Abstract
This paper considers unit root tests of the index of British industrial production 1700–1913. For the full sample the data are found to be I(1). However, three distinct phases are identified with alternating stationarity properties. One period, 1780–1851, is identified as the British Industrial Revolution. The results support the findings of Perron (1989) and Reichlin (1989) on the dangers of unit root testing in models with structural change.