A motored, unfired CFR engine was red with a rich heptane/air mixture. The jacket temperature was maintained at lOOcC with a few experiments at Oee. The ·exhaust vapours were totally condensed by a novel exhaust recovery system. After completely removing cylinder deposits. no reaction products were round at compression ratios up to 14: I. At a compression ratio or 16: I the charge autoignited/detonated. Subsequently, reaction products were round at a compression ratio or 6: I and thereafter in increasing quantities up to 14.6: I, the reproducible threshold of autoignition. This entire sequence or experiments was reproduced four times. Samples of the condensate were concentrated by vacuum distillation and subjected to qualitative and quantitative analyses. A very limited range of peroxides and carbonyls were formed in very low concentrations in all experiments where autoignition did not occur. These were identified and their rates of formation calculated. Some thermodynamic inferences were drawn as to their heats of formation. The maximum compression temperatures were calculated. At compression ratios below 16: I, it was concluded that the preflame reaction was initiated on the surface or the combustion chamber, after it had been subjected to autoignitions/detonations,