4 Color diagrams

When available, the spectral types attributed to the WN stars in all subsequent figures come from Smith et al. (1996) and those of the WC stars come from Koesterke & Hamann (1995). Otherwise, for the galactic stars, they come from van der Hucht et al. (1981) with additions or revisions by Massey & Conti (1983b) while for the LMC stars, they come from Breysacher (1981), with additions or revisions by Massey & Conti (1983c). The reason of these choices is to include the weak-strong line distinction in our classification scheme.

The total data set consists of the results of the synthetic photometry performed on the spectra discussed above and of the results of the observing runs. In order to keep consistancy between Sects. 4 and 5, synthetic data were prefered to the observed ones when both were available (even though the precision is higher on the photometric data).

4.1 Calibration of the synthetic photometry

In the whole data set, there are 25 WR stars with both observed and synthetic photometry. Moreover, two of the observed non - WR stars make part of the Hamuy et al. (1992) catalog of spectrophotometric standard stars (LTT 7379 and HR 7596). The calibration of the synthetic photometry was established on the basis of these 27 stars. A first iteration gave a mean zero point for each filter and deviations for every star. The most deviant stars (2 $\sigma$) were excluded and a second iteration provided the accepted zero points. Standard deviations are $\approx$ 0.1 mag in all filters which is compatible with the accuracy claimed in TM (Table 2).

   Table 2: Precision of the zero points of the synthetic photometry. s is the standard deviation of the mean and n the number of stars. We do not list the actual values which are unit dependent

4.2 The color indices

Color diagrams based on very simple color indices, e.g., (c₂-$r_{\mbox{\tiny C IV}}$) vs. (c₁-$r_{\mbox{\tiny He II}}$), are able to separate WN stars from WC stars, but they do not discriminate them perfectly from non - WR stars, i.e. the constant stars observed to calibrate the photometric parameters of our system and the stars taken from JHC. To achieve a better separation, color indices are needed in which the continuum flux below the emission lines is more correctly taken into account.

Morris et al. (1993) indicate that, in the relevant wavelength domain, the continuum of WR stars can be approximated by a straight line in a $(\log{F_{\lambda}}$ vs. $\log{\lambda})$ diagram. On the other hand, in the same wavelength domain, the interstellar reddening can be approximated by a straight line in a $(\log{F_{\lambda}}$ vs. $\lambda)$ diagram. In both cases, the continuum under each line can be evaluated by a linear combination of the c₁ and c₂ magnitudes. The difference between the two approximations appears only in the values of the parameters included in the linear combinations. Nevertheless, simulations indicate that these differences are quite small because the wavelength domain concerned is rather limited ($\approx 1400$ Å).

  

Figure 1: WN-WC separation. The symbols are explained in Table 3. The big circle around the origin encloses most of the non - WR stars

In this paper, the continuum magnitudes under the different lines were computed in the framework of the second approximation here above $(\log{F_{\lambda}}$ vs. $\lambda)$, i.e. with the following relations:

${\rm cont}_{\mbox{\tiny He II}}$ = c₁ + 0.373(c₁-c₂)

${\rm cont}_{\mbox{\tiny C IV}}$ = c₂ + 0.2475(c₁-c₂)

${\rm cont}_{\mbox{\tiny He I}}$ = c₂ + 0.177(c₁-c₂).

The indices linked to the normalized intensities of the lines are defined as:

l( He II) = ${\rm cont}_{\mbox{\tiny He II}}$ - $r_{\mbox{\tiny He II}}$

l( C IV) = ${\rm cont}_{\mbox{\tiny C IV}}$ - $r_{\mbox{\tiny C IV}}$

l( He I) = ${\rm cont}_{\mbox{\tiny He I}}$ - $r_{\mbox{\tiny He I}}$.

4.3 Separation between non - WR stars, WN and WC stars

This separation is easily achieved by a plot of l( C IV) vs. l( He II), as shown in Fig. 1. The meaning of the various symbols is given in Table 3. Figure 1 shows a neat separation between the WC stars and WN stars, which are distributed along two distinct branches. Some kind of rough classification is operated along both branches and the WN/WC stars lie either in the WC branch or between both branches. It is interesting to note the unusually large range of the color indices compared to other more conventional systems, e.g., UBV.

