Discussion of “Annual floods and the partial duration flood series”

The writer is interested in finding from this paper a similar idea which he had in mind when engaging in a statistical study of hydrologic data, a part of the Highway Drainage Research Project being carried out in the Department of Civil Engineering, the University of Illinois. However, a complete presentation of the study will be reserved for future publications, and only fragments of thought are discussed here.

In treating hydrologic data for the study of their frequency of recurrence, two phases of study are generally met. One is the frequency of recurrence for the observed distributions, and the other is the corresponding frequency of recurrence for the theoretical distributions of best‐fit. The former is usually required for the purpose of plotting observed data, and hence is called the “plotting positions,” a term used by Foster [1934, p. 1217]. The latter is treated by the mathematical theory of probability, and generally serves as a theoretical interpretation of the observed phenomena.