The effect of alternating temperatures on the times taken by seeds of lentil (Lens culinaris Medikus) to germinate was investigated using a two-way temperature-gradient plate. Between 5 and 25 degrees C, warmer temperatures increased the rate of germination. Variation among the individual seeds in the times required for germination at different constant temperatures within this range were described well by a log-normal distribution of thermal times, accumulated above a base temperature of 1.5 degrees C. Even with amplitudes as great as 20 degrees C, no effect of alternation pei se on the thermal time required for germination was detected-whether the cool temperature was applied for 8 or 16 h d(-1). Similarly, in alternating temperature regimes where the minimum temperature of the diurnal cycle was between 0 degrees C and the base temperature, the thermal times required for germination (where no thermal time accrued during the periods when temperature was below T-b) were in close agreement with those values provided by the model determined at warmer constant temperatures. However, where the minimum temperature applied was < 0 degrees C the germination of all but the earliest germinators was delayed beyond model predictions, and more so where the sub-zero minimum temperature was applied for 16 rather than 8 h d(-1). The results, therefore, contradict the view that alternation in temperature pet se reduces the thermal time required for seed germination. Rather, rate of germination responds instantaneously to current temperature, but prolonged exposure to sub-zero temperatures can result in damage sufficient to delay germination when seeds are returned to regimes warmer than the base temperature.