portion. The east-west tread of roughness Indicates Instead relief- 

 forming processes which act relatively constantly through time, but 

 quite variably along the ridge strike. Francheteau and Ballard (1982) 

 describe the along-strlke variability of processes along the East 

 Pacific Rise, and relate the changes to distance from the ridge crest/ 

 fracture zone Intersection. The change In process Is expressed by the 

 relative Importance of fluid lava flows and pillow lava flows. These 

 petcologlc changes are In turn associated with the elevation of the rift 

 valley along the ridge crest; topographic highs are associated with high 

 ratios of fluid lavas and topographic lows associated with pillow lavas. 

 Indeed the shallowest portion of the Gorda Ridge segment Is located near 

 latitude 42"45'N, which shows relatively low roughness values as one 

 would expect from the sheet-like flow of fluid lavas. Ridge flank areas 

 become rougher adjacent to deeper axial valley segments, which might 

 reflect the rougher surface of pillow lavas. Although other explana- 

 tions for the trend of roughness on the ridge flanks could be put forth, 

 the distributions found in this study are consistent, although not nec- 

 essarily typical. Extension of the model to more thoroughly investi- 

 gated ridge crests should shed light on this particular hypothesis. 



The patterns apparent in the distribution of a are not as evident 

 when one examines the distribution of the spectral slope (b). It is 

 clear that the "universal" value of -1 for the spectral slope Inferred 

 by Bell (1975b) and others is not supported by this modelling effort. 

 The reason for this discrepancy is not readily apparent, however the 

 attention given to defining stationary sample space in this study does 

 represent one major difference in method. In order to test this hypoth- 



71 



