Steps are now being taken to define in more detail the phenomenology of coal liquefaction and to provide a scientific basis for empirical correlations previously established between liquefaction conversion and basic compositional characteristics of coals. The rates of production of oils, asphaltenes and preaphaltenes have been determined at four temperatures for three coals, two of Carboniferous and oneof Creaceousage. Productsareformed moreslowlyfrom theyoungercoal (which isof slightly lower rank) than from the others, but oxygen, partly as OH but probably mostly in a type of ether, is lost more rapidly. It is estimated that the maximum content of 0 as cleavable ether is 7.7 atoms/l 00 C atoms for the younger coal (from Wyoming) and 4.1 and 5.1 for the other two (from Oklahoma and Ohio, respectively). Until ~505/0 of the amount present in the Oklahoma coal is lost, the rates of removal of oxygen and organic sulphur are approximately equal; beyond this level, the removal of S is more rapid. The loss of organic sulphurfrom the Ohio coal is slightly faster. Even so, the data do not support the idea that cleavage of thioethers is more rapid than that of ethers and that this is the basic reason why a high organic sulphur content tends to promote liquefaction. Conversion of the pyrite in the Ohio coal to pyrrhotite occurs considerably more rapidly than the pyrite in the Oklahoma coal. In preliminary experiments, it is shown that a curve-resolving programme allows two aromatic and five aliphatic C-H stretching vibrations to be distinguished in FTIR spectra of the hexane-insoluble products, and the distribution changes with degree of conversion. In particular, there is evidence that new aryl methyl are generated during liquefaction, in agreement with evidence from oxidation studies.