We present spectroscopic and photometric observations of the luminous
narrow-line Type IIP (plateau) supernova 1994W. After the plateau phas
e (t greater than or similar to 120 days), the light curve dropped by
similar to 3.5 mag in V in only 12 days. Between 125 and 197 days afte
r explosion, the supernova faded substantially faster than the decay r
ate of Co-56, and by day 197 it was 3.6 mag less luminous in R than SN
1987A. The low R luminosity could indicate less than or similar to 0.
0026(-0.0011)(+0.0017) M-. of Ni-56 ejected at the explosion. The emis
sion between 125 and 197 days would in this case be dominated by diffu
sion of emission from the mantle region, or by an additional power sou
rce, presumably circumstellar interaction. Alternatively, the late lig
ht curve could have been dominated by Co-56 decay. In this case, the m
ass of the ejected Ni-56 would have been 0.015(-0.008)(+0.012) M-., an
d the rapid fading between 125 and 197 days most likely due to dust fo
rmation. Although this value of the mass is higher than in the first c
ase, it is still lower than estimated for any previous Type II superno
va. Only progenitors with M(ZAMS)similar to 8-10 M-. and M(ZAMS)greate
r than or similar to 25 M-. are expected to eject such low masses of N
i-56. If M(ZAMS)similar to 8-10 M-., the plateau phase indicates a low
explosion energy, while for a progenitor with M(ZAMS)greater than or
similar to 25 M-., the energy can be the canonical similar to 10(51) e
rgs. As SN 1994W was unusually luminous, the low-mass explosion may re
quire an uncomfortably high efficiency in converting explosion energy
into radiation. This favors a M(ZAM)greater than or similar to 25 M-.
progenitor. The supernova's narrow (similar to 1000 km s(-1)) emission
lines were excited by the hot supernova spectrum, rather than by a ci
rcumstellar shock. The thin shell from which the lines originated was
most likely accelerated by radiation from the supernova.