The resulting ratios in our program stars are given in the fourth and eighth columns of Table 3. The histogram of the ratios is shown in Fig. 6, and it shows a rather broad peak from 1 to 6. The average of the ratios is (standard deviation). Two exceptions in our sample are VX And and HD 52432, both of which have a ratio of 12. It should also be noted that some of the stars have ratios less than 34, the value expected at the equilibrium of the CN-cycle. Especially, Lee 99, HD 75021, RY Dra, and CG Vul have extremely low ratios less than 2. Though the uncertainties of the ratios are relatively large for some of those stars, such low ratios cannot be explained by the operation of the CN-cycle in equilibrium.
We now turn our attention to the silicate carbon stars. Willems & de Jong (1986) proposed that all silicate carbon stars might have enhanced 13C abundances, noting that at least two of the silicate carbon stars they identified are J-type. Later, Lloyd-Evans (1990) identified seven stars which show the 9.8 µm silicate emission feature and the photospheric spectra of carbon stars. He showed that five of the seven stars are J-type and the classification of the others remained to be further examined. Lambert et al. (1990) identified these two questionable stars as J-type, based on the spectroscopic observation at the K-band. We confirm these previous identifications more quantitatively. The ratios are for GCCCS 447, for BM Gem, for NC 83, for V 778 Cyg, and for EU And. These results are perfectly consistent with their previous identifications as J-type. Moreover, it is worth noting that the five silicate carbon stars have the ratios which are the most common in our sample. In other words, these stars exhibit no peculiar ratios which would be associated with the presence of the silicate emission feature.
Our sample includes seven stars previously analyzed by other authors. Table 4 shows a comparison with the previous results. Y CVn and RX Peg were analyzed by Fujita et al. (1969) and Fujita & Tsuji (1977), and their results are for Y CVn and for RX Peg. They determined the excitation temperatures so that the lines due to different bands should form as smooth a curve-of-growth as possible, and the uncertainties of the ratios are reportedly about a factor of 2. Thus, given the accuracies of the results and the difference about how to determine excitation temperatures, our results should be preferred to theirs. Our sample also includes three stars analyzed by Lambert et al. (1986) (Y CVn, RY Dra, and VX And). The ratios derived by both authors show fair agreement, though our result tends to be somewhat smaller than theirs, except for the value for VX And determined from the CO lines. Abia & Isern (1996, 1997) derived ratios in 11 J-type carbon stars from CN lines in almost the same wavelength region as we observed, but using the spectral synthesis method. Six stars on the list of Abia & Isern (1997) are included in our sample. The agreement is rather fair, except for UV Cam and BM Gem. Our results are for UV Cam and for BM Gem, while theirs are 6 and 9, respectively. The reason for the disagreement remains to be further investigated. Lambert et al. (1990) estimated that the ratios of EU And, V 778 Cyg, NC 83, and BM Gem might be similar to that of VX And ( = 13 by Lambert et al. 1986and by the present work). But their estimates are based only on the mean equivalent widths of 12CN and 13CN lines located around 2 µm, while our analysis is done on a line-by-line basis, using the model atmospheres in order to take into account the correction for the excitation effect. Therefore, our quantitative analysis finally confirms these silicate carbon stars as J-type.
Table 4. Comparison with the results derived by the previous authors
This agreement of ratios among the authors shows a marked contrast to the case of N-type carbon stars discussed in Paper I, where our results of ratios are by a factor of 2 to 3 smaller than those derived by Lambert et al. (1986). We pointed out in Paper I that the difference of the model atmospheres used in the analyses might be the reason for the disagreement. However, as we have demonstrated in Paper II, it cannot explain the disagreement. In the case of J-type carbon stars, the enormous strength of 13CN lines might help us measure equivalent widths or central depths relatively accurately both in their analysis and in ours. It might minimize the effect of blending, whether ratios are determined by the use of equivalent widths or central depths, while it might have an effect on the analyses of the spectra of N-type carbon stars to some extent. Moreover, thanks to the 13CN lines as strong as the 12CN lines, the ratios can be determined from lines with almost the same excitation potentials in the case of J-type carbon stars. The iso-intensity method can best be applied to the case where the line intensities are truly the same. It is also easy for any other method to interpret the lines with similar intensities in terms of abundance ratios. This might be one of the reasons for the agreement of the resulting ratios derived by both authors.
© European Southern Observatory (ESO) 1999
Online publication: April 12, 1999