22 



TRANSPORTATION OF DEBRIS BY RUNNING WATER. 



The sixth column of the table contains an 

 index of the range in diameter within each 

 grade, and this is also the range in linear fine- 

 ness. The index is the ratio between the 

 diameters of the apertures of the two sieves by 

 means of which the grade was separated. It 

 would express accurately the ratio of the 

 diameter of the largest particle to that of the 

 smallest particle in the same grade if all the 

 particles were spherical, or if all had precisely 

 the same shape. But there are actual differ- 

 ences of form sufficient to modify materially the 

 character of the separation. For the same 

 mean diameter a prolate form will pass a sieve 

 that will arrest a sphere, and an oblate form 

 may bo arrested where a sphere will pass. By 

 reason of this qualifying condition, the actual 

 range is somewhat greater than the tabulated 

 estimate. The range for bulk fineness, given in 

 the last column, is the cube of the range of 

 linear fineness. 



METHODS OF EXPERIMENTATION. 



t 



The methods here described are those used in 

 the investigation of stream traction. Those 

 used in studying flume traction are set forth in 

 the chapter on that subject. 



GENERAL PROCEDURE FOR A SINGLE EXPERI- 

 MENT. 



The experiment trough stands horizontal. 

 The width of channel has been fixed by the 

 placing or the omission of the partition. The 

 head gate is open. The openings in the hopper 

 have been set to a particular size. The outfall 

 contractor has been adjusted to a width pre- 

 viously found suitable for the conditions of the 

 experiment. The two sand-catching boxes 

 stand on the car, one of them being in position 

 under the opening in the trough bottom. 



The pump is started. The measuring gate is 

 opened until its index reaches the point corre- 

 sponding to the desired discharge. A valve 

 associated with the pump is turned, if neces- 

 sary, to adjust the water level in the high 

 trough. When the flow in the experiment 

 trough has become steady, or nearly steady, the 

 sand feed is started by opening the valves 

 which admit water to the hopper. 



The "run" has now begun, and it is con- 

 tinued without change until the slope of the 

 deposit constituting the channel bed has 



become stable. The car in the cross trough is 

 now moved so as to bring the reserve sandbox 

 into position, a stop watch is started, and the 

 deflector in the settling tank is shifted. These 

 changes begin the period of load measurement. 

 During this period measurements are made of 

 the depth of the current and, under certain 

 conditions, of the slope of the water surface, 

 and the character of the sand bed is recorded. 

 The measuring period is terminated by again 

 shifting the sand boxes and the deflector, and 

 the watch is stopped. The head gate is now 

 closed, sand feed and pump are stopped, and 

 the discharge-measuring gate is closed. Noxt 

 the slope of the channel bed is measured, and 

 the sand caught during the period recorded by 

 the watch is weighed. 



DETAILS OF PROCEDURE. 



In the following paragraphs some details and 

 variants will be described, with comments on 

 methods and apparatus. As the investigation 

 had no precedents to follow, its methods were 

 necessarily developed by a process of trial and 

 and failure. Many defects of method were 

 remedied as the work went on. A few that were 

 recognized after much work had been done were 

 tolerated to the end, because their correction 

 would have made the body of results less homo- 

 geneous. 



WIDTH OF CHANNEL. 



The full width of the 31-foot trough and the 

 150-foot trough was 1.96 feet. By the use of 

 partitions the channel was given widths ap- 

 proximately two-thirds, one-half, one-third, 

 two-ninths, and one-ninth of a full width. The 

 scale of width was, in feet, 0.23, 0.44, 0.66, 1.00, 

 1.32, 1.96. A few experiments employed also 

 the widths 1.2, 1.4, 1.6, and 1.8 feet. 



DISCHARGE. 



As the discharge was determined by flow 

 through an aperture of adjustable size, under an 

 adjustable head, the precision of its measure- 

 ment involved (1) the precision of setting the 

 gate at the aperture, (2) the precision of cali- 

 brating the aperture for different positions of 

 the gate, and (3) the precision of adjusting the 

 head. 



The rating (described in Appendix B) was 

 eiTected by a volumetric method, believed to be 

 adequate, the average probable error of its 

 determinations being about 1 in 500. 



