('. B. STARRING. 189 



stats'' was more efficient than its opponent, and hence 

 carried it with it. 



But apart from all theory, it is not difficult to show 

 that such is not the case. We have only to set one, two, 

 three, or four wheels going, and vary them in every pos- 

 sible manner, taking care only that the wheels shall re- 

 volve in the direction of the arrow ellipses, and in every 

 case the whole system will revolve in the direction indi- 

 cated by the arrow ellipse A. 



It may, however, be thought that the horizontal rota- 

 tion is due to the friction of so many axles, and conse- 

 quently that if these were frictionless, there would be 

 no such rotation, and that Sir William Thomson is right 

 after all. 



But, in fact, this motion is in direct opposition to the 

 friction. A very pretty proof is the following. Set the 

 instrument in operation, placing the bottom so that the 

 four axes make a kind of square, as in fig: 45, and leave 

 it to itself. The bottom will descend, and soon the axes 

 become vertical, as in fig. 44, while the wheels are yet in 

 rapid motion. The machine which had, up to this time, 

 been rotating in the direction indicated by the arrow el- 

 lipse A, fig. 45, will now gradually come to a stop, and 

 then begin to revolve in the opposite direction ; in other 

 words, go backwards, increasing in speed to a very con- 

 siderable velocity. Therefore, the friction of the bear- 

 ings tends to prevent the gyroscopic motion, and to pro- 

 duce the opposite. 



Attach a cord to the bottom piece, and pass it up 

 through a little eyelet attached to the upper piece near 

 the conical point. By means of this it is easy to raise or 

 lower the bottom without otherwise disturbing the instru- 

 ment. Hold the cord when the wheels are in operation 

 and the instrument gyrating as shown by fig. 45, so that 

 the bottom cannot fall at all ; the horizontal rotation will 

 cease. Slacken the cord, it will commence again. 



173 



