The spectroscopic information on BN Ori from the period before 1960 is scarce. The first spectral classification of A7, was made by Cannon (1931) for the HDE catalogue, from objective prism spectra on plates exposed between 1925 and 1931. This classification was later confirmed by Parenago (1933), Hoffmeister (1949) and Payne-Gaposhkina (1952). Herbig (1954) estimated a type A5- A7 from a low-dispersion slit spectrogram, but from an additional spectrum ( = 75, taken on Mar.6, 1955), he noted that the spectrum of BN Ori is peculiar and that it may be composite with types A5- A6 and mid-F9 (Herbig 1960a). In 1968 and 1969 Zajtseva (1971) took a number of blue-slit spectra ( = 140- 250) from which she classified the spectrum of BN Ori as F0. A microphotometer tracing in that paper shows absorption lines near 4525, 5163 and 5265Å (identified by her as FeI), CaIIK and MgII at 4481Å. From the slope of the continuum between 6000 and 4000Å a spectrophotometric temperature of 9700K was derived, which is consistent with spectral type A7- A9. Photoelectric spectral scans in the range 6600- 3200Å made in Jan'75 allowed Kolotilov & Zajtseva (1976) to determine the residual strengths of the H, H, H and CaIIK lines, which led the authors to a classification F2- F3. The only emission line in the visual spectrum of BN Ori is H. Between Jul'71 and Dec'75, 18 spectrograms of H ( = 20, = 1) were taken by Kolotilov & Zajtseva (1976). The line profile of H in these spectra is double-peaked with an absorption trough at -50. Its blue emission-component varies strongly and has been observed to change into an absorption-component (with maximum velocity -600) within 24
equivalent widths (Å), full widths at half maximum ( ) and source ( ) of the lines in the red part of the visual spectra of BN Ori, HR 5999 and FU Ori. See Table 10 (click here) for the description of the used symbols
Table 6: Identification, equivalent widths (Å), full widths at half maximum ( ) and source ( ) of the lines in the red pa rt of the visual spectra of BN Ori, HR 5999 and FU Ori. See Table 10 (click here) for the description of the used symbols
Figure 3: Part of the low-dispersion blue spectrum of BN Ori, taken with the ESO 1.5m telescope in Dec'94
Since 1983 we have obtained a number of high-and low-resolution spectra of BN Ori. The observational details are listed in Table 5 (click here). We will compare the spectral information of BN Ori with that of FU Ori (Herbig 1966; Reipurth 1990) and that of the bright Herbig A5- 7III e star HR 5999 (Bessell & Eggen 1972; Tjin A Djie et al. 1989; Baade & Stahl 1989).
Two low-dispersion blue spectra have been taken: one in 1983 at Mt. Maidanak and one in 1994 at ESO (Fig. 3 (click here)). From the strength of the H, H, H and CaIIK lines and the presence of the G-band in the spectra we confirm the classification F2- F3 by Kolotilov & Zajtseva (1976). However, some additional lines, such as the strong CaI 4227Å line, may point to a later type, e.g. F3- F5. These blue spectra again show many absorption lines, among which the FeII multiplet 42, the G-band, the NaID doublet and the BaII (6497Å) and LiI (6708Å) lines which are also conspicious lines in FUOR spectra (Reipurth 1990). The EW of the LiI line is comparable (Table 6 (click here)) with that of LiI in the spectrum of FU Ori (Herbig 1965).
The red emission-component of H is stronger than the blue one, in contrast with the H-profile in the low-dispersion IDS spectrum of 1986, which shows roughly equal red and blue emission-components.
