^{page 42 note (1)} American experience is reported at length in Business Cycle Indicators, edited by Geoffrey H. Moore, Princeton University Press, two volumes, 1961.

^{page 44 note (1)} The selection was made by using probability measures which are briefly described in the Appendix.

^{page 44 note (2)} The lead fell substantially in the case of the index of security prices, from 14.8 months in the pre-war period to 4.7 months in the post-war period.

^{page 44 note (3)} The tests applied are briefly discussed in the Appendix.

^{page 45 note (1)} This series is based on annual company reports. The more up-to-date figures of quarterly profits published by the CSO have not been running long enough to be subjected to the type of analysis used here. The same applies to a number of other series which have become available in the last few years.

^{page 45 note (2)} The roughly coinciding group of indicators was chosen in the same way as the group of leading indicators.

^{page 45 note (3)} For method of construction see Appendix.

^{page 45 note (4)} For this purpose all the series (leading, lagging and roughly coinciding) were taken.

^{page 46 note (1)} After making the best allowance possible for the effects of the very cold weather.

^{page 46 note (2)} For example, the changes in producers' stocks of steel may reflect still earlier changes in consumption of steel, or in orders received; but steel consumption figures are available only at 6-monthly intervals over most of the period covered and steel orders are not published at all.

^{page 48 note (1)} G. H. Moore, ‘Statistical indicators of cyclical revivals and recessions’, Business Cycle Indicators, edited by G. H. Moore, Princeton University Press 1961, vol. 1, pages 206-7.

^{page 49 note (1)} For convenience, the cyclical factor is combined with the trend in this analysis.