Astron. Astrophys. 331, 815-820 (1998)

4. Discussion

By introducing a color correction, in addition to the usual shape correction , we have obtained a very substantial improvement in the fit to 29 distant supernovae. The best-fit parameters are R = 2.09 0.38 and b = 0.52 0.15, to be compared to b 0.8 found previously for R = 0.

The preliminary value for the Hubble constant remains H₀ = 60, pending further investigation of the effect of the color correction on the mean absolute magnitude of the Cepheid-calibrated SNe. The uncertainty in H₀ comes from many sources but is dominated by the still-uncertain distance scale and metallicity dependence for the Cepheid calibration. Following Branch et al. (1996), we continue to assign 0.15 mag uncertainty to the Cepheid zero point. This contributes 4 to H₀. The statistical uncertainty coming from the 29 Cal n/Tololo supernovae along with the uncertainty in b, R, and q₀ combined in quadrature give . Apart from the Cepheid calibration, we assign, for the time being, another mag uncertainty to the mean absolute magnitude of the seven calibrated SNe. Adding all these in quadrature yields an overall uncertainty H .

The best fit is much too good, having a reduced , and a confidence level CL = 0.98. This suggests that the measurement uncertainties of the Cal n/Tololo supernovae may have been overestimated. Alternatively, as discussed previously (Tripp, 1997), there may be correlations between the errors , and now that are not calculated by the observers and which, if introduced, could reduce the overall uncertainty. In any case, reducing these uncertainties by 20% produces a more reasonable CL = 0.9, while a scale factor of 0.55 yields the most likely CL 0.5. If errors are indeed smaller than presently believed and if cosmological SNe Ia measurement uncertainties are found to be similarly overestimated, then smaller errors will lead to a smaller uncertainty in q₀. Also, by applying the same type of color correction to cosmological supernovae even without knowing whether reddening is intrinsic or due to dust, one will be able to completely standardize the light output of each explosion and thereby get substantially better values for q₀ and the mass density of the universe.

That there should be a color dependence to the SNe Ia absolute magnitude is not surprising: Höflich & Khokhlov (1996), employing a range of acceptable carbon-oxygen white dwarf models of SNe Ia explosions, have demonstrated, assuming the correctness of the models, that the absolute magnitude should depend both on and on . These models are compared with data from 40 SNe Ia in Höflich et al. (1996). According to van den Bergh (1995), one expects from these models that the color dependence should yield an R value of about 4. By coincidence, this is quite similar to the dependence of reddening due to interstellar dust as measured within our Galaxy and presumed to be true in other regions of space as well. This reddening dependence as expected from dust outside of the Galaxy has recently been verified by Riess et al. (1996) using a multi-color analysis of extragalactic SNe Ia. Since our analysis combines both of these quite different color mechanisms into one phenomenological parameter R, we would have expected to find R 4; instead we find R to be about 2. This puzzle has beset the study of nearby SNe Ia for many years and has been discussed at length in the review by Branch & Tammann (1992) of Type Ia supernovae as standard candles. Here it has resurfaced in an even more compelling form, coming as it now does from the high quality Cal n/Tololo distant supernova data and in addition appearing to be in conflict with some of the favored theoretical models of Type Ia supernova explosions.

Note added in Proof: In a multi-color analysis of SNeIa, Riess et al. (1996, ApJ 473, 88) use a parametrization which implicitly assumes a tight correlation between B - V and in the approximate form B- V = 3.5 . This is very different from that observed in the data displayed in Fig. 1. They obtain a value of corresponding to an , the value to be compared to our R = 2.09.

© European Southern Observatory (ESO) 1998

Online publication: March 3, 1998
helpdesk.link@springer.de