Caffeine sodio-benzoate, 200 mgm., was injected subcut. into the author before each of 17 tests of 15 2-move chess problems each under controlled conditions; 17 tests after saline injection were irregularly interspersed. From the 9% improvement after caffeine one can not safely conclude that it improves chess problem solving, though an unusually long solving time may be somewhat less frequent; this latter was more marked in a previous study upon "Fletcherizing." Correlation between efficiency and barometric changes was not found, and body-temp. did not rise to account for the caffeine improvement. It was confirmed that 17-21 pieces per problem cause maximum difficulty, that accuracy is higher with fewer pieces per problem (8-13), that the queen is preferred as the solving piece and that results with such problems are better.