Two low-dispersion spectra in the far-red wavelength range have been secured by us: one in 1983 at Mt. Maidanak and one in 1992 at ESO. Figure 4 (click here) shows part of the spectrum obtained at ESO and Table 6 (click here) lists the EWs and FWs of the most important lines in this spectrum and in those of FU Ori (Shanin 1979) and HR 5999 (Tjin A Djie et al. 1989). Figure 4 (click here) shows the CaII (2)IR-triplet (8498, 8542 and 8662Å) in absorption, which is blended with the Paschen absorption lines P16, P15 and P13, respectively. We have used the method of Bychkov et al. (1978) to make a spectral classification from the EWs of the blends of the CaII (2)IR-triplet and the EW of the unblended P14. In the assumption that none of
Figure 4: Part of the low-dispersion red spectrum of BN Ori, taken with the ESO 1.5m telescope in Dec'92, with the most prominent lines identified
these lines are filled-in by emission one can derive a lower limit for the effective temperature or the spectral type of the region of formation of these lines. From Table 6 (click here) we find for BN Ori an EW ratio of 12, which points to a spectral type around G5. With the same method Shanin (1979) classified FU Ori as a K0 star. A reticon spectrum of HR 5999 (Tjin A Djie et al. 1989) in the same spectral region gives us an EW ratio of 3.1, which indicates a spectral type around A5. High-resolution observations of only the 8498 and 8542Å lines (Hamann & Persson 1992) show the unblended profiles with indications for some filling-in by emission. After correction for emission the EWs would be somewhat larger, which could move the spectral type of HR 5999 to a slightly later type, e.g. A7. Similar (unknown) emission corrections to the CaII (2)IR-triplet for BN Ori and FU Ori could shift their classifications to types later than G5 and K0 respectively. In the far-red spectrum we also observed the two OI IR-triplets around 7774 and 8446Å (blended with P18 at 8438Å). The EWs of these lines can be compared with those of standard stars of well-known spectral class (Slettebak 1986; Figs.2 and 3, Faraggiana et al. 1988). High-resolution profiles of these lines for HR 5999 have been given by Felenbok et al. (1988) and by Hamann & Persson (1992). The EW, measured from the best profile of the OI (1)IR-triplet is 1.38Å, which according to Slettebak (1986; Fig.2) is consistent with A2- A7III sh. A correction for the emission-component would not change this conclusion. The EW of this (unblended) triplet for BN Ori is 0.22Å, which would classify the star as F6- G0, unless the contribution by emission is very large. A comparison of the EW of the blended OI (4)IR-triplet for HR 5999 with Fig.3 of Slettebak's paper gives again A2- A7III sh, even if the filling-in by emission (Hamann & Persson 1992) is taken into consideration. The very small EW of this line for BN Ori suggests again a type G, but F0 could be possible if the correction for emission fill-in would raise the EW from 0.1 to 0.4Å. For FU Ori the EW of this line is 0.3Å (Shanin 1979). This could lead to a classification F3IV. Since FU Ori has strong shell features and high luminosity-class (see next section) its spectral type must be later than A6- G0 (Slettebak 1986; Fig.3), unless the emission contribution to the OI (4)IR-triplet\ is very strong.
A high-dispersion visual spectrum of BN Ori was obtained on Nov.21, 1991 with the main stellar spectrograph (MSS) mounted on the 6m Big Azimuthal Telescope (BTA) of the Special Astrophysical Observatory (SAO) in the Northern Kaukasus. Various parts of this spectrum are given in Fig. 5 (click here) and the values of the EWs and FWs of the stronger lines are listed in Table 11 (click here), together with those of HR 5999 determined from the Coudé spectrum G9359 ( = 12.4, Tjin A Djie et al. 1989), and some qualitative information concerning line strengths from the Coudé spectra of FU Ori ( = 16, Herbig 1966).
In this table the identification of the lines with wavelengths longer than 5200Å is tentative, except for the BaII (6497Å) and the LiI (6708Å) lines. From the width of several photospheric lines (MgII (4), FeII (42) and FeI (43)) we estimate a rotation rate of = 180- 230 for BN Ori. A similar estimate for HR 5999 gives 135. However, a comparison with the MgII(4481Å) line-width with those of rotation standard stars (Slettebak et al. 1975), taken with the same instruments as used for HR 5999, leads to = 180 20 for this star (Tjin A Djie et al. 1989). This suggests that the rotation rate of BN Ori is probably somewhat higher, perhaps around 250. For FU Ori Herbig (1966) derived 60.
