1. Experiments with Biloxi soybean (Glycine max [L.] Merr.) showed that length of photoperiod as well as length of dark period influenced the degree of flower initiation. With effective dark periods there were lower-critical, optimum, and- upper-critical lengths of photoperiods. With effective photoperiods there were lower-critical, optimum, and upper-critical lengths of dark periods when cycles were shorter than 36 hours long. 2. These critical and optimum lengths of photoperiod became shorter when accompanied by longer dark periods; the converse was also true for critical and optimum lengths of dark period in relation to the length of the accompanying photoperiod. 3. Cycles with photoperiods 8 hours long and cycle lengths ranging from 18 to 56 hours showed two distinct optimum lengths of dark periods. These optima differed in duration by about 24 hours. With relatively short dark periods there was a lower-critical duration below which, and with very long dark periods there was an upper-critical duration beyond which, flowering did not occur. In addition, flowering did not occur at dark periods of intermediate lengths. 4. When these results were plotted on the basis of length of cycle, maximum flowering occurred on cycles either 24 hours long or 48-52 hours long. Therefore, the flowering response in Biloxi soybean is apparently influenced to a marked degree by some endogenous rhythm of 24 hours' duration. 5. Cool temperature during part of the cycle markedly influenced the response of plants to variations in photoperiod, dark period, or length of cycle. When exposed to cool temperature during a portion of cycles of 32-36 hours, the plants flowered. If not exposed to cool temperature in similar cycles, they failed to flower. Cool temperature during a portion of a cycle seemed to have some influence on the endogenous rhythm, so that the length of the rhythmic cycle was other than 24 hours. 6. A discussion is given of the possible nature of this endogenous rhythm and its relationship to the photoperiodic results. Two tentative hypotheses are presented which might explain the results obtained with cycles of different durations at both normal and low temperatures.