30 The: Colorado Experiment Station. 
walls below the weir, in order that air may freely circulate underneath. 
For short weirs it is sufficient that the lateral walls of the lower canal 
be free from the sides of the weir. In such case, when air freely 
passes underneath, the level of the water in the lower canal has no 
effect' on the discharge unless it reaches to the level of the crest of the 
weir. 
5. The depth of the water should be measured accurately at a 
point where the suction of the flow does not affect the level of the 
surface and where it is free from influences like the wind, or the move¬ 
ment of the water, which affect the true level. The height should be 
read to within 1-300 of the depth in order that the error may be within 
one-half of 1 per cent., or 1-67 to be within 1 per cent. This means to 
one-fifth of an inch when depth is one foot. 
6. The weir needs to be carefully constructed. It should not 
vary more than 4° from being perpendicular to the channel. Its sill 
should be horizontal. 
Table I. will aid in obtaining proper proportions between the 
channel of approach and the weir. 
The conditions affecting the flow may be divided into three 
classes; those which always tend to increase the discharge, those 
which always tend to decrease, and those which may either increase 
or decrease, as the case may be, and thus in the long run, tend to 
balance. 
The flow of the weir is decreased by eddies or by obliquity 
of the weir to the sides. 
The quantity is increased by the velocity of the approaching 
water, by nearness of the sides to the end of the weir, or the bottom 
of the box to the crest, or by any condition which prevents a com¬ 
plete contraction at the sides or at the crest. If sand or silt collects 
in front of the weir board, this decreases the contraction and 
increases the quantity. If the water below the weir board backs, 
up so as to prevent air from having access underneath the sheet 
of falling water, then the vertical contraction is lessened and the 
discharge is increased. 
If the crest is not sharp or its thickness more than one-fourth 
of the depth of the water on the crest, the vertical contraciion is 
lessened and the flow is increased. When the thickness still further 
increases the effect is to decrease the flow. 
It is evident that the changes that are likely to happen increase 
the flow of water over that given by the formula. These are dif¬ 
ficult to avoid and in general cannot be entirely eliminated. It is 
because of this that Cippoletti proposed to make an arbitrary in¬ 
crease of one per cent, in the formula. This is not necessary and 
it would seem to be better to use the ordinary Francis formula, 
making the weir as near perfect as possible and making allowance, 
if necessary. 
The correction to be allowed to the tables may be estimated in 
some cases by the following considerations: If the weir is not ver¬ 
tical, the discharge is increased or diminished according as the in- 
