We present detailed non-LTE (NLTE) synthetic spectra for comparison with a
time series of observed optical spectra of the Type Ic supernova SN 1994I w
hich occurred in M51. With the exceptions of Si I and S I, we treat the imp
ortant species in the formation of the spectrum in full NLTE. We present re
sults for both a hydrodynamic model that has been fitted to the light curve
and for an illustrative custom-crafted model that is more massive. Both mo
dels give reasonable fits to the overall observed spectra; however, neither
is able to reproduce all the observed features. Some conspicuous observed
features are absent, and some predicted features are unobserved. No model t
hat we have explored is able to reproduce satisfactorily the observed infra
red feature near 1 mu m on 1994 April 15 (+7 days), which has been attribut
ed to the triplet He I lambda 10830 transition. The low-mass hydrodynamic m
odel produces an infrared feature with a blend of He I, C I, O I, and Si I-
Si II lines, but it predicts a strong unobserved absorption feature near 61
00 Angstrom due to Fe III, and the observed feature just blueward of 6000 A
ngstrom most likely due to Na D is not reproduced. The more massive model d
oes a better job of reproducing the observed infrared line shape, but also
predicts the unobserved feature near 6100 Angstrom. The early-time spectrum
of the low-mass model is far too blue; thus, a more massive model may be s
lightly favored. Since the predicted infrared feature is produced by a blen
d of so many elements, and there is no overwhelming evidence for other heli
um features such as lambda 5876, it may be premature to conclude that SNe I
c unambiguously contain helium. Thus, we conclude that pure C + O cores are
still viable progenitors for SNe Ic.