Although we can foresee a time when there will be more SNe Ia at redshifts above 0.3 than nearby ones, we will have to learn from the nearby samples with their superior data coverage and quality. The distant supernovae are still observed rather sparsely and lack the wavelength and spectral coverage we can obtain for local events.
The capabilities for detailed supernova studies have increased continuously over the past decade and detailed spectroscopic and photometric data sets will become available at a rapid rate. This will allow us to address very specific questions and focus on model predictions. However, simple model predictions have been lacking so far and the complications in the explosion models and the radiation transport have proven to be veritable road blocks.
With the extensive and homogeneous data sets which have become accessible in the last few years the general discussion of global parameters of SNe Ia is possible. The detailed statistics of luminosity, rise times, decline rates, and spectral line evolutions have made the systematic investigations of supernova energetics, explosive nucleosynthesis, and more detailed inferences on progenitors a possibility. The increased and coordinated access to telescopes of all sizes has brought the field forward substantially. With the statistics on global supernova parameters the tedious comparison with explosion models has been supplemented.
In the following an attempt is made to assemble the available information from the observations to point out future directions of SN Ia research.
The differences of the various luminosity corrections and absorption determinations (see section 3.1) are very disconcerting. They clearly are not due to evolution, but will likely be traced to technicalities of the fits. The most obvious culprit is the degeneracy of reddening and intrinsic color of SNe Ia, which has to be lifted for a reliable measurement. The discrepancies apparent in Fig. 4 and discussed in section 3.1 are most likely due to this degeneracy. The influences on the cosmological conclusions drawn from SNe Ia will have to be investigated in much more detail and the discussion has already started (e.g. Drell et al. 1999). The different corrections also play an important role in the exact determinations of the Hubble Constant (Suntzeff et al. 1999, Jha et al. 1999, Saha et al. 1999, Gibson et al. 2000) and are responsible for the remaining discrepancies.
Despite the technicalities of the light curve fitting the variations among SNe Ia are real and have to be explained. It is very interesting to consider the invariants among SNe Ia. Each of them tells a different part of the overall story and should help in piecing it together.
With the brighter SNe seemingly rising and declining more slowly this implies that the energy release is retarded throughout the maximum phase. The width of the peak phase in the bolometric light curve also correlates with the peak luminosity (Contardo et al. 2000) and so does the occurrence of the shoulder (Contardo 2000). The more luminous objects are so through the whole known evolution, i.e. they are emitting more energy than the fainter supernovae. It is also striking to see that the very luminous SNe Ia show the characteristic Si II line appear relatively late (typically only after maximum light).
The color at maximum is a measure of the opacities in the ejecta. Since these are dominated mostly by lines and not continuum processes, the colors are not a direct indicator of the temperature in the ejecta (Pinto & Eastman 2000). Interestingly, the B - V color is very well defined and depends very little on the light curve shape whereas V - I shows a stronger dependence on the decline parameter (Phillips et al. 1999).
The velocities of the ejecta can be measured from the emission lines several months past the explosion (Mazzali et al. 1998). The fact, that they correlate with the light curve decline is remarkable. The ejecta velocities derived from the nebular lines are an indicator of the ratio of kinetic energy and total mass. Strictly speaking, this only applies exactly for spherically symmetric explosions, but at the late phases any asymmetry should be damped out. Since the decline correlates with the maximum luminosity and this in turn is connected to the Ni mass synthesized in the explosion (Arnett 1982, Arnett et al. 1985, Branch 1992) we have a direct connection of the explosion strength and a product of the power source of the supernova emission. More luminous SNe Ia also are more powerful explosions. The ejecta mass of these powerful explosions must be higher as well, as the slower release of the energy can only be achieved by a larger column density. This immediately rules out a single mass progenitor system.
The host galaxy morphology and the galaxy color provide information on the star formation of the parent population. SNe Ia in elliptical galaxies are on average less luminous than their counterparts in late spirals (Filippenko 1989, Hamuy et al. 1995, Schmidt et al. 1998). Also SNe Ia in bluer galaxies seem to be brighter (Branch et al. 1996). All of the most luminous objects (like SN 1991T) occurred in spiral galaxies, and most of them suffer reddening in the host galaxy and are possibly loosely connected to star forming regions or spiral arms (Bartunov et al. 1994). It thus appears as if the more luminous objects are connected to a younger parent population and the fainter SNe Ia come from old progenitor systems. Yet, the correction from the decline rates applies to all SNe Ia independent of the host galaxy morphology (Schmidt et al. 1998). Thus, although the parent population may be different, most likely due to age differences, most, if not all, SNe Ia come from the same type of progenitor system. It has been proposed that the explosion energy depends on the precursor composition and could be observed from samples which span sufficiently long look back times, i.e. shorter progenitor life times (Höflich et al. 1998b, Kobayashi et al. 1998, Umeda et al. 1999). Such experiments will require better and more extended data than are currently available.
