Properly weighted (expenditure-share) measures of the money supply behave nothing like the official simple-sum accounting data and give 'fantastically different' inferences about testable hypotheses such as the future course of inflation and the relationship between money and the business cycle.