214 



If the average drift was really no greater tliau that obtained when 

 special effort was made to determine its amount it was not sufficient to 

 balance the considerable excess of yositive readings. 



Summary: Of the total number of readings, 61 per cent, indicated 

 positive results, 20 per cent. Avere negative, and 13 per cent, were zero, 

 i. e., gave no deflection. Of the total amount of the deflections (omitting 

 the rather questionably large ones in Table I), 82 per cent, were positive 

 and 18 per cent, negative. Including those of Table I. 90 per cent, were 

 positive. 



