154 Trans. Acad. Sci. of St. Louis. 
hung two bars of wood, at right angles to each other, 
each being loaded with about ten pounds of rock. This 
was done in order to prevent lateral vibrations of the 
magnetometer case. The arrangement is fairly well 
shown in Plate XXXVI, Fig. 1. When this view was taken 
two of the bars bracing the torsion-head had been re- 
moved, in order to make the details of the arrangement 
more clearly visible. The tent had also been removed. 
The magnets serving to deflect the needle into an east 
and west position were steel bars, four feet in length 
and having a cross-section 3/8x1.5 inch. These bar-mag- 
nets were enclosed in heavy rubber tubing with plugs 
closing the ends, and mounted in V-shaped supports 
within wooden boxes. The boxes were filled with dry and 
fine grained saw-dust, which was packed closely around 
the tubes enclosing the magnets. These boxes are shown 
at the ends of the framed structure forming the table. 
They were inclined at an angle of 45° with the magnetic 
meridian. See Fig. 1. 
The compensating magnets, which served to reduce 
the strength of the resulting field thus formed, were of 
four like bars of steel. On either side of the needle were 
two bars in reversed position, bound together by strong 
cords. They were wound with adhesive rubber tape. 
They were separated from each other by blocks of wood 
having a thickness of about one em. They were mounted 
on edge within the boxes. They thus acted differentially 
upon the needle. Single bars on either side, when place 
anywhere within the available space, would reverse the 
needle, and give too strong a resultant field. The two 
boxes, each of which contained such a doublet of such 
compensating magnets, are shown on either side of the 
magnetometer in Plate XXXVI, Fig. 1. These boxes were 
also packed with saw-dust. The four boxes were clamped 
to the supporting framed structure by heavy wooden 
clamps. The boxes and frame were loaded with fully 
half a ton of rocks. An end view of the south end of the 
table is shown in Plate XXXVI, Fig. 2. The framed struc- 
