3 Results

3.1 Line identifications

The spectra have been normalized using spectral parts free of emission and absorption lines. Identification of the spectral lines was done on the basis of a recent catalog of Coluzzi (1993), a compilation of Johansson (1978) for Fe II lines, and a paper by Quinet et al. (1996) for [Fe II] lines. A comparison of our data obtained with different resolutions in the same spectral intervals shows that they are very similar. Because of this, the data obtained with higher resolution and/or higher signal-to-noise ratio was used for identification.

In totaly 406 emission lines were identified in the spectral range between 4197 and 8864 Å. Their observed wavelengths, identifications, peak intensities, equivalent widths, and integrated fluxes (for the lines detected in the OHP 1997 spectrum) are presented in Table 1.

The 63 absorption lines detected in the spectrum seem to be of photospheric origin, they have not been observed previously. They are listed in Table 2. We found 60 absorption features (Table 3), which were identified with diffuse interstellar bands (Jenniskens & Désert 1994). All of them are rather strong confirming a heavy reddening already indicated by the color-indices. We did not attempt estimating E_B-V from their equivalent widths because of large scattering in corresponding relationships (e.g., Herbig 1975) and the fact that MWC 314 is one of the most reddened stars for which diffuse bands have been measured. At the same time, we have to note that the resolution of our spectra reveals many details of the diffuse features which can be used in their future statistical studies. Seven unidentified lines are listed in Table 4. Data on some photospheric lines of MWC 314 and several other supergiants are presented in Table 5. Radial velocities of the spectral lines were measured by matching their direct and mirrored profiles. Mean radial velocities determined for the lines of different elements are presented in Table 6. Parts of the whole spectrum obtained at different time are presented in Figs. 1-3.

  

Figure 1: The 1996 OHP spectrum of MWC 314 with identified lines. The wavelengths are given in Angströms, the intensity is normalized to the continuum level

  

Figure 2: The 1997 November SAO spectrum of MWC 314 with identified lines. Units of the wavelengths and intensities are the same as in Fig. 1

  

Figure 3: The 1997 OHP spectrum of MWC 314 with identified lines. Units of the wavelengths and intensities are the same as in Fig. 1

3.2 Characteristics of the spectral lines

Hydrogen. Three emission lines of the Balmer series were observed in our spectra, the Paschen lines are also seen in emission from n = 11 up to n = 37. The Balmer lines do not show any noticeable variability in comparison with the data obtained by Miroshnichenko (1996) in 1991. Even with the higher resolution spectra the H$\alpha$, H${\beta}$, and H${\gamma}$have no P Cyg-type absorptions. However, at a resolution better than 1 Å H$\alpha$and H${\gamma}$appear double-peaked with a separation of 28 and 42 km s^-1 respectively, which is near the resolution limit.

Helium. All the lines of neutral helium in our spectra are in emission and have a mean velocity of 21 $\pm$ 3 km s^-1, except for the strongest lines at 5876 and 7065 Å. The resolving power was not high enough to detect a possible double-peaked structure. It was suspected only for the 5876 Å line. The lines at 4921, 5015, 5876, 6678, and 7065 Å lines display shallow blue-shifted P Cyg-type absorptions ($\le$ 0.15 the continuum level) extended up to $\sim -300$ km s^-1 (Fig. 5). Very weak emission features are seen inside the absorption components. A very weak line of ionized helium at 4686 Å, which has not been reported before for MWC 314, is suspected to be present in the spectrum (Fig. 4).

  

Figure 4: Spectral parts containing the $\lambda$4686 He II line (the 1996 OHP spectrum, left) and the Fe III lines (Mults. 115 and 117, right). Solid line in the right part represents the spectrum of MWC 300 shown for comparison, dashed line represents a part of the November 1997 SAO spectrum of MWC 314. Units of the wavelengths and intensities are the same as in Fig. 1

  

Figure 5: A spectral part containing He I 5876 Å line and sodium D1,2 lines. The solid line represents the SAO spectrum obtained in July 1997, while the dashed line that of November 1997. Units of the wavelengths and intensities are the same as in Fig. 1

Iron. This is the element, which exhibits the greatest number of lines. Almost all permitted lines of Fe II show a double-peaked structure. The separation of the peaks and their intensity ratio depend on the line strength. For the weak lines ($I_{\rm peak} \leq$ 1.2 $I_{\rm cont}$) they are 70 km s^-1 and 0.65 ($I_{\rm blue}$/$I_{\rm red}$), while for the strong lines ($I_{\rm peak} \sim$ 1.5 $I_{\rm cont}$) 40 km s^-1 and 0.95 respectively (see also Sect. 4.1). 37 rather weak emission lines of [Fe II] were detected in comparison with only one line found by Miroshnichenko (1996). Eight permitted lines of doubly ionized iron and six [Fe III] lines were also identified.

