In this article time series on labour productivity in the manufacturing sector are the starting point of the analysis. Using a growth accounting framework (Cobb-Douglas production function) we try to assess to what extent increases in labour productivity are due to rises of total factor productivity or to a growth of capital intensity. The results of this accounting exercise seem to contradict the views expressed by a great number of historians who have written on this subject. By relaxing some of the strong assumptions regarding perfect competition, a new view on the determinants of long-term industrial growth can be developed. The increase in the capital/labour ratio — which undoubtedly has been the engine in the process of industrial growth — cannot be explained exclusively by changes in relative factor prices. Also changes in institutional arrangements as well as demand shocks (resulting in economies of scale) play an important role. By taking imperfect competition (and non-equilibrium prices) into consideration, the puzzling results of the Cobb-Douglas accounting experiment can be reconciled with traditional literature on this subject.