28 JOSEPH BARRELL 



forming roughly a harmonic series. The theory apphes best where 

 elongated unit areas are flanked by areas of opposite sign. But 

 even where a single axis of uncompensated elevation or depression 

 is surrounded with a region of mean elevation it may be regarded 

 as a half of a wave sustained by the rigidity of the earth. The 

 stresses in the vertical plane through its crest line would appear 

 to be less than, but not so greatly different from, what they would 

 be for a continued series. Where the load is of oval instead of 

 zonal distribution the stresses would also be somewhat diminished 

 in case the oval is surrounded by a neutral region, but if a series 

 of ovals of opposite signs is analogous to two intersecting wave- 

 series, although the stresses would be complicated and are not in 

 general the sum of the stresses due to the separate series, yet it 

 does not appear that the extreme maxima would be necessarily 

 less than the sum, and many of the maxima would be greater than 

 the maxima of the separate series. 



Harmonic loads with short wave-lengths. — Horizontal compres- 

 sion builds mountain folds of which the individual ranges are 

 clearly the results of compression and not of isostatic elevation. 

 Erosion dissects an elevated country on a pattern of a certain scale, 

 deep valleys of erosion separating the crust remnants above base- 

 level which are not yet consumed. These actions produce a surface 

 relief which corresponds roughly to various harmonic series of 

 appreciable amplitude and wave-length, but in this section will be 

 considered only the geodetic evidence of variable mass not iso- 

 statically compensated, much of the variation being due to the 

 concealed heterogeneities of density. 



The distance between Washington, D.C., and Hoboken, New 

 Jersey, as estimated in Part V, Section B, is 326 km. Within this 

 distance are three intermediate stations and the two limiting 

 stations, each station showing a gravity anomaly opposite in sign 

 to that of the adjacent stations. There must be then at least 

 two wave-lengths. The average change of anomaly between the 

 adjacent stations is 0.021 dyne. As it is wholly improbable that 

 the stations are located at the crest Hnes of the waves, the whole 

 amplitude may safely be taken as at least 0.026 and the wave- 

 length 160 km. 