Sodium. Only neutral sodium is present in the spectrum. Interstellar components of the D_1,2 lines of multiplet 1 are almost saturated indicating a very large reddening. Each of them consists of two components of almost equal intensity at $\sim$ +7 and +25 km s^-1. At the same time, both lines contain emission components of circumstellar nature. The only line of multiplet 4 was found in pure emission in the red part of the spectrum.

Titanium and Chromium. These elements are mainly found in the form of singly ionized permitted lines and are, in general, weaker than the Fe II lines. Many of them are blended with the Fe II lines, which prevents determination of their profile shapes. Only in several cases are they undoubtfully double-peaked. Several lines of [Ti II] are seen in the red region, while no [Cr II] lines were identified.

Magnesium. Two emission lines of Mg II were detected near the red edge of the spectrum. Neutral Mg is present in emission as two rather strong lines of Mult.2 at 5173 and 5184 Å, the unresolved triplet at 8806.7 Å, and the doublet at 6319 Å, which is blended with a line of Fe II at 6318 Å. The latter was also found in the spectra of B[e]-supergiants in the Magellanic Clouds (Zickgraf et al. 1986).

Calcium. Calcium is represented by the singly ionized infrared triplet in emission which is much stronger than the Paschen lines blended with it. Two relatively weak emission lines of [Ca II] are also present in the red region.

Oxygen. Only five emission lines of neutral oxygen were detected in the spectrum. The permitted doublet at 8447 Å is relatively strong ($I_{\rm peak}$ = 2.2 $I_{\rm cont}$) while the permitted triplet at 7775 Å is much weaker (1.2 $I_{\rm cont}$). The forbidden lines at 5577 and 6300 Å are extremely weak.

Silicon. Six lines of Si II (multiplets 2, 4, and 5) were found in emission. The line at 6370 Å is blended with a stronger Fe II double-peaked line, while the line at 6347 Å is probably double-peaked, what can be revealed at higher resolution (Fig. 6).

Nitrogen. Several weak permitted emission lines of neutral nitrogen were found longward of 8600 Å. Forbidden emission lines are located at 5755 Å (very strong line) and at both sides of H$\alpha$. A number of N II absorption lines as well as three of N III were detected between 4197 and 6523 Å, which seemed to be photospheric.

Photospheric lines. Besides those of nitrogen, the photospheric lines of S II (26 lines), Ne I (8), C II (2), Al III (2), P II (1), and Si III (1) were detected. They have the same velocities as the stellar N II absorptions.

Other elements. Emission lines of singly ionized Mn (8 lines), V (4), Sc (3), Ba (1) were also found in the spectrum.

In our spectra we found almost all the emission lines reported by Swensson (1942). However, some of them might have different identifications than those suggested by Swensson. For example, a double-peaked emission centered at 4657.9 Å is more likely due to a Ti II transition than to that of [Fe III] 4658.1 Å, because forbidden lines have single-peaked profiles in our spectra and positive radial velocities. Similarly, an emission at 5184 Å is double-peaked and is more likely to be a Ti II line of multiplet 86 rather than that of Mg I 5183.6 Å. The presence of the Mg II line in a blend at 4481-4482 Å, which probably consists of a Ti II (30) and a Fe II (20) line appearing in other blends as well, is uncertain.

Nearly 20 weak emission features (mainly with $I/I_{\rm cont} <$ 1.05) can be identified with Fe I lines of multiplets 15, 60, 169. Some of them have double-peaked profiles and radial velocities similar to those of the permitted singly ionized metallic lines. Although this is not unexpected, we consider these identifications as doubtful because some strong lines of the mentioned multiplets are not seen in the spectrum (for example, Fe I (169) $\lambda$ 6252.6 Å or Fe I (60) $\lambda$ 8688.6 Å).

3.3 Variations of the line intensities and profiles

Variations of the spectral line intensities on a timescale of weeks were reported by Swensson (1942), who was going to analyse them later. However, this discussion was not published. Our observations show no significant variations of the line intensities or profiles, except for those in some very weak lines, which can be due to noise.

The profiles of the majority of the emission lines are very narrow ($\Delta v \sim$ 130 km s^-1 FWHM), which make their appearance dependent on the spectral resolution. However, comparison of our spectra obtained at different times but with the same resolution shows no significant changes (Fig. 6). The broad lines, like H$\alpha$(Fig. 7), reveal their fine structure at higher resolution but appear stable even on more extended period of time. Widths of some emission lines reported by Swensson (1942) are almost the same as those in our spectra. This fact along with the absence of significant photometric variability (Miroshnichenko 1996) allow to suggest a relative stability of the object's wind on a timescale of decades.

  

Figure 6: Comparison of the observational data obtained at different time and with different spectral resolution. Solid line represents the July SAO spectrum, the OHP 1996 spectrum is shown by dashed line and the OHP 1997 spectrum by dot-dashed line. Units of the wavelengths and intensities are the same as in Fig. 1

  

Figure 7: Section of the spectrum around H$\alpha$. The SAO 1997 spectrum is shown by solid line, the spectrum obtained by Miroshnichenko (1996) in 1991 at SAO by dashed line. Intensity is normalized to the continuum level