A. Type II Take SN 1979C, a type II-L , for example (Branch et al. 1981), the spectrum
is nearly a continuum near maximum, with weak ones of hydrogen and neutral
helium (figure 4). After maximum light, lines of Ca II, Na I, Fe II, Ti II
and Sc II
appear in the form of resonant scattering profile (Patchett & Branch
1972; Branch 1981), while hydrogen lines, with net emission, are not well
represented by resonant scattering. H 
becomes prominent along with
multiple features of Fe II a month after maximum, but has no blue
absorption. This may be the result of large emissivity in the dense wind
which, from existence of radio emission (Fransson 1986), have surrounded
this supernova. Six months later, the lines of [O I] 
becomes observable, indicating
perhaps, the exposure of oxygen-rich core material.
Figure 4: The development in time of spectra of type II-L SN 1979C (Branch
et al. 1981). The spectra shows the successive emergence of H , Fe II lines
after maximum, and [O II] at onset of the nebular phase.
There is insufficient data to determine whether there's systematic difference
between SN II-L and SN II-P. SN 1986I, as a Type II-P, showed qualitatively
the same spectral features except for a blue-shifted P-Cygni absorption
component for H .
B. SN 1987A Figure 5 shows spectra of SN 1987A at three typical epoches: the first stage is 1.85 days after explosion when the supernova is in its photospheric phase, and the spectral lines are formed with pure scattering source function showing P-Cygni profiles, as is a characteristic feature of comoving atmospheres; the second stage is 50 days after explosion when a transition from photospheric phase to nebular phase is occurring. Emission and absorption must be considered in the source function along with scattering, because emission lines and residual blackbody continuum exist together. The third epoch is nebular phase when emissions dominate the emergent spectrum. The fact that Balmer series is well developed as early as t=1.85 days is one thing that makes SN 1987A prominent. Another unique feature is that SN 1987A developed distinct P-Cygni behavior in late-time spectra. The modeling by Swartz et al. (1993) in the effects of composition mixing and the placement of photosphere suggests that O and Ca in hydrogen-rich envelope is enough for observed emission line strength and no mixing from inner layers is needed for Balmer lines.
Figure 5: Spectra of Type II SN 1987A in the Large Magellanic Cloud,
obtained from the photospheric phase to the nebular phase.
C. Type Ia The spectrum of Type Ia supernovae, such as SN 1981B (Branch et al. 1983)
and SN 1992A (Kirshner et al. 1993) contains resonance scattering lines
of Ca II, Si II, S II, Mg II and O I, and perhaps Co II at the blue
end of the spectrum, near maximum, and is dominated by blends of permitted
Fe II lines afterwards for as long as 100 days (Branch et al. 1983;
modeling; Branch et al. 1985; Harkness 1991). The case of SN 1992A is shown
in figure 6.
The spectrum in nebular phase consists of forbidden emission lines from
first several ionization states of iron, such as [Fe II] and [Fe III], as
well as [Co III] 
5894 multiplet. Ca II 3945 multiplet
absorption and Ca II 8579 multiplet absorptions are also
present. It is not well known whether [Co III] 5894 multiplet
is blended with persistent Na I lines in nebular phase, which, though
usually occurs hundreds of days after explosion, is not clearly defined.
Departure from LTE with source function 
times its LTE
strength is suggested for [Co III] 5894 multiplet in
synthesizing a semi-nebular spectrum of SN 1992A (Kirshner et al. 1993).
Figure 6: Early spectral evolution of type Ia SN 1992A. The spectra are
labeled by epoch relative to maximum light. From Kirshner et al. (1993).
D. Type Ib The photospheric spectra of Type Ib SNe 1983N and 1984L show strong lines
of He I ( 4471, 5015, 5876, 6678 and 7065), along with other
spectral features attributed to C II, O I, Ca II and Fe II. Harkness et al.\
(1987) found that a large non-LTE overpopulation of the excited states
of He I were needed to fit the spectra. It is suggested that the huge
non-LTE excitation of He I could be the result of high-energy photons
from nickel-cobalt decay (Harkness et al. 1987). Figure 7 shows the
spectral evolution of typical Ib SN 1984L in NGC 0991.
Figure 7: The spectral evolution of the prototypical SN Ib 1984L.
The strong He I features are apparent two weeks after maximum. The appearance
of oxygen forbidden line in the last spectrum labels the onset of the
nebular phase.
A spectrum in nebular phase, as shown in figure 8 for SN 1985f,
may contain [O I] 5577 (Harkness et al. 1987) earlier,
and develops strong broad emission lines of [O I] 
6300, 6364, [Ca II] 7291, 7324, Na I
5890, 5896, and [Mg I] 4562, [C I]
8727 later (say, 8
months of SN 1983N and 1985f; Filippenko & Sargent 1986; Gaskell et al.
1986).
Spectral features in nebular phase had induced conclusions by Fransson &
Chevalier (1989) that approved a massive ( 
M 
) core collapse
mechanism rather than exploding white dwarfs for Type Ib. The appearance
of Type IIb supernovae connects SN Ib and SN II more closely, though
the explosion mechanism for SN Ib remains a problem.
Typical spectra for both Type Ia and Ib are shown in figure 2, in which
the characteristical lines described above are easily recognized.
Figure 8: A nebular spectrum of type Ib SN 1985f is shown. From
Filippenko & Sargent (1986).
E. Type Ic SNe Ic may be named "helium poor Type Ib" because they never show the
strong absorption of He I 5876 at early times. With absence of strong 6150Å\
feature, the spectra of SN Ic are certain only in Ca II, Fe II and O I lines.
It is of much importance that small minimum at about 3200Å in early
spectra of SN 1987M may result from Co II in atmosphere models of SN Ia
events. Data on SN 1987M supports that SNe Ic are closely related to SN Ib
events despite the distinct spectral differences in the early time spectra.
By comparing SN 1994I and 1987M, both type Ic's, Clocchiatti et al. (1996)
suggested that type Ic supernovae display He I lines in optical region of
spectra, and that there exists high velocity helium in outer portion of
ejecta. Spectral synthesis by Zhang et al. (1995) and Baron et al. (1996)
show supports to Nomoto et al. (1994) model of C+O star as its origin, with
H and most of the helium layers are blown away.
Filippenko (1988) and Filippenko et al. (1990) have supposed
that extremely weak H appears to be
present in SN 1987M, the most completely studied Type Ic, as well as
SN Ib 1983I and 1988L, near their maximum light, but that this possible
presence of weak H must be verified. The mass of hydrogen envelope is a
main variable in these objects.
Further evidence of weak H in SN 1987M was provide by Filippenko (1992)
with comparison of early-time optical spectra of Type II SN 1985L and
SN 1987M. The presence of H in Type Ic SN 1991A and 1990aa was also
suggested. This phenomenon is not widely accepted, but if true, would again
imply some generic connection between SN Ib/c and SN II.
A more definite link is provided by Type IIb supernovae (SN 1993J and SN 1987K) which have spectral characteristics of Type II near maximum brightness and those of Type Ib long past maximum. More detailed description of SN IIb (i.e. SN 1993J) will be in next section.