6 
The: Colorado Experiment Station 
fore the water reaches the end of the partition board. The par¬ 
allel divisor board is preferable to the swing divisor board. 
In addition to the tests on divisor boxes, 196 tests were made 
on dividing the flow over rectangular and Cipolletti weirs. Crest 
lengths of 2 and 4 feet were used for both types of weirs. A thin 
metal plate was placed on the downstream side of the weir so 
that its edge touched the crest of the weir and extended vertically 
above the weir crest into the weir notch. This plate was set for 
different experiments at intervals of 2 inches across the entire 
width of the weir, and separate channels caught the flow over the 
weir on the two sides of the plate. These channels were placed 
far enough below the crest of the weir to allow a free passage of 
air under the over-pouring sheet of water. Under these condi¬ 
tions both types of weirs give reasonably accurate divisions, the 
greatest error being with the rectangular weir set to divide the 
flow between two parties on a basis of *4 and 24, when the actual 
deliveries will be 24 and 76 per cent for a head of 0.2 foot, and 
22*4 and 77y 2 per cent for a head of 0.8 foot. The errors with 
Cipolletti weirs used as divisors were in the opposite direction 
and about one-half as great. When either weir would be used to 
divide the water equally between three parties the error would 
be quite negligible. 
If the divisor plate is placed out some distance from the weir, 
or the edge is placed in a horizontal position below the weir crest, 
the discharge for the end division would be considerably short 
and the flow for the middle division would be accordingly greater 
than the desired amount. 
Fig. 1 
