Flagellate bacteria move by rotating semi-rigid helical flagella in either a clockwise or anticlockwise direction, using the frequency of switching between the two directions of rotation to bias their overall movement in a favourable direction (Macnab, 1978). It is probable that flagellar rotation involves the movement of protons down the membrane proton gradient, between two complex protein rings in the cytoplasmic membrane. One protein ring appears to be fixed to the flagellum and free to rotate, and the other is fixed in the membrane. The movement of protons via an unidentified proton pore in the ring system causes the two rings to move against each other. This rotation can use either the electrical or chemical potential to power the rotation (Khan & Macnab, 1980). Although the energetics of flagellar rotation have been investigated, the mechanism by which flagellar rotation is initiated, or the rotation in one direction is stopped and restarted in the opposite direction, by using a unidirectional proton gradient, is unknown. Photosynthetic bacteria become non-motile if incubated anaerobically in the dark, but regain motility upon actinic illumination. If a useful protonmotive force is the only requirement for initiation of flagellar rotation, as is assumed for its continued rotation, then the time taken for motility to resume upon illumination after incubation in the dark should be similar to the time for the creation of a stable increased membrane potential. Rhodopseudomonas sphaeroides possesses an endogenous indicator of membrane potential, the carotenoids. When there is a change in the potential across the cytoplasmic membrane, there is an electrochromic shift in the absorbance of the carotenoids (Jackson & Crofts, 197 1). Photosynthetically grown R . sphaeroides were isolated into either Hepes [4-(2-hydroxyethyl)-l-piperazine-ethanesulphonic acid1 or potassium phosphate buffer (lmM, pH7.0) with IOpM-disodium EDTA. After several hours anaerobic incubation in the dark, the bacteria were actinically illuminated and the carotenoid bandshift was measured in an Aminco dual-wavelength spectrophotometer as previously described (Armitage & Evans, 1979). There was a very fast rise in membrane potential as measured by the carotenoid bandshift (half-time, ti 5 5 ms). Similarly, if the bacteria were incubated in 1 O p ~ buffer for several hours in the dark in an anaerobic stirred vessel there was a decrease of about 0.1 pH unit within 15 s of illumination. However, when R. sphaeroides, isolated under similar conditions and sealed into micro-slides (50mm, with O.lmm pathlength), was examined with similar illumination conditions on the microscope, the first bacterium did not start to move for several seconds and the normal percentage of motile bacteria was not observed until more than 60s illumination. The effect could be enhanced by the addition of low concentrations of uncouplers. If 1 ~ M C C C P ' (final concn.) was added to the bacteria in the dual-wavelength spectrophotometer, there was a gradual blue shift in carotenoid absorbance taking several minutes. When the new baseline absorbance stabilized (the decrease in potential being presumably the result of residual respiration in the dark), the actinic light-induced bandshift remained unaltered or increased. The bacteria therefore retained their ability to maintain a light-induced membrane potential. Measurement of external pH under similar conditions showed the same result as for untreated bacteria. However, examination microscopically of the CCCP-treated cells showed an increase in the time taken for motility to resume. It was now several minutes before the first bacterium moved, and many more before 50% were motile. When motility restarted, individual bacteria moved at normal speed and their behaviour was similar to that of untreated bacteria. The length of the delay was related to the concentration of uncoupler, which could also be related to the decrease in the size of the bandshift as the concentration of uncoupler was increased. As the light-induced membrane potential, measured by the bandshift, decreased, the time taken for motility to resume increased (16 s for first motile bacterium at 0.5 ~ M C C C P , 30s at l p ~ , 65s at 5pM). However, the time taken for motility to resume at each uncoupler concentration was independent of the length of dark incubation after uncoupler addition, the maximum delay remaining constant after 12 min dark incubation. The length of the delay was also pH-dependent; at pH6.5 O. I~M-CCCP caused a delay of 2min, whereas at pH7.5 the same concentration only delayed the onset of motility by 7 s. These results suggest that the protonmotive force alone is not enough for the initiation of flagellar rotation, but there is an additional potential-dependent component, possibly a protein modification. It is possible that this is also required for the switching mechanism, as bacteria with artificially low membrane potentials have been shown to be unable to change direction (Khan & Macnab, 1980).