DAILY RHYTHMS 493 



are longer or shorter than the natural period of A depending on the 

 phase of B when A is reset. A mathematical formulation of the model 

 has been developed (Pittendrigh, Bruce, and Kaus, 1958). 



To explain the fly-resetting data it is unnecessary to assume that B 

 feeds back on /4 ; we may, that is, assume B to be completely driven 

 by A . Nor do we need any special assumptions concerning its tempera- 

 ture sensitivity. However, it soon became evident that the coupled- 

 oscillator model would also explain several additional types of Droso- 

 phila data if we allowed two further broad assumptions: ( 1 ) that there 

 is some slight feedback of B on A; and (2) that B is temperature- 

 sensitive in the senses that it can be entrained by an external tempera- 

 ture cycle, and its period (unlike A\) is temperature-dependent. 



Temperature Dependence of the B Oscillator 



One of the most challenging features of the Dwsophila data has 

 been the strange mixture of temperature dependence and independence 

 they have revealed. Pittendrigh (1954) showed that immediately fol- 

 lowing a temperature drop the overt rhythm of the fly was strongly 

 affected: the period lengthened (Fig. 10). However, it immediately 

 and spontaneously reverted to its natural period and to nearly its 

 original phase. A special temperature-dependent terminal clock, opera- 

 tive only in the last day of pupal life, was adduced to explain these 

 facts. This hypothesis has subsequently been withdrawn (Pittendrigh 

 and Bruce, 1957), and explanation of the behavior has been sought in 

 terms of an oscillator's transients. Perplexing features have neverthe- 

 less remained. As noted in our previous paper (Pittendrigh and Bruce, 

 1957, p. 98) : "There are . . . great differences in the response of the 

 Dwsophila clock to light and temperature stimulation; . . . sharp 

 temperature perturbations produce spectacular transients with rela- 

 tively little phase shift in the ultimate steady state; hght perturbations 

 have, on the other hand, spectacular effects on the steady-state phase 

 shift." 



The coupled oscillator model offers the most promising interpreta- 

 tion of these facts so far available if the B oscillator is taken to be 

 temperature-dependent. The transients produced by temperature drops 

 reflect temporary lengthening of the period of the B oscillator, which 

 is promptly reentrained by the temperature-independent A oscillator. 



