90 0. E. SCHIOTZ. [NORW. POL. EXP. NO. 8.] 
— more than 200 units in the 7th decimal place — as compared with the 
others determined on that day. The observation in question was the first 
made in the evening when the observations were resumed after an interval 
of about 3 hours. The disagreement may therefore possibly be explained 
by assuming that the imperceptible trembling movements in the ice-covering 
had gradually subsided during the time of repose, but that the motion recom- 
menced when this observation was completed, and continued with increasing 
violence through the night and day following. If this explanation is correct, 
this observation with pendulum 33 should about correspond with the obser- 
vation made on June 8th with pendulum 34, as the results also seem to 
indicate; for if these two observations are combined, they give as the 
value for the acceleration, 9°83131, while the second observation alone gives 
9°83136 (see p. 60). 
In accordance with the above, I have, in the preceding calculations, started 
with the assumption that no appreciable error will be committed by leaving 
out of consideration possible simultaneous movement in the substratum of 
the pendulum apparatus. 