The question of what is a normal SN Ia has been raised many times in the last few years (e.g. Branch et al. 1993). Selections based on color or spectra have been proposed and used. To what extent such subclassifications describe physical differences is, however, unclear.
The extreme cases of SN 1991bg and SN 1997cn can not yet be accommodated within the simple schemes proposed. Could it be that they emerge from different mechanisms? The answer is still outstanding and we need more objects of this sort to investigate.
5.2. Nickel masses
The nickel mass of individual supernovae differs by up to factors of several (section 3.4). With such large differences, it is clear that the explosions are not as uniform as assumed only a decade ago. It will be an important task for the next years to determine whether these differences are due to different explosion mechanisms (e.g. double detonations, deflagrations, etc.) or are variations of a single mechanism (e.g. the density at transition from deflagration to detonation (Höflich et al. 1996)). The distribution of nickel masses may provide a first indication of what the exact distribution of the explosions is. It has been claimed that most SNe Ia emerge from a fairly narrow range of luminosities (and hence nickel masses), but the numbers are still small. The distribution of the decline parameters has not yet yielded a clear picture (e.g. Drell et al. 1999).
It is of course also possible that we are observing two or more explosion mechanisms, each with variations on its own. A single explosion mechanism which produces differences of a factor of 10 as observed in the nickel masses (Table 1) has not been proposed yet.
5.3. Future developments
Three major questions about SNe Ia will have to be solved: the influence of reddening, the progenitor systems and the explosion mechanism. Observationally, reddening should be the easiest to either measure or avoid. Many interesting questions can be addressed with a statistically significant near-IR sample. The luminosity corrections are smaller in the I band (Phillips et al. 1999), at late times the near-IR is the prefered region for the determination of the Ni mass (Spyromilio et al. 1992, Bowers et al. 1997), and the near-IR Hubble diagram will also provide a reddening free determination of the Hubble constant. The reddening law in external galaxies will be another topic which could be addressed by SNe Ia observed in the optical and the near-IR.
Signatures of SN Ia progenitors should be discovered soon. Either signs of the companion transfer to the white dwarf are detected or possible progenitor systems can be ruled out on statistical grounds. Neither has so far been the case. Dedicated programs for the search of possible progenitor systems are needed.
The constraints on the progenitor models will have to be increased and improved. The fact that we do not know whether some SNe Ia come from sub-Chandrasekhar or Chandrasekhar mass explosions is embarrassing. In fact, not even the relative ejecta masses are known. If the above argumentation is correct (section 5.1), then we have a first sign of explosions with different masses.
The first direct detection of the -rays from the nuclear decay of 56Ni and 56Co would be a major success. Such a detection is within reach of the current instruments on CGRO or INTEGRAL (Höflich et al. 1998a, Georgii et al. 2000). The combination of the -ray detection with the observed (UVOIR) bolometric light curve will be a powerful tool to measure the escape fraction and hence the energy release in SNe Ia.
Statistical studies of the bolometric luminosity of SNe Ia will further delineate their true luminosity and hence nickel mass distribution. With such information it will become possible to decide what the major stellar evolution channels for SNe Ia are. We are entering an interesting phase, where searches will be volume limited even for faint SNe Ia and 'fair' samples can be established.
The future is bright for SN Ia research. The extension to high-redshift searches and the inherent possibility to probe SNe Ia over significant look back times offers the opportunity to follow a specific tracer of individual stars over a large fraction of the age of the Universe. This addition to the current supernova research will tell us about the history of stars beyond the cosmological implications championed so far.
Parts of this review have been written during visits at the Astronomical Institute in Basel and the Stockholm Observatory. I would like to thank A. G. Tammann and C. Fransson for their hospitality. Many discussions with M. Phillips, N. Suntzeff, B. Schmidt, R. Kirshner, A. Riess, P. Meikle, K. Nomoto, W. Hillebrandt, and A. Filippenko are acknowledged. Special thanks go to G. Contardo for letting me show some of the results of her PhD thesis and Jason Spyromilio for continuous critical conversations on supernovae.