William C. Valine, an atmospheric scientist at the University of Arizona, explains.
Ever since Benjamin Franklin's time lightning has been understood to be a large electrical discharge similar to that seen when a conductive object (like a metal doorknob) is touched after a static electric charge is picked up (by feet scuffing across carpet, for example). But whereas the spark from static electricity measures a centimeter or less in length, a lightning channel can span five kilometers or more. (Also, cloud-to-ground lightning involves electrical currents on the order of tens of thousands of amps. In contrast, a circuit breaker for a common household circuit is usually rated at 20 amps.) Because of its extreme scale, lightning is a complex physical phenomenon.
During a thunderstorm the lower portion of a cloud contains a region that accumulates a large negative charge and the upper portion becomes positively charged. Also, a positive charge is induced on the surface of the earth, because the negative charge in the cloud is closer to the ground. Although some lightning strikes, or lightning flashes, transfer the positive charge to ground, such positive flashes are rare in comparison to lightning flashes that transfer negative charge to ground. From the appearance of a lightning flash it is clear that this charge transfer occurs in a channel with a width that is small compared to its length. Because air is normally an electrical insulator, it must break down so that the conductive channel can form in order for a lightning flash to occur. This breakdown of the air between the cloud and the ground does not happen all at once, however. Instead, it happens in discrete steps of about 50 meters, with each step taking about one microsecond and about 50 microseconds elapsing between steps. Because of the discrete nature of this process, the initial channel of a lightning flash is called a stepped leader.
Although the formation of a lightning channel is a result of the attraction between negative charge in the cloud and positive charge induced in the ground, the individual steps are often far from vertical. (Note the small horizontal section of the channel shown in the photograph above.) This variability arises because the conductivity of the air is not uniform. The channel will tend to extend out to regions of higher conductivity (as shown in the figure by some of the branches actually pointing slightly upward). As a result, the relatively short step size and the random distribution of such regions of higher conductivity render the channel jagged rather than smooth. Also, the point where the channel contacts the ground may be displaced a considerable distance horizontally from the point inside the cloud where the channel began.
Answer originally published July 7, 2003.