Using data from a mobile photoelectic device, we present two-dimensional return-stroke velocities from 130 strokes that consist of 86 negative natural, 41 negative triggered, one positive triggered, and two positive natural return-strokes. The velocity measurements are divided into two groups: ''short-channel'' values with channel segments starting near the ground and less than 500 m in length and ''long-channel'' values that start near the ground and exceed 500 m in length. The average long-channel velocity is 1.3¿0.3¿108 m s-1 for natural return-strokes and 1.2¿0.3¿108 m s-1 for triggered return-strokes. For the short channels, natural return-strokes have a statistically higher (at the 98% confidence level) average velocity of 1.9¿0.7¿108 m s-1 than triggered strokes with an average velocity of 1.4¿0.4¿108 m s-1. We compare our results with previous measurements of the return-stroke velocity for comparable channel segment lengths and find that there is no statistical difference between our results and the previous ones.

A qualitative analysis indicate that the optical waveform from a return-stroke generally increases in risetime and decreases in amplitude at greater altitudes. Using the transmission line model (TLM), the shortest segment one-dimensional return-stroke velocity, and either the maximum or plateau electric field, we find natural strokes have a peak current distribution that is lognormal with a median value of 16 kA (maximum E) or 12 kA (plateau E) while triggered lightning has a median peak current value of 21 kA (maximum E) or 15 kA (plateau E). We find substantial differences between TLM and the shunt-measured peak currents and find no significant correlation between shunt peak currents and triggered and return-stroke velocities. We find correlations between TLM peak currents and velocities for triggered and natural subsequent return-strokes but not between TLM peak currents and natural first reture-stroke velocities. In both cases where a correlation is found, the trend is for greater peak currents to be associated with lower return-stroke velocities.

Our peak current calculations are approximately half as large as previously reported. This difference is attributed to the greater return-stroke velocity measured by our photoelectric device and used in the transmission line model. Hence previous estimates of the return-stroke peak current seem too high. Two natural single-stroke positive cloud-to-ground flashes had velocites of 1.0 and 1.7¿108 m s-1 and peak currents of 120--150 kA. ¿ American Geophysical Union 1989