The M9.8 solar flare of 1991 October 24 22:30 UT presents several inte
resting characteristics: (1) energy release starts high in the corona;
(2) the primary chromospheric ribbons are initially well separated an
d do not move apart at an observable rate; (3) no evidence is found fo
r an erupting filament or other driver. To explain this flare, we cons
ider several canonical flare models, including a filament eruption, a
confined filament eruption, current interruption, and interacting loop
s. We conclude that none of these scenarios unequivocally explains thi
s flare. Two possibilities which cannot be ruled out are (1) the erupt
ion of a filament unobservable in Ha which starts high in the corona a
nd produces no ribbon motions smaller than our detection threshold and
no perceptible expansion of the coronal X-ray source, and (2) energy
release due to spontaneous, propagating reconnection which allows the
system to essentially brighten in place.