[ 35 ] 
The Silt Problem and Storage Reservoirs. 
3 
set, one from the top, one from the bottom and two intermediate points. 
The discharge was to be measured at the time the collections were made, 
than if the settling process had been carried on for one year, and will 
very probably be greater. If there is an error in this assumption it is 
on the safe side. 
Most of the samples collected were taken with a tin can which was 
provided with a bale at the top and a wire loop soldered to the bottom. 
To this loop a weight could be attached and another attached to the 
bottom of a cord to which the can could be fastened at any desired dis- 
tance above the weight at end of cord. A rubber stopper, having an 
arrangement to which another cord could be tied, was fitted tightly into 
the top of the can. On being lowered into the water the stopper could 
be withdrawn and the can filled at the desired depth. • I made some col- 
lections myself with a special water sampler constructed for the Depart- 
ment. This sampler consisted of a brass cylinder, or drum, having slid- 
ing gates at the ends that could he held open bv means of stops. A 
vane attached to the apparatus served to keep it end-on to the cur- 
rent. By pulling a cord the stops could he withdrawn and the gates 
suddenly closed hv strong springs, thus imprisoning a cylinder of th^ 
water which had been flowing under normal conditions at the time of 
closing the gates. For the sake of comparing- the results obtained bv 
the two methods a few samples were collected in both wavs at the same 
time, hut no material difference was found to exist. Also a few samples 
were taken near the banks of the streams, where the current was slow, 
on the same occasions that samples were collected in midstream, and 
while some difference in the distribution of the silt in vertical planes 
was observed this was not decisive and the average percentages were 
practically identical. 
I did not find it necessarv to take four samples each time, but varied 
this somewhat as the conditions of the streams varied. If the water 
were very turbid four samples were collected, whenever possible, no mat- 
ter how much sediment was carried. For comparatively small amount® 
of sediment three samples for each set only were collected and for still 
smaller amounts of sediment two, or even one, sample was deemed suf- 
ficient. 
Two Texas streams were selected in which the observations were con- 
ducted under my own direction — the Brazos river, at the Jones bridge, 
six or seven miles west of College Station, selected because of its acces- 
sibility, and the Wichita river at Wichita Falls, chosen because of the 
fact that an immense irrigation system has been projected^ there, the 
construction of which the recentlv defeated constitutional amendment 
will very seriously delay. Had this amendment carried, the State at 
large would probably have had a splendid opportunity to witness the 
results of an extensive experiment without cost to any one save those 
