THE RIGIDITY OF THE EARTH 589 



in the case of the solar tide), and taking the mean of the first, 

 second, third, etc., ordinates. The resulting values were plotted 

 and any error in computation was usually indicated by the failure 

 of a point to fall on a smooth curve. It is important to treat the 

 observed and calculated tides both in the same way, as any dis- 

 tortion in the resulting sine curves due to lack of complete ehmi- 

 nation of other periods affects the two alike. This is, of course, 

 most noticeable in the case of the diurnal tide on account of its 

 smaller amplitude and the smaller number of periods. Mr. Fred 

 Pearson gave valuable assistance in. measuring the fihns, in plotting 

 the curves, and in deducing the various tides from the curves. 



Very little trouble was caused by sudden erratic changes in the 

 fringes. Occasionally, however, earthquakes would cause the 

 fringes to disappear for from ten minutes to half an hour. Once 

 the effects of an earthquake were evident for about six hours. 

 During three hours of this time the fringes were completely 

 obliterated. 



The most serious disturbance was a gradual change in the slope 

 of the observed curves. This would often be fairly uniform and 

 gradual for a month or two. At some times the curves would rise 

 and at others fall. Sometimes the N-S and E-W slopes had the 

 same sign, and sometimes opposite signs. We have been able to 

 discover nothing systematic about this drifting. It may have been 

 caused by unequal settling at the ends of the pipes, by temperature 

 changes in the pits, or by tilting in the earth's strata. There were 

 always large shifts of the fringes when the lights came on after 

 having been interrupted by the power company for a half-hour or 

 so. The change of slope was eliminated in reducing each monthly 

 tide, as given in Tables I-VI, and the tide for the whole year, 

 opposite Y in the tables, by distributing the change of level uni- 

 formly throughout the period. This change of slope is quite 

 conspicuous in Figure 3, where the observed E-W tide showed a 

 fairly uniform and distinct downward trend throughout nearly the 

 whole month. The change in slope pf the N-S curves for the same 

 period is comparatively small, as shown in Figure 2. 



Plate III is from four photographs taken simultaneously in 

 the four pits. The reproductions are positives on the same scale 