  

Table 3: Symbols in the figures

In Fig. 1, as in every subsequent ones, the normal stars cluster near the origin because they have no strong features in the filter passbands. Incidentally, this implies that binaries containing a WR star and a normal star are expected to be closer to the origin than the single WR stars. The only non - WR stars that do not really belong to that cluster of points are K and M stars and supergiants Of. However, nearly all of them do appear in regions where they cannot be mismatched with WR stars (too close to the cluster of other non - WR stars or even not in the same quadrant as the WR stars). The only non - WR star of the simulations that seems hard to disentangle from the WN stars is a dwarf M1 (located at l( He II) $\approx$ 0.5, l( C IV) $\approx$ 0.2). A giant K4 (at l( He II) $\approx$ 0.1, l( C IV) $\approx$ 0.2) is also badly placed in Fig. 1 but we will see that it can be rejected on the basis of Fig. 2 because all non - WR stars (the M1V excepted) lie at l( He I) < 0.1 and l( He II) < 0.1. The measurable positive value of the l( He II) index in the case of the M1V star finds its origin in a molecular absorption band around our c₁ filter which depletes the reference continuum used to evaluate the He II $\lambda$ 4686 emission.

  

Figure 2: Intra-WN separation. A downward arrow indicates a binary. A plus sign above another symbol indicates a "+ abs'' star. The other symbols are explained in Table 3

4.4 Intra-WN separations

Figure 2 shows a plot of l( He I) vs. l( He II), with the sample limited to the WN stars as selected from Fig. 1 and the synthetic photometry on non - WR stars. In this figure, the binaries and the WN+abs are plotted with the symbol corresponding to the spectral type of their WR member, except that the former are identified by a downward arrow and the latter by a plus sign.

  

Figure 3: WN8-9 separation based on line ratios. A downward arrow indicates a binary. A plus sign above another symbol indicates a "+ abs'' star. The other symbols are explained in Table 3

This graph shows a rather clear separation between the WNEs stars, the WNEw stars, and the WNL stars. Among the latter, there is also a tendency to have a segregation of the WN7 stars and the WN8-9 stars in two separate regions, with an intermediate zone. As expected, the binaries are clustering around the origin and the horizontal axis of the plot. Indeed, almost all of the stars nearest to the origin are known binaries or exhibit absorption lines in their spectra. The only exception is Brey 58 (diamond shaped point at l( He II) $\approx$ 0.6). This star has a rather uncertain classification: WN5-6 or Of according respectively to Breysacher (1981) or Smith et al. (1996). Some WN9 stars also lie close to the non - WR star region which is normal as most of these objects are considered as closely related to the LBV stars, i.e., to be representative of a transition stage between Of and WN stars.

It may be interesting to mention that, when plotted on this diagram, the WN/WC stars place themselves according to their WN features, except WR 26 (WN7s/WCE, at l( He II) $\approx$ 2.4 and l( He I) $\approx$ 1.1) which exhibits strong WC characteristics (Fig. 1) and Brey 72 which is not a single object (Breysacher 1981). We get:

• WR 98 (WN8w/WC7) in the WN8-9 stars
• WR 8 (WN7w/WCE) in the WN7 stars
• WR 58 (WN4s/WCE) in the strong-lined WNE stars.

A better separation between the WN8 stars and the WN7 stars is achieved on the color diagram in Fig. 3 which uses more complex color indices. On this graph, we no longer have an intermediate strip where the WN8-9 stars and the WN7 stars are mixed. WR 12, recently reclassified as a WN8 (Rauw et al. 1996; Eenens et al. 1996; Smith et al. 1996), is lying well within the WN8-9 part of the plot. The only WN7 star lying in this same region is WR 82 that Smith et al. reclassify as a weak-lined WN7 star but which was previously known as a WN8 star.

On this diagram, all the WNEw stars lying at l( He II) - l( He I) < 0.9 are binaries, except Brey 60, classified as peculiar WN3 (Smith et al. 1996) and WR 28 and Brey 47, which have recently been subject to a reclassification by Smith et al. (1996).

  

Figure 4: Intra-WC separation. A downward arrow indicates a binary. An upward arrow indicates a weak lined WC star. A plus sign above another symbol indicates a "+ abs'' star. The other symbols are explained in Table 3

4.5 Intra-WC separations

In Fig. 4, the l( He I) index is plotted as a function of l( C IV). Some groups can easily be identified: the WC9 stars are located in the upper left corner while the WCE stars lie in the upper right corner, though not far apart from the WC7 -  8 stars. The representative point of the WO (Brey 93 = Sand 2) shows up near the WCE region, but clearly apart from it.

The lower part of the diagram is occupied by the binaries, the 3 weak-lined WC stars of our sample (WR 39: WC6w; WR 86: WC7w and WR 50: WC6w+abs), plus WR 72, classified as peculiar WC4.

Concerning the WN/WC stars, the situation is reversed in comparison to Fig. 2 since only the strong-lined WR26 is located according to its WC features.