24 



TRANSPORTATION OF DEBBIS BY SUNNING WATER. 



the trough and gradually fed to the current by 

 means of a scraper. Hand feeding had the 

 defect of discontinuity, as well as irregularity 

 in detail, but it had the advantage of measure- 

 ment, and in certain experiments its meas- 

 urement of load gave an important check on 

 the measurement of debris delivered at the 

 outfall end of the trough. With a perfect and 

 stable adjustment of conditions the two should 

 agree, and their disagreement served to show 

 that the slope of the channel bed had not 

 become perfectly adjusted, or else that its 

 adjusted condition was subject to rhythmic 

 oscillation. 



In some of the later work the rate of feed 

 was measured from time to time by inter- 

 cepting the stream of sand falling from the 

 hopper during a definite number of seconds and 

 weighing the sample thus caught. 



THE COLLECTION OF SAND. 



In the original construction of the apparatus 

 for arresting the sand the opening in the 

 bottom of the trough was covered by a coarse 

 wire screen, which lay flush with the trough 

 bottom. This was intended to separate the 

 current above from the still water below and 

 prevent the formation of eddies, which might 

 keep the sand from settling to the collecting 

 box and might also check the current. It 

 fulfilled its purpose and was altogether satis- 

 factory for currents of moderate velocity, but 

 with high velocities it interfered with the 

 arrest of the sand, letting a considerable 

 fraction pass on to the settling tank. It was 

 accordingly removed, apparently without bad 

 results. Eddies were formed, but the antici- 

 pated difficulties were not realized. 



On the whole the apparatus for arresting 

 sand was successful. It was only with the 

 finer debris and at the highest velocities that 

 the fraction of load escaping to the settling 

 trough was too large to be neglected in the 

 weighing. 



DETERMINATION OF LOAD. 



The sand collected, in sand box and settling 

 tank, during the period recorded by the stop 

 watch was weighed without drying, and the 

 gross weight was afterward corrected by an 

 allowance for the contained water. In order 

 to determine the proper allowance a prelimi- 

 nary study had been made, and as a result of 



that study a definite procedure was adopted 

 for bringing the wet sand to a particular 

 "standard" condition. After the sand-collect- 

 ing box had been lifted from the trough all 

 water which would drain from it by gravity 

 alone was allowed to escape. It was then 

 removed to smaller boxes for weighing. These 

 boxes were jarred by tapping, which caused 

 the sand grains to readjust their contacts and 

 settle together, excluding a part of the inter- 

 stitial water, which appeared at the surface 

 and was poured off. The sand was then 

 weighed. It is of interest to note that in the 

 condition thus adopted as a convenient stand- 

 ard sand occupies less space than when dry, 

 moist, or supersaturated; its voids are at a 

 minimum. 



The period recorded by the stop watch was 

 ordinarily about 10 minutes but was made less 

 when the current was most heavily loaded, 

 because of the limited capacity of the sand- 

 collecting box, and was extended for the 

 lightest loads. Its beginning was sharply 

 defined by the shifting of the sand boxes, 

 which could be made to coincide within a 

 second with the starting of the watch. Its 

 end was somewhat less definite, but the error 

 in time is believed to be small in comparison 

 with the whole period. 



The load per second was computed by 

 dividing the total load, namely, the corrected 

 weight of sand, by the number of seconds in 

 the stop-watch reading. Its error included 

 (1) the error of timing, (2) the error of stand- 

 ardizing the sand and correcting for contained 

 water, and (3) the error in weighing. There 

 are no definite data bearing on its amount, 

 and nothing better can be recorded than a 

 general impression that the results are reliable 

 within 2 per cent, that the precision is lower 

 than that of the discharge measurement, and 

 that the error in determination of load is 

 notably less than the error, presently to be 

 considered, in correlating load with slope. 



When the rate of feed was regulated by the 

 periodic contribution of a measureful of d6- 

 bris, the weighings of the unit, from time to 

 time, showed inequalities from which precis- 

 ion could be estimated. A computation indi- 

 cated the average probable error, for a run, as 

 about 1 per cent. This depended chiefly on 

 the standardization, and as that was less per- 

 fect for the debris as fed than for the debris 



