Prior to the time the Committee on Runoff of the Boston Society of Civil Engineers published their report (Report of the Committee on Runoff, J. Bos. Soc. C. E., v. 9, No. 8, 1922), duration‐curves of stream‐flow had been prepared almost universally by plotting the runoff in cubic feet or in cubic feet per second per square mile as ordinates against the cumulative percentage of time as abscissae. [Note: Throughout this paper the basic unit of stream‐flow is defined as rate of discharge in cubic feet per second.] This report showed that if the ordinates were expressed as ratios to the mean flow, the duration‐curves for different streams in New England would be surprisingly similar, regardless of any differences in the size and character of the drainage‐basin, or in the total flow. This procedure, originally proposed by Hazen (Allen Hazen, Discussion of Power‐estimates from stream‐flow and rainfall‐data, by Dana Wood, J. Bos. Soc. C. E., v. 3, No. 3, 299–303, 1916), is confirmed by Barrows (H. K. Barrows, Water power engineering, p. 128, 1927), who maintains “…that for the rivers of the east and south the flow, as a per cent of the mean, could be expressed, for the dryer 80 per cent of the time, by the formula, log Q = (2.40 ‐ 0.011 T), where Q is discharge as a per cent of the mean and T is per cent of time.”