methods, a logical modification in interpretation brings them into accord with the conclusions, concerning detailed mechanism, of recent flame-structure sampling studies. A similar analysis of the acetylene decomposition results yields effective orders and activation energies which differ from those of both low-temperature pyrolysis and shock-tube studies. A proposed reaction mechanism, which is likely to be dominant only in the case of flames and in the deliberately contrived absence of soot deposits, accounts for these differences as well as for the numerical values obtained. In the presence of soot, an alternative heterogeneous surface-reaction is likely to become important and experimental support for this concept is provided by the behaviour of acetylene during its interaction with hot-wire igniters above the burner.