178 BRUSH— A KINETIC THEORY OF GRAVITATION. 



when applied to the single (or real) zinc-bismuth effect was inad- 

 vertently overlooked. 



Table II. shows that this correction (subtractive) for either single 

 or double bobs on the long pendulums is 22 per cent., and, of course, 

 grows rapidly as the pendulums are shortened. 



It is seen that the very consistent results of single and double bobs 

 of equal weight on the long pendulums indicate a zinc-bismuth effect 

 of about one part in 32,000. But in all the pendulum and other 

 experiments the end sought is not so much quantitative accuracy as 

 qualitative certainty. In the present stage of the general investiga- 

 tion it matters little whether the true zinc-bismuth effect is one part 

 in ten thousand or a hundred thousand; the great point is to make 

 sure that it is something tangible. This is the purpose of my long 

 course of experimentation. 



Another, though small, correction might be made in Table II. 

 (additive), due to the fact that the pendulum rods are of different 

 material from the bobs. 



Effect of Unequal Air Resistance. — I am often asked if per- 

 formance of the pendulum experiments in vacuum might not greatly 

 affect, or even obliterate, the apparent zinc-bismuth effect. To this 

 I can confidently answer no. As I see it, the only thing to be gained 

 by working in vacuum would be elimination of unequal air resistance 

 due to the unequal diameters of the zinc and bismuth cylinders. 



But this is easily compensated when working in air by attaching 

 to the smaller bismuth cylinder small and very light vanes of paper 

 or very thin aluminum normal to the line of swing of suitable area 

 experimentally found or computed. This was always done in the 

 foregoing experiments. 



Careful experiments have shown that wholly uncompensated in- 

 equality of air resistance would increase the apparent zinc-bismuth 

 effect only about 10 per cent. The effect of the air vanes on the 

 bismuth is to slightly lengthen its period. 



It is well known that wind pressure on a cylinder, normal to its 

 axis, is approximately half that on a plane surface of a width equal 

 to the diameter of the cylinder. Therefore, if we were to place on 

 either double bob cylinder shown in Fig. 2 a plane air vane normal 



