( 454 ) 



connected with the frictions presenting themselves, i. e. these values 

 will be independent of the initial condition. At the same time the 

 two clocks will show a different rate ^), of which clocks one therefore 

 will have to sustain the rapid principal oscillation, the other the 

 intermediate one. Periodic transference of energy will then take place. 

 Probably it will not be easy to realize this condition, character- 

 izing itself particularly b}^ the fact, that one of the clocks goes consi- 

 derably faster than would be the case when placed independently ^). 

 The initial conditions will then have to be chosen in such a manner 

 that from the very beginning one oscillation will predominate for one 

 clock, the other for the other clock. And this will become all the 

 more diflicult as c^ and c^ become more and more equal, therefoi-e 

 according as the two clocks become more and more alike and 

 are suspended in a more symmetric way. For, so much smaller 

 will, according to what was mentioned in note (2) p. 453 be the 

 difference in proportion of the amplitudes y.^ and y.^ at each of the 

 oscillations. ^) 



17. Finally we wish to point out how we must represent to our- 



1) So this differs agnin from what Elligott observed in his last experiments, 

 so that these cannot be regarded as the reahsation of this case, though they have 

 the transferences of energy in common with it. However, between the fact of 

 those transferences and the assurance that both clocks have entirely the same 

 rate exists a contradiction, as we have already seen, which is not to be solved, 

 hideed, those transferences can be explained by interference only, so they require 

 the cooperation of two oscillations of different periods ; but these oscillations must 

 both be sustained if the state is really to continue indefinitely, and then each 

 of them by one of the motive works where the oscillation referred to will predominate 

 the other one. See also the last note. 



To me it seems most probable that with Ellicott the transferences of energy existed 

 only at first indicating the temporary presence of the rapid principal oscillation. 

 Ellicott's wording is not emphatically against this conviction. 



2) The difference from case A is of course only quantitative. In both cases the clocks 

 go faster than when placed independently, but in case Cthe acceleration of the quickest 

 clock becomes much greater than that of the less rapid one (see § 13). A gradual 

 transition presents itself then, and the case of Ellicott was probably situated on 

 that transition-line. 



3) The idea that perhaps each of the motive works might be able to take over 

 one principal oscillation and the other in turns had to be set aside after a closer 

 investigation. If we compose in the well-known graphical way two oscillations of 

 unequal amphtudes and of periods of oscillation differing but little, it is evident 

 that the motive work will go alternately somewhat quicker and somewhat slower 

 than will correspond to the period of oscillation of the greatest amplitude, but this 

 can never go so far that the rate of the smaller amplitude is taken over, not even 

 for a short time. 



