An approximately spherical particle of coal (diam. 13-14 mm) was made by filing a larger piece. Next a hole (diam. 0.6 mm) was drilled through the centre of the sphere to end at 3 mm from the opposite face. Into this hole a fine thermocouple (o.d. 0.2 mm) was cemented. The coal was immersed into an electrically heated bed of sand, fluidised by nitrogen at 850 degrees C. During the subsequent pyrolysis, the temperature was measured at 3 mm inside the coal. Minima were found in the local internal late of increase in temperature, when plotted against either time or this measured internal temperature. Here is new evidence for large coals thermally decomposing by a sequence of "endothermic waves' moving radially inwards through the coal to release volatile matter. These minima are each associated with a particular temperature and 21 of them were found (apart from that for drying) between 160 and 820 degrees C. Altogether four bituminous coals and one lignite were studied. Some "decomposition temperatures" were common to them all. Lower rank coals have major losses of volatile material at lower temperatures. These experiments thus support the modelling of pyrolysis using a suite of approximately 20 parallel reactions, each contributing to different extents and with different kinetic parameters and Delta H. Experiments in an identical bed, but fluidised by air, revealed endothermic waves in a coal. In this case, some of the volatiles burned in a counter-flow flame in the recently discovered cushion of air underneath such a relatively large coal particle. Also, towards the end of devolatilisation, the resulting char started to burn; different chars burned at different temperatures, all above that of the bed, even though burning was controlled by external mass transfer of O(2) to the char particle. It appears that underneath a char particle, there is again a counter-flow flame, where CO (from burning the char) is oxidised by OH radicals to CO(2) at similar to 900 degrees C. Most probably the carbon in the char is also oxidised by OH radicals to CO. The temperature at which a char burns is partly controlled by how much CO is oxidised by OH radicals close to the underside of the char particle. The oxidation of both CO and carbon in a char accordingly appears to be catalysed by hydrogen from the char. (C) 2011 The Combustion Institute. Published by Elsevier Inc. All rights reserved.