DAILY RHYTHMS 491 



15 and by advances thereafter. In the hamster the switchover from 

 delays to advances occurs at about hour 13; and in no case is the 

 steady-state shift saturated. 



A COUPLED OSCILLATOR MODEL FOR DAILY RHYTHMS 



We can now be more explicit about the interest attaching to the 

 pattern of phase shifts. First, it lacks any obvious adaptive meaning; 

 and we take it, therefore, with some caution, as evidence of a common 

 feature in the mechanism of the diverse daily rhythms it characterizes. 

 Second, its analysis leads to a more restrictive and explicit form of the 

 oscillator model for rhythms. 



The Drosophila case, for which fullest data are available, is con- 

 sidered first. Any model based on a single oscillator seems unable to 

 explain the concurrence of the three features that strongly characterize 

 resetting in the fly: (1) ultimate determination of phase by a signal 

 seen three cycles previous to the new steady state, (2) the presence of 

 transients, and (3) the dependence of transient length on the time of 

 the cycle at which the signal fell. These features are, on the other hand, 

 all explained by a model for the system based on two coupled oscilla- 

 tors. 



Essential features of the two-oscillator scheme are given in Fig. 9. 

 The principal departure from an earlier model (Fig. 1) is addition of 

 a second oscillator, B. The A oscillator is self-sustaining; it can be 

 coupled with and entrained by the light regime of the environment; 

 when free running in aperiodic conditions its phase can be shifted by 

 single light signals; it is temperature-independent; and it is coupled 

 with and drives a second {B) oscillation. The B oscillator is the system 

 whose rhythmic behavior is more immediately reflected in the fly's 

 overt rhythm; in some sense the fly reads phase (hence time) from 

 the motion of B. 



The peculiarities of phase shifting are represented schematically in 

 the lower part of the figure: ( 1 ) the light signal immediately resets the 

 phase of the A oscillator; the new phase, which is expressed by the fly 

 much later, is instantaneously registered by the A oscillator; (2) the 

 transients observed in the fly reflect the gradual reentrainment of the B 

 oscillator (which the fly follows) by the A oscillator; (3) the transients 



