OVERFLOW OF RED RIVER OF THE NORTH. 63 
flow is placed at the extreme east end. Both of these structures are 
to be of reinforced concrete. For the location, elevation, and dimen- 
sions of the dam see figure 3, sheet 2, and figure 4. A Levee at the 
south end of the lake, near Browns Valley, will be necessary to pre- 
vent overflow into Bigstone Lake. A channel extending about 2 miles 
back into the reservoir is planned to facilitate flow to the sluiceway. 
The embankment of the railroad spur from Browns Valley to the 
grain elevator at the south end of Lake Traverse must be raised to 
avoid interruption of traffic by water in the reservoir. (See fig. 
3, sheet 1.) On account of the navigation on Lake Traverse it is not 
permissible to lower the water surface of the lake artificially below 
elevation 972, the present low-water stage. As the breaking of the 
reservoir when full would cause a great flood and much damage to 
property and crops, it has been made of such dimension as will give a 
large factor of safety, and it will store the run-off from a storm much 
larger than any on the Red River of which we have a record. 
RUN-OFF TO BE CONTROLLED. 
Fairly complete precipitation records from about 1893 to date are 
available, as are also data (April, 1917, to date) from which the dis- 
charge of the Mustinka River into Lake Traverse can be computed, 
and for the discharge of the Bois de Sioux for 1919 and 1920. In 
addition there are some data relative to flood heights and a few 
measurements of run-off made in the watershed. Also, there is avail- 
able considerable information in regard to run-off from other drain- 
age areas in the watershed of the Red River. The records of dis- 
charge are shown in Table 3. In view of the paucity of actual flood 
discharge measurements it was necessary, in order to decide upon 
the probable run-off from the watershed under consideration, to es- 
tablish some relation between precipitation and run-off. 
In studying the available records of run-off, as shown in Table 3, 
it was found that by far the greatest run-off per square mile that 
has ever been recorded in any watershed tributary to the Red River 
occurred in the Wild Rice watershed in Minnesota as a result of 
the storm of July 21-23, 1909. (See Table 9.) Computations of 
monthly and annual run-off for the period July, 1909, to December, 
1917, also show that the maximum monthly and half-yearly run-off 
measured during that period occurred on the Wild Rice River during 
1909. 
As the run-off from January to June, 1909, was not measured and. 
a maximum year could not be estimated therefrom, the records from 
January to June, 1916 (showing maximum half-yearly run-off since 
1909), and from July to December, 1909, were combined as though 
they had occurred in one year. The period January to June, 1916, 
