352 H. H. ROBINSON 



ment on a beach, the other on a terrace) . Such discrepancies as just 

 noted may give rise to large errors (50 per cent, or more) in the calcu- 

 lated gradient of a shore-line when the distance between the two points 

 of observation is short. These larger fluctuations in water-level would 

 prove especially troublesome where strong cutting or filKng was occur- 

 ring. When the water-plane is gradually dropping a blurring of 

 shore forms frequently results, although this is not always the case as 

 Goldthwait's profile at Jacksonport shows.' On the other hand, if 

 the water-level remained constant for several years a well-marked 

 terrace might easily be cut or a strong beach built, while with a drop 

 to a new level, at which the water might again stand for some time, a 

 second terrace or beach might be formed. As the record shows, 

 differences of 2 or 3 feet would be common between such terraces and 

 beaches. It seems difficult to escape the conclusion that a source of 

 error in the reconstruction of water-planes exists here that cannot be 

 eliminated. It may range from o to 6 and possibly a larger number 

 of feet. 



4. The quotation from Taylor, above given, raises the question 

 whether the level of a water-plane can be judged more closely from 

 cut terraces or beaches. From the fact that beaches show such wide 

 variations in position with respect to the water-plane, it would seem 

 probable that terraces furnished the best guide. Upham has stated: 

 " Cliffs eroded by the lake waves give more definitely the plane of the 

 water surface"^ than beaches. This applies, however, only where 

 the forms are well preserved. When they are not well preserved, it 

 seems likely that a beach would be a better guide than a terrace, since 

 the original surface of a terrace is more liable to become obscured 

 than that of a beach. This brings up the point that both beaches 

 and terraces in time become more or less eroded and covered with 

 vegetation, thus making it difficult to tell just what point should be 

 chosen in measuring their elevations, and leading to uncertainty in 

 correlation. Lawson has said, in speaking of a certain locality on the 

 north shore of Lake Superior, that "a series of terraces can be seen 

 scoring the hills to the north at a distance of probably a little more 

 than a mile and a half at the farthest. An effort was made to locate 

 these terraces by running a line of levels from the railway, but this 



I Op. cit., p. 76. 2 op. ciL, p. 277. 



