The most important results of the STNEI (this work) and CIE (Paper I) spatially resolved spectral analysis are: 1) the one temperature models fail to fit the data as a whole; 2) in those cases where the 1T STNEI model is appropriate the resulting ionization time is unexpectedly low; and 3) a 2T CIE model is usually adequate.
The first and third point were already reported in Paper I, where we
analyzed the same data set using
CIE models. Indeed, Itoh (1979) suggested that the X-ray
spectrum of a hot plasma in NEI condition could be approximated by a
two temperature CIE plasma model.
Therefore, on the basis of the
results reported in Paper I, a single component nature of the X-ray
emission could not be definitely excluded.
As we have seen in the
previous section, when considering the new results of the STNEI spectral
analysis, we find that the 1T STNEI model fails in the same location
where the 2T CIE model was appropriate (for instance, consider the
d3 bin and zone C in Paper I, or b6 and zone H in Paper I).
However, this does not necessarily imply that the Itoh
suggestion is wrong in our case:
the NEI model used by Itoh for
synthesizing the X-ray SNR emission was different from our STNEI model,
because it was a sum of Sedov shells with different 
values.
Nevertheless, we interpret our results as a growing
evidence that the medium in not homogeneous. If the medium was really
homogeneous and the Sedov solution applied, then
the peak in the NEI plasma emissivity (Fig. 1) and the peak
in the Sedov density profile would yield a spectrum very similar to a
STNEI spectrum, given the moderate PSPC spectral resolution.
We stress here that, in spite of the high counting statistics of our spectra, these results are not affected by uncertainties in the detector response, since they have been studied (Appendix A of Paper I) and accounted for by inclusion of systematic errors in the analysis.