I apply the method of Caballero and Lyons to industry-level data for UK manufacturing. I find evidence for a positive external effect, arising from the expansion of manufacturing as a whole. But expansion at the sectoral level was found not to generate external effects. Nor was any evidence found for increasing returns which are internal to the industry: here returns appear to be constant. I argue that external effects at the level of aggregate manufacturing cannot be explained away as due to measurement error or cyclical effects. However, alternative explanations, such as economy-wide shocks, cannot be totally excluded.