58 BULLETIN 1017, U. S. DEPARTMENT OF AGRICULTURE. 
work has not been comprehensive enough to provide for extensions 
that have or may become desirable ; moreover, proper attention to the 
outlet facilities has been neglected in some, and in others the design 
of the system has been faulty. Many of the early ditches were shal- 
low, and too small to afford good drainage. The ditches that are now 
being constructed are generally deeper and larger, and are there- 
fore much more efficient as drains. The snow and ice in the spring 
fill many of the drains, but there seems to be no way to overcome 
this difficulty. 
Damaging floods have occurred in Thief and Roseau River basins 
at times of extraordinary storms such as that of July, 1919 (see fig. 
12 and Table 9). A part of this area is extensively ditched, many 
of the land sections having ditches on two or three sides. Never- 
theless at the time mentioned the ditches and streams overflowed, 
the water completely submerging large areas and doing great- dam- 
age. The trouble seems to have been due, mainly, to the inability 
of the two rivers in their present condition to carry off the water 
brought to them through the artificial channels, and to some extent 
to the over-taxing of these artificial channels by the run-off from 
too large areas. Under normal conditions most of the land affected 
is free from damaging water. 
The development of graded highway construction throughout the 
watershed has grown to be a somewhat important factor in the re- 
moval of run-off. As it has progressed, more and more attention has 
been given the proper drainage of road grades until at present prac- 
tically no new construction is undertaken without provision for the 
removal of water which might prove injurious to the roads. The 
network of road ditches which is rapidly spreading over the entire' 
watershed has a marked and increasing effect upon the rapidity with - 
which the run-off from spring thaws and storms collects in the 
streams and drainage ditches and is a factor not to be neglected 
when considering the rates of run-off over the drainage basin. On 
the other hand, road embankments in some localities have interfered 
with drainage outlets. 
TILE DRAINAGE. 
Tile drainage has barely passed the experimental stage in the Red 
River Valley, but the practicability of this method of local drainage 
for the farm lands of this valley has been fully demonstrated by tests 
that have been carried on continuously since 1907 at the Minnesota 
State Experiment Farm at Crookston and by several private under- 
takings. The soil in which these tests have been made is typical of 
that occurring generally over the flat .areas along the river and the • 
results obtained are indicative of what may be accomplished through- 
out the low-lying lands. The private tile systems that have been 
installed have operated with varying degrees of success. It is doubt- 
