Fig. 3

Our model correctly predicts majority of peaks in power spectrum of fly locomotion. a Power spectra of individual wild type and clock mutants of Drosophila measured in constant darkness for 5–7 days. X-axis given as ratio \(T_{0} /T\), with the circadian period \(T_{0}\) indicated in each case. Increasing values indicate shorter periods of oscillation. For each \(T_{0}\), prominent secondary peaks are found at \(T_{0} /T = 2,3, \ldots\) accompanied by lower power Dirichlet kernel peaks. b Comparison of peaks detected in the data to peaks predicted by the model was obtained by analyzing wt (N = 29), per ^S (N = 22) and per ^L (N = 19) flies. Only peaks higher than \(p = 0.005\) were used in the analysis. \(y = x\) is shown as a solid line, 10 % deviation shown as dashed lines. For wild type and clock mutants more than 88 % of the data peaks for T = 2–35 h can be explained by the model with ± 10 % error. c Power spectra of a wild type fly measured in LL and a per ⁰ mutant measured in DD. Neither spectrum shows peaks higher than \(p = 0.005\) significance level (dashed line). For both graphs \(T_{0} = 24\)  h was used for scaling the abscissa