174 BRUSH— A KINETIC THEORY OF GRAVITATION. 



The new and enlarged value of the zinc-bismuth effect found, as 

 above described, clearly does not invalidate the usefulness of Table I. 

 as a check on the accuracy of the work done before correcting the 

 weight of the zinc bob. If the new value were introduced in a similar 

 table, it would simply increase the zinc-iron effect (less hours for 

 one beat gain of iron) as well as the zinc-bismuth effect, and show 

 that the iron is more closely related to bismuth than to zinc in eft'ect 

 producing qualities. 



Inequality of bob weights is much the largest and most dangerous 

 source of error I have found in the pendulum experiments ; dangerous 

 because it is not very obvious (no one has pointed it out), and if 

 overlooked may lead to wholly erroneous results — even reversing the 

 sign of the zinc-bismuth effect. I have said so much about it for the 

 guidance of others who may wish to try pendulum experiments for 

 themselves. 



The next step was to make a new pair of zinc and bismuth bobs 

 as nearly as possible exactly like the original pair, except that the zinc 

 cylinder is slightly larger in diameter so as to have the required weight 

 without banding, and the bismuth cylinder purposely has a coarser 

 crystalline structure. 



The new castings were made like the old ones, in molds consisting 

 of vertical iron cylinders somewhat larger in diameter than the fin- 

 ished bobs, with an asbestos paper tube extending above. Thus the 

 castings were chilled at their lower end, and rather rapidly cooled 

 throughout. 



But before machining, the bismuth casting only, was packed in 

 molding sand in an iron box and gradually heated in a large electric 

 muffle furnace above the melting point of bismuth, and then allowed 

 to cool very slowly through the freezing stage of the metal by suitably 

 reducing the heating current in the furnace. This procedure surely 

 resulted in a crystalline structure of very large grains. 



Finally, the new cylinders were worked to weight, length, paral- 

 lelism of ends, and adjustment of center of gravity with all the care 

 and patience exercised with the earlier cylinders. The weights of 

 the new and old cylinders are now as follows : 



Weight of old zinc 1,383.899 grams 



Weight of new zinc 1-383-893 



Weight of old bismuth 1,383795 " 



Weight of new bismuth 1.383-798 " 



