Stomatal opening towards the end of the night under natural conditions, which is followed, if darkness is artificially prolonged, by reclosure and in a few cases by a second (much reduced) opening and closure about 24 h later, is thought to represent part of the first `transient' cycle of an endogenous rhythm. This is supported by observations that the phase of such opening, in terms of onset and peak, can be shifted by appropriate `light-on' and `light-off' signals. Possible biological advantages of `night opening' in nature are discussed, but these involve maintaining the phase of the rhythm relative to the changing time of the light-on signal (dawn). Phase-shift of night opening had apparently only been investigated in detail for Tradescantia virginiana and here it was only caused by the light-on or light-off signals of photoperiods far outside the range in which this species may be presumed to have evolved. It is not clear how, in less extreme daylengths, the phase of the rhythm in this species can be corrected either for `periods' not exactly 24 h or for changing times of dawn and the rhythm thus `entrained'. The present experiment was carried out with the nearly related Commelina communis to look for interactions between the light-off and light-on signals in 24 h dark-light regimes with a range of photoperiods (7-17 h) thought to exceed that within which the various species used so far for investigating night opening are likely to have evolved. The 3 $\times $ 2 factorial experiment (in four `randomized blocks of time') showed no significant interaction between light-off and light-on for onset or for peak of night opening, in analysis either on the actual times or on their logarithms. Thus the results were consistent both with an additive relation and a product relation between the effects of the two types of signal, but the latter was thought the more probable. Light-off and light-on caused almost the same highly significant mean phase shift per unit change of signal time (`relative phase shift') with a value very close to + 0.5. This had the effect that onset and peak of night opening came at almost constant mean times (about 3 h and 8 h respectively) after `mid-dark', that is after the time corresponding to the middle of the previous controlled dark period, for the whole range of controlled photoperiods from 7 h to 17 h. Thus duration of light or darkness were without effect and phase bore no fixed relation to the time of `dawn'. It appears however that the system could prevent rhythms having transient half-cycles with half-periods other than 12 h from getting more than certain limiting amounts out of phase with the dark-light regime. These results are in marked contrast to those for T. virginiana, but some results for Vicia faba appear to be consistent with peak night opening occurring 8 h after mid-dark, for plants from 8 h, 12 h, and 16 h photo-periods in 24 h dark-light cycles. It seems that too much importance may have been attached to a relation between the phase of the rhythm and dawn.