Circadian behavior can selectively synchronize to daily temperature gradients that are combined with continuous increases or decreases in absolute temperature. Median activity records for each experimental condition described in Figure 7 were used to generate circadian activity profiles (mean ± S.E.M.) from which the phase of peak and offset of circadian activity was then determined by interpolation. The phase shifts in peak and offset of circadian activity resulting from the various environmental protocols are summarized in panels (A) and (C). The values and error bars indicated in these panels represent the midpoint and range of the intervals of interpolation for the phase of peak and offset. Phase-dependent resetting of the circadian activity peak and offset was only consistently observed for temperature protocols based on the 4°C daily temperature gradients ('zigzag'). This is illustrated by a side-by-side comparison of the circadian activity profiles (mean ± S.E.M.) found for the 'ramp up' and 'zigzag up' protocols after 12:00 to 0:00 and 0:00 to 12:00 light phases (B) and the 'ramp down' and 'zigzag down' protocols after 6:00 to 18:00 and 18:00 to 6:00 light phases (D). The bars underneath the activity profiles in (B) and (D) indicate the phase of light (gray) and dark (black) during the prior light/dark cycles. Please note that, to facilitate direct comparison, the activity profiles are plotted relative to the previous light period rather than experimental time, which differs by 12 h for the top versus bottom diagrams in (B) and (D).