We compare two observations of gamma-rays before, during, and after lightning
flashes initiated by upward leaders from a tower during low-altitude winter
thunderstorms on the western coast of Honshu, Japan. While the two leaders
appear similar, one produced a terrestrial gamma-ray flash (TGF) so bright that
it paralyzed the gamma-ray detectors while it was occurring, and could be
observed only via the weaker flux of neutrons created in its wake, while the
other produced no detectable TGF gamma-rays at all. The ratio between the
indirectly derived gamma-ray fluence for the TGF and the 95% confidence
gamma-ray upper limit for the gamma-ray quiet flash is a factor of
$1\times10^7$. With the only two observations of this type providing such
dramatically different results -- a TGF probably as bright as those seen from
space and a powerful upper limit -- we recognize that weak, sub-luminous TGFs
in this situation are probably not common, and we quantify this conclusion.
While the gamma-ray quiet flash appeared to have a faster leader and more
powerful initial continuous current pulse than the flash that produced a TGF,
the TGF-producing flash occurred during a weak gamma-ray "glow", while the
gamma-ray quiet flash did not, implying a higher electric field aloft when the
TGF was produced. We suggest that the field in the high-field region approached
by a leader may be more important for whether a TGF is produced than the
characteristics of the leader itself.
