The author proposes to determine what features of spark-plug construction cause preignition and how this preignition manifests itself. To this end observed conditions on an Hispano-Suiza aviation engine following 4 hr. of an intended 6-hr. run are reported, with supplementary tests and observations. This resulted in experiments made to determine the cause of preignition, using spark-plugs constructed so that different features of their design were exaggerated. Illustrations of these plugs are shown and the results obtained from their tests are described. The different observed peculiarities are then stated, analyzed and compared with normal spark-plug performance. The experiments serve as a means of identification of special forms of preignition and as an indication of the abnormally high temperatures to which valves and combustion-chamber walls are thus subjected. They also serve as a basis for predicting the performance of a given design of spark-plug or for designing a spark-plug to meet known conditions.