strongly and clearly on a/D than on either ((> or r. Figure 32 may 

 be compared with Figure 9 in Inman (1957) which attempts to show 

 the domain o£ ripple occurrence on a field of U versus D. With the 

 present dimensionless variables, this is roughly equivalent to <^'i versus 

 r^/^. In Inman 's plot the effects of a/D Cor of T) are not indicated; in 

 Figure 32 r is absent, except as a parameter in the criteria for sheet 

 flow of Manohar (1955) and Chan, Baird, and Round (1972). 



VIII. SUMMARY 



This study has discussed observations of ripples from a series of 

 laboratory experiments, without any attempt at theoretical models; the 

 direct results of these observations are summarized in this section. A 

 guide to this summary is provided by the list of "emerging questions" 

 in Section I,3,h. These questions and related findings are put in two 

 groups. 



The first group of findings relates to the initiation of grain 

 motion and ripples on a flat bed. Their primary application is in 

 support of physical models of ripple development: 



(a) ipQ has been found to be a function of both a/D 

 and r (Fig. 12). 



(b) A prior compression of the sand surface was found 

 to have only a small effect upon (})(. (Fig. 12). 



(c) A Shields criterion for the initiation of grain 

 motion under oscillatory flow was found in need of further 

 modification (Fig. 13). 



(d) ^j. was found to be distinct from cj)^, but the 

 difference was small and detectable only with the finer 

 sands (the arrows in Figs. 10 and 11). 



(e) Rolling-grain ripples were observed only with the 

 two finer sands. These ripples appeared not to play an 

 active or necessary role in the formation of vortex ripples. 



(f) The initial length of vortex ripples was found 

 to depend slightly on D but not on a (Fig. 15). This 

 independence suggests that vortex ripples are not caused 

 by a prior disturbance moving with the water. 



The second group of findings relates to the character of bed 

 forms, whether in growth or in equilibrium. In large part, they con- 

 firm the design curves of Mogridge and Kamphuis (1972), while enlarging 

 their experimental base with other sand grain sizes at moderate and 

 large values of a and T. 



81 



