The contribution of steam to British economic growth in the nineteenth century is estimated using growth accounting methods similar to those recently employed to examine the role of ICT. The results indicate that steam contributed little to growth before 1830 and had its peak impact about a hundred years after Watt's famous invention. Only with the advent of high‐pressure steam after 1850 did the technology realise its potential. Compared with ICT, steam's impact on the annual rate of growth was modest. It is unlikely that these conclusions are vulnerable to quantification of hitherto unmeasured TFP spillovers.

n the early 1980s we published revised estimates of aggregate economic I performance during the British industrial revolution which have stimulated reappraisal of the beginnings of modern economic growth.2 These new estimates have led most scholars to abandon the previous orthodoxy which had been based on the pioneering work of Deane and Cole.3 We have subsequently explored further aspects of the industrial revolution using our 1982/3 estimates of the rate of growth as acceptable best g~e s s e s .~ Recently, however, several papers published in this journal have, with varying degrees of hostility, expressed doubts about our estimate^.^ We think it is now opportune to respond to these comments. Our critics have questioned the appropriateness of our estimation techniques and have suggested that our overall view of the industrial revolution is misconceived. While we do not wish to dismiss the points that have been made, we feel both that many of the specific criticisms of our estimating techniques have been erroneous and that our overall view of the late eighteenth and early nineteenth centuries has been misunderstood.Initially in section I we present a brief summary of a coherent new view of British growth that emerged from our aggregate estimates. The bulk of the paper then takes up the technical issues that have recently been raised. In particular, we argue the following main points. First, that there is inevitably a range of possible estimates for industrial output growth given the inherent index number and data problems. Both Crafts's and Harley's original estimates fall within this range, as do Jackson's new estimates, whereas Deane and Cole's do not. Second, that we do not accept Jackson's case for preferring what he calls a Crafts rather than a Harley view of industrial output growth nor do we accept his dismissal of the indices of industrial output presented by Crafts, Leybourne, and Mills (CLM) and by Harley. Third, that Hoppit and Berg and Hudson substantially overstateWe are grateful to Maxine Berg, Brian Mitchell, Joel Mokyr, Patrick O'Brien, and an anonymous referee for helpful comments on earlier drafts. Any errors are, of course, our responsibility.

No abstract

No abstract