We present observational data for the 1989 March 7 white-light flare (WLF), including SMM/HXRBS hard X-ray measurements and CCD optical images in 5000 angstrom continuum and at 3 angstrom in the red and blue wings of the Halpha line. The optical data were acquired at a 0.5 s rate under good seeing and were further processed to remove image motion and distortion. The flare kernel shows an impulsive rise in both Halpha and continuum that is well correlated with impulsive hard X-rays. The Halpha emission shows a red asymmetry which we interpret as Doppler redshift associated with explosive heating of the chromosphere. These results are consistent with previous observational descriptions which traditionally have been interpreted in terms of heating and ionization of the flare chromosphere by a nonthermal electron beam. However, new observational results which have not been reported previously include (1) the separation of the flare kernel into a bright inner core and a fainter outer region, where the two components display distinctly different temporal behavior and amount of Halpha red asymmetry, and (2) a delay, relative to hard X-rays, of approximately 1 s in the impulsive rise of the Halpha wing emission, followed by an additional 1-2 s delay in the 5000 angstrom continuum. It is concluded that the observed fluxes and timing of the hard X-ray and WLF optical emissions are consistent with chromospheric heating by nonthermal electrons, with additional visible light continuum possibly being contributed by a backwarmed photosphere irradiated by intense chromospheric recombination continua (principally Balmer continuum). Power sufficiency of nonthermal electrons in balancing WLF radiative losses is shown to be more easily satisfied when the latter irradiation process applies, than in the case of a purely chromospheric model. The observed delays in impulsive Halpha wing emission relative to hard X-rays may be related to the nonthermal ionization time scale of the chromosphere, or may involve the formation of a chromospheric condensation. A satisfactory explanation for the continuum delay is not obtained.
