UNCLASSTFTRn 



SECURITY CLASSIFICATION OF THIS PAGE(T*7l»n Data Enterad) 



The remaining 80 experiments investigated the response of previously established 

 bed forms to changes in T or a or both. The ripple length, X, and height, n, 

 were measured from photos, except when bed forms were three dimensional. 



Sand grains began to move on the flat bed at flow velocities generally com- 

 patible with earlier observations. A Shields criterion for motion initiation 

 was found in need of modification for use in oscillatory flow. Prior compres- 

 sion of the bed had small effect. 



With the grains in motion, ripples were initiated by and would spread from 

 slight irregularities on the bed or around its edge, or could be generated 

 everywhere by a slight increase in velocity. Flow separation and vortex forma- 

 tion occurred at an early stage. Rolling-grain ripples as described by Bagnold 

 (1946) did not always form and seemed unnecessary to vortex ripple development. 

 The initial length of the vortex ripples was small and remarkably independent 

 of a. 



During growth, the dimensionless ripple length and height. A/a and n/a, were 

 expressible as functions of (D/a)n, where n is the number of periods since rip- 

 ples first appeared on the flat bed. Ripple growth was approximately complete 

 when (D/a)n = 1. 



Profiles of two-dimensional ripples in equilibrium with the flow were gener- 

 ally as described by Mogridge and Kamphuis (1972) . In particular, X/a and n/a 

 were fairly constant over a wide range of intensity of flow. However, with end 

 crests fixed, and a varying, a profile could be "strained" into various stable 

 forms over a considerable range of X/a. 



Both two- and three-dimensional final bed forms were observed. Criteria for 

 these forms were not clearly established, but appeared to depend on both the 

 amplitude and intensity of flow, and the grain size. 



UNCLASSIFIED 

 2 



SECURITY CLASSIFICATION OF THIS PAGEflVTien Date Entered) 



