192 Trans. Acad. Sci. of St. Louis. 
the suspended masses could be detected. If any effect 
was produced it was very small. The large masses were 
turned 90° in position, with a like result. 
An alternating current of 20 amperes was applied. No 
effect could be detected when the masses were in the lat- 
ter position. When the lines through the points of con- 
tact of the brushes with the large spheres was coincident 
practically with the line through the centers of gravity 
of the two masses nearest to each other, the gravitational 
attraction was quickly reduced to zero, and made nega- 
tive. The repulsion was about 50 per cent of the normal 
attraction. To bring about this result required about 
fifteen to twenty minutes of time. When the double 
switch was opened, so that the large masses were wholly 
separated from the source, and the masses were grounded 
it required between two and three hours for the large 
masses to recover from the shock which they had re- 
ceived. This has been repeated many times, with no dis- 
cordant results. It may be that the parts of the large 
spheres which are most affected by the alternating cur- 
rent are those parts near the brushes. If there are heat 
effects, they tend to oppose the observed effect. Burning 
candles replacing the large masses cause a change in the 
reading in the opposite direction from that of the al- 
ternating current, the glass plates being removed. This 
is due to convection currents within the screen. 
The work here described has been done in a private 
laboratory in the second story of Eads Hall, now oc- 
eupied by the physics department of Washington Uni- 
versity. 
My thanks are due to the Carnegie Institution of 
Washington for meeting the expense of this work. 
EXPLANATION OF PLATES. 
Plate XLV. Variation in Gravitational Attraction. 
Plate XLVI. Variation in Gravitational Attraction. 
Plate XLVII. Variation in Gravitational Attraction. 
Issued November 8, 1917. 
