We have imaged the ultraluminous infrared galaxy Arp 220 in light of t
he near-infrared [Fe II] 1.257 mu m and Pa beta lines, and have obtain
ed spectra in the J- and H-band atmospheric windows. Arp 220 is a stro
ng source of [Fe II] and Pa beta emission, with luminosities of 1.3 x
10(41) and 9.2 x 10(40) ergs s(-1), respectively. The [Fe II] and Pa b
eta emission are both extended over the central 2''-3'', but with diff
erent morphologies. The Pa beta line is strongly peaked at the positio
n of the western nucleus seen at 2.2 mu m (Graham et al. 1990) with a
fainter ''spur'' in the direction of the eastern nucleus. The [Fe II]
emission line shows a weak peak at the western nucleus along with diff
use emission extending to the east, but with no indication of a second
ary maximum. The [Fe II] is more extended in the north-south direction
than the Pa beta line. Nearly 75% of the detected [Fe II] emission is
spatially resolved. The overall [Fe II]-to-Pa beta line flux ratio in
Arp 220 is consistent with that seen over similar spatial scales in S
eyfert 2 galaxies, yet larger than what is measured in galaxies with n
uclear starbursts. The [Fe II]-to-Pa beta line flux ratio is spatially
variable being approximately 0.8 at the position of the western nucle
us and approximately 2.0 at radii of up to 500 pc. We suggest that the
extended [Fe II] emission is produced through the interaction of fast
shocks with ambient gas in the ISM at the base of the outflowing, sup
ernovae-driven superwind mapped by Heckman et al. (1987). The bolometr
ic luminosity of the starburst required to power this wind is estimate
d to be at least 2 x 10(11) L.. If the spatially unresolved [Fe II] em
ission is produced via a large number of supernova remnants, the impli
ed rate is similar to 0.6 yr(-1). The overall luminosity of such a sta
rburst could account for a large fraction (1/2-1/3) of the Arp 220 ene
rgy budget, but the large deficit of ionizing photons (as counted by t
he Pa beta luminosity) requires that the starburst be rapidly declinin
g and/or have a low upper mass cutoff. Alternatively, dust may effecti
vely compete with the gas for ionizing photons, or much of the ionizin
g radiation may escape through ''holes'' in the ISM. It is also possib
le that a buried AGN produces a large fraction of the unresolved [Fe I
I] and Pa beta emission. We briefly discuss these possibilities in lig
ht of these new imaging and spectroscopic data.