From analysis of simultaneous electric field and TV records of 76 negative cloud-to-ground lightning flashes in Florida, various lightning properties have been determined and several new facets of lightning behavior inferred. Only 17% of the flashes were single-stroke flashes, less than half the commonly claimed percentage (e.g., Anderson and Eriksson, 1980). The initial electric field peak (and, by inference, current peak) for the only strokes in single-stroke flashes was smaller than for first strokes in multiple-stroke flashes. Half of all flashes, single- and multiple-stroke, struck ground at more than one point, with the spatial separation between the channel terminations being up to many kilometers. One third of multiple-stroke flashes had at least one subsequent stroke whose distance-normalized initial electric field peak exceeded that of the first stroke in the flash. Thus such flashes are not unusual, contrary to the implication of most lightning protection and lightning test standards. Subsequent strokes of the order of 2 through 4 were more likely to create a new channel termination on ground than strokes of the order of 5 and higher.

Further, leaders of lower-order subsequent strokes following previously formed and not-too-good (100 ms or less) channels were more likely to show stepping, as opposed to continuous propagation (i.e., to be dart-stepped leaders rather than dart leaders), than were leaders of higher-order strokes. Finally, lower-order subsequent return strokes exhibited a larger initial electric field peak than did higher-order strokes. The second leader of the flash (the first subsequent leader) encounters the least favorable propagation conditions of all subsequent strokes: more than half of the of the second leaders either deflected from the previously formed path to ground or propagated in a stepped, as opposed to a continuous, fashion along the lowest part of that path. It is important to note that interstroke intervals preceding second strokes are similar to or shorter than those preceding higher-order strokes. These observations indicate that channel conditions for the propagation of a subsequent leader are determined not just by the immediately preceding channel heating and cooling processes but rather by the entire channel history.

In particular, the status of the channel apparently depends on the number of strokes that have participated in its cumulative conditioning. The overwhelming majority of long continuing currents, those with a duration longer than 40 ms, were initiated by subsequent strokes of multiple-stroke flashes as opposed to either the first stroke in a multiple-stroke flash or the only stroke in a single-stroke flash. Strokes that initiate such long continuing currents were (1) relatively small (in terms of both return-stroke field peak and as determined from an independent study in New Mexico, stroke charge), (2) followed relatively short interstroke intervals, and (3) showed a tendency to be preceded by a relatively large stroke. Millisecond-scale K and M electric field changes appeared different in terms of both microsecond-scale pulse content and interevent time intervals. Often no microsecond-scale K and M field pulses were detected. When they were present, such pulses were highly variable and sometimes irregular in waveshape, as opposed to the alleged characteristic K-pulse waveform described by Arnold and Pierce (1964), which has been extensively used in atmospheric radio-noise studies. There is a remarkable similarity between many lightning characteristics in Florida and in New Mexico. ¿ American Geophysical Union 1994