REPULSION RESULTING FROM RADIATION. 
507 
centre of gravity could be altered by a screw, and a pointer projecting upwards 
enabled me to see a movement of the beam. A brass ball ( c , c ') 6 millims. diameter, 
was soldered to each end of the beam. The balance was adjusted so that it made 
one oscillation in about five seconds. 
19. The case consisted of a rectangular box of brass, 10 mm. thick, the front of which 
was replaced by plate glass. At one end two holes were drilled (d, e) about 20 mm. 
apart, and a curved piece of glass tube was cemented in, so that both ends were out- 
side. At the top a hole (f) was drilled to receive a thermometer, and another (g) to 
receive the tube attaching the instrument to the Sprengel pump. By sending a current 
of hot water through the bent glass tube, heat could be communicated to one of the 
brass balls, and the movement, if any, of the beam could be seen by a micrometer in 
front. Many experiments were tried with this apparatus, and the result appeared to be 
that warming the ball caused it to sink (49). This action might, however, be due to 
the expansion of the brass beam by heat ; so to obviate this source of error, I sought 
for a material wherewith to make a beam which should be as little affected as possible 
in this manner. 
20. A stick of fine-grained charcoal was worked up into the shape of a beam, and 
fitted at each end and in the centre with appropriate metallic collars for the needle- 
points and brass balls. To get rid of absorbed gases, which experience showed were 
evolved unequally in a vacuum, and thereby threw the beam out of adjustment, it was 
first heated strongly in an exhausted tube, and then soaked in an alcoholic solution of 
shellac. When quite dry the charcoal was heated till the shellac fused. After much 
difficulty the beam was adjusted, and being enclosed in the brass case described above, 
exhaustion was effected. With this charcoal beam I also found that heat generally 
caused the brass ball to sink (49). 
21. These results were opposed to what I had before noticed, but many anomalies 
were observed (49). Thus the diminution of gravity did not appear to vary as the 
rarefaction increased ; and the position of the hot body, in relation to the brass ball, 
seemed to have considerable influence on the direction and amount of movement. It 
was also difficult to make the brass case, with its numerous joints, sufficiently tight to 
hold a Sprengel vacuum*, even by painting it over with gold-size when partially 
exhausted, and I therefore decided to form the beam of some other material. 
22. A mica beam was at first tried, but it was found to be liable to split across. 
Magnesium possesses the advantages of lightness and rigidity, but was inapplicable, 
* As I shall have to speak of various kinds of vacua, it will be best to name them distinctively to avoid 
periphrasis. I shall call the best vacuum which my air-pump will give an air-jpump vacuum — this is one or 
two mill i metres below the barometer. The ordinary vacuum produced by the Sprengel pump I shall call a 
Sprengel vacuum — in this the gauge is appreciably level with the barometer. A so-called “ perfect ” vacuum, 
produced by potash and carbonic acid, as subsequently described, or by similar means, I shall call a chemical 
vacuum. I object to the term perfect, as applied to any vacuum at present known, as I believe that where 
force can travel we are not justified in assuming the absence of matter — imponderable it may be, and unaffected 
by ordinary forms of force— but none the less matter. 
