surface may be very considerable; goods may be damaged, business interrupted, building foundations weakened, and health of citizens impaired by dampness after the floods have withdrawn. The decision as to whether or not to provide capacity for these unusual storms must be made for each case by the engineer or the city authorities as their judgments dictate; but probably in nine cases out of ten the wisest decision would be not to endeavor to provide for very infrequent storms of unusual intensity-say those occurring less often than once in ten years.

If it is decided that storms occurring less frequently than once in ten years may be disregarded, plot as above the maximum rate for each year for each record for 5-, 10-, etc., minute periods; add together the durations, in years, of the several records, divide this by ten, and the quotient will give the number of points plotted for each precipitation period that may be disregarded. This number of highest rates having been eliminated, a curve can then be drawn that will embrace all of the remaining high rates. (See Fig. 5.)

A number of diagrams showing records of this kind for several sections of the country, and curves embracing the maximum rates for which formulas have been calculated, have been prepared and made public from time to time by a number of engineers. Some of these formulas are as follows:

Maximum (not exceeded more than once a century), r =

For the United States east of the Rocky Mountains (by Talbot):

360
1+30

Vicinity of New York City (by Kuichling), heavy rainfalls of uniform in

Rate in Inches per Hour (Y)

Table No. 9 gives a number of the maximum rates of rain

fall recorded; and in the last column, for comparison, the rates corresponding to Webster's "extraordinary," which might be adopted as reasonable for sewer design.

FIG. 5.-PHILADELPHIA RAINFALL DATA AND CURVE.

Each dot represents a rate (at the date indicated) that was equaled or exceeded not more than ten times during the period.

In each of the above, r is intensity of rainfall in inches per hour and is the minutes of time for which this intensity was the average.

The records seem to show, where any information on the subject is given, that the maximum intensity usually lasts but a few minutes, seldom more than ten; that it sometimes occurs at the beginning of a storm, sometimes at the middle or end of it, quite a number of storms stopping within ten or twenty minutes after the maximum rate is attained.

It is to be borne in mind that a rate for any given timesay thirty minutes-is the average for that time, during which

CINCINNATI RECORDS.

Each shows the rates for successive five-minute periods of the storm that gave the maximum total for the period stated.

there may have been a maximum rate for five or ten minutes several times this average. For use in calculating run-off from small areas, actual rainfall for each of a number of successive minutes is desirable; but for general purposes the amount is usually expressed as the rate per hour. One advantage of this is that one inch per hour on an acre equals almost exactly one cubic foot per second.

Few data are available showing how large an area is experiencing a maximum rate at a given moment. Frequently,

FIG. 7.-MAXIMUM RAINFALL CURVES AT CINCINNATI; WITH SUGGESTED CURVES FOR USE IN DESIGNING.