This paper uses new micro-level US data to re-examine productivity leadership in cotton spinning c. 1900. We find that output aggregation problems make the Census unreliable in this industry, and that Lancashire, not New England was the productivity leader for almost every type of yarn. This is...