MOLECULAR PRESSURE, AND THE TRAJECTORY OF MOLECULES. 
141 
At a pressure of 1 *4 millims. the bright velvety light has spread over all four vanes 
on their metallic sides. With the screw of the coil arranged to give a faint spark 
the movement is negative. If a Leyden jar is inserted, or the contact screw turned 
to give strong sparks, the movement is positive. Inserting the jar makes the halo 
fainter. The best pressure for negative rotation is 1T4 millims. 
When the pressure is '51 millim., the dark space on the metallic side extends nearly 
to the glass, and the whole inside of the bulb is luminous; the rotation is now always 
positive with the coil. Introducing a piece of wet string into the circuit causes the 
vanes to go negatively. 
As the exhaustion proceeds, the dark space widens out and flattens itself against the 
glass, the positive rotation getting faster until the maximum is reached. Exhaustion 
beyond that point causes the speed to diminish, but I cannot get negative rotation 
again by any amount of over exhaustion. 
505. Two radiometers of this kind were made and sealed off, one at a pressure of 
T9 millim., showing good positive rotation, and the other at a pressure of IT4 
millims, showing negative rotation. Connecting the two in series, that is to say, the 
vanes forming the negative pole of one with the positive of the other, and passing 
the induction current through, caused them both to rotate at the same time, one 
positively and the other negatively. 
506. The vanes of the radiometer were now made of aluminium cups (334) slightly 
favourably presented to the side of the glass. They were almost hemispheres, bright 
on both sides. The other parts of the radiometer remained as before. 
When the exhaustion is low and the induction coil is connected, the fly rotates 
negatively. This is caused by the electrified air flying off the edges of the cups, and 
driving them round like the well-known electrical fly. Luminosity appears on the 
cups at a pressure of 3'5 millims., and as the exhaustion proceeds it spreads over each 
side and is separated by a dark space. This widens out, retaining as much as possible 
the shape of the cup, passing successively through the appearances shown in Plate 14, 
fig. 11, a, b, c, d, and e. The luminous margin to the dark space is concentrated at the 
concave side of the cup, forming a luminous centre, and widens out at the convex side. 
As soon as the luminous boundary of the dark space reaches the side of the glass 
positive rotation commences. On continuing the exhaustion, the dark space becomes 
flattened against the glass, and the rotation becomes more rapid. At higher exhaus¬ 
tions, the dark spaces which surround each cup in the form of an irregular ellipsoid 
drawn in at the focus of the cup, touch one another. They now act as if they were 
solid elastic bodies, and become flattened out along the lines of contact (the distance 
between two cups not being large enough for complete ellipsoids) and form broad 
longitudinal lines of light down the bulb between each pair of cups, the rest of the 
bulb being comparatively dark. These lines are nearer the concave than the convex 
side, and turn with the fly. 
507. The train of reasoning carried out in pars. 498, 499, and 500, is therefore seen 
