The water tunnel used in these experiments has been described by 

 Lofquist (1977). Essentially, the tunnel is a U-tube with the middle 

 horizontal part comprising the test section. The water is moved 

 sinusoidal ly back and forth over the sand by the positive displacement 

 of tight-fitting pistons. The amplitude a, which can be changed only 

 when the apparatus is at rest, and the period T, which is continually 

 adjustable, can both be selected within wide limits. These limits con- 

 tain most of the normal prototype range. The test section and the 

 sand bed which fills its lower part are 253 centimeters long and 20.8 

 centimeters wide, with a normal depth of flow over the (undisturbed) 

 sand bed of 29.8 centimeters. At either end of the test section the 

 flow continues horizontally through smooth-bottomed "spools" for 41 

 centimeters before beginning its upward curve. 



In this study, even the largest ripples were too low to reveal 

 any flattening of their crests by the presence of the top of the test 

 section. Thus, it is assumed that effects of the top were generally 

 negligible. Possible effects of the sidewalls on three-dimensional 

 bed forms are discussed in Section VII, 4. 



Because of the probable significance of "end effects," the 

 geometry of the ends of the test section and of the sand bed deserves 

 special attention. A drawing of a longitudinal section through one of 

 the spools and the adjacent end of the test section is shown in Figure 

 6. The sand bed is bounded by a curved ramp, of 10.2 centimeters 

 (4 inches) radius, which is flush with the floor of the spool. This 

 curvature reduces local scour. Once disturbed, the sand surface might 

 end either against the curved ramp or at some point on the floor of the 

 spool, possibly even reaching the nearer screen. This screen is 31 

 centimeters from the crest of the ramp and has openings of 0.15 

 millimeters; the farther screen at the back of the spool has openings 

 of 0.25 millimeters. The first screen serves to contain the sand while 

 both together act to make the oncoming flow more nearly uniform. To 

 inhibit premature grain motion and scour at the ends of the 'sand bed 

 the flow velocity near the ends was lowered relative to that over the 

 major part of the bed by placing a scow-shaped constriction against 

 the top of the tank. This constriction, which was made of Lucite, with 

 a maximum thickness of 4.6 centimeters, reduced the depth of flow over 

 the level of the undisturbed sand bed to 25.2 centimeters. Its shape 

 is shown in Figure 6. The curved transition at either end of the scow 

 extends for 32 centimeters. 



To define conditions for the initiation of grain motion or for 

 spontaneous ripples on a flat bed requires that the surface itself be 

 first well defined. To this end a special device leveled the submerged 

 bed while giving it a nearly constant and repeatable degree of 

 compaction. This was a scraper held in a motor-driven carriage which 

 ran along a track and the tops of the tank walls with the top of the 

 tank removed. 



28 



