180 BRUSH— A KINETIC THEORY OF GRAVITATION. 



muth double the air resistance of the zinc, the time required for the 

 bismuth to gain one beat was 274 minutes. 



Thus it is seen that when we give the zinc more than five times 

 its uncompensated normal excess of air resistance the apparent zinc- 

 bismuth effect is increased only one half ; and when we give the bis- 

 muth 100 per cent, excess of air resistance, instead of its normal 18 

 per cent, deficiency, the apparent zinc-bismuth effect is still very con- 

 siderable after correcting for radius of gyration effect. 



I hope these reductio ad absurdum experiments make it clear that 

 no error need accrue from swinging the compensated pendulums in 

 air instead of in vacuum. It is true that in vacuum the periods of 

 both pendulums would be slightly shortened; but they would be 

 equally shortened. 



Effect of Local Distortion of the Earth's Magnetic Field. — It has 

 been suggested that distortion of the earth's field by large masses of 

 iron in the laboratory and in the building structure may have some- 

 thing: to do with the observed zinc-bismuth effect. 



There are no considerable masses of iron nearer than about 2 

 meters from the swinging bobs, and their distorting effect must be 

 very nearly uniform in the small space in which the bobs swing; so 

 that both bobs will be affected substantially equally. However, to 

 test out the effect of great distortion of the earth's magnetic field in 

 the immediate vicinity of the bobs, and very unequally for the two, 

 the following experiment was made : 



After synchronizing single zinc and bismuth bobs on the short 

 pendulums, a large cylinder of very pure soft iron 31 cm. high and 

 7 cm. diameter, shown on right front corner of table in Plate VIII. 

 (never left there, however, during other experiments), was placed 

 vertically with its axis in the plane of the two pendulums at rest, and 

 with its axis 10 cm. away from that of the nearer bob ; and later in 

 the same relation with the other bob. Then the bobs were started 

 swinging together. 



In each case the nearer bob was accelerated, but only about 11 

 per cent, as much as bismuth is always accelerated when the bobs are 

 exchanged after synchronizing. The amplitudes were slightly dimin- 

 ished. When a glass cylinder of similar size was substituted for the 

 iron, no disturbance of period was observed. 