Figure 5: Various parts of the high-dispersion visual spectrum of BN Ori obtained at SAO in 1991. Shown are from top to bottom: CaII H and K, H to H14 and the TiII lines at 3760Å, H and its blue neighbourhood (VII, SrII, FeI, TiII, MnI (2) triplet), H and the red neighbourhood, the red neighbourhood of H. The identification of the marked lines can be found in Table 11 (click here)
The most important conclusion from the spectrum of BN Ori is that different lines suggest different spectral types for the star:
Together with the classifications G0 from the OI (1)IR-triplet and G5 from the CaII (2)IR-triplet (Sect. 4.2.2 (click here)) this variety in spectral type indicates a considerable thermal stratification of the atmosphere of BN Ori, which must consist of an A6- A7 photosphere, surrounded by a rather dense and cooler envelope or disc. Several lines such as SrII (4078Å) and the semi-forbidden TiII (13) lines near 3760Å are typical shell lines and are probably formed in the low-density outer part of the envelope.
Most of the lines mentioned above are also present in the spectra of FU Ori and HR 5999 which were obtained with comparable resolution. The visual spectrum of FU Ori has been classified as F2p:I-IIe (Herbig 1966). If we add the classification F6- G0 from the OI (4)IR-triplet and K0 from the CaII (2)IR-triplet we can conclude that also the atmosphere of FU Ori is thermally stratified. Similar to BN Ori the spectrum of FU Ori shows a strong G-band and strong photospheric lines of BaII and LiI. However, in contrast with BN Ori, strong components with velocities around -50 are present in the spectrum of FU Ori for all lines of Hydrogen and neutral or singly-ionised metals (with exception of the FeI lines with higher excitation levels). Their presence has led to the qualification peculiar in the spectral classificaton. Because of their difference in velocity with the photospheric lines these components are probably formed in an outer shell region. Since we have no EWs from the visual spectrum of FU Ori, with the exception of LiI, NaID and H (Table 6 (click here)), we only give qualitative estimates from Herbig (1966). With exception of the emission in CaIIK and H, the spectrum did not seem to vary significantly in time between 1945 and 1963.
The visual spectrum of HR 5999 has been classified as A5- A7III e (Bessell & Eggen 1972; Tjin A Djie et al. 1989). The classification from the far-red spectrum (Sect. 4.2.2 (click here)) is again A7III, which indicates an absense of significant thermal stratification of the envelope of HR 5999. This means that the stratification of the atmosphere as found in BN Ori and FU Ori can not originate from the disc component, because this component is significant in the case of HR 5999 (Sect. 5 (click here)). We will discuss this further in Sect. 6 (click here).
In contrast with FU Ori and BN Ori we do not find a G-band and LiI line in the spectrum of HR 5999, but similar to FU Ori, it has strong blue-shifted (-30 to -50) components of the Balmer- and metallic- lines, which are formed in a cool, dense outer shell (Tjin A Djie et al. 1989). Such an dense cool circumstellar shell does not seem to be present around BN Ori.
This is confirmed by the high-resolution observations of the NaI D (Fig. 6 (click here)) with the CAT at ESO ( = 0.11) and the MSS at SAO ( = 0.65) and the KI lines (not shown). The observations at ESO show the presence of interstellar components only, with a heliocentric-velocity of +26, which is of the same order as those of other stars in the Orion complex. The interstellar NaI D lines have EWs equal to 0.13Å () and 0.09Å (). Since these lines are not saturated, we can use the Strömgren (1948) method, with which we derive an interstellar (foreground) column density for NaI of 1.27 cm. With the relation between E(B-V) and NaI D column density of Hobbs (1974) we find for the foreground of BN Ori an E(B-V) of 007. From the NaI D\ lines, observed at SAO in 1991, we obtain the same value. These values are in agreement with those determined for the Orion complex by Walker (1969). In contrast with the NaI D lines of BN Ori, those of HR 5999 and FU Ori have several velocity-components which are formed in the circumstellar envelope of the star. The velocities measured for HR 5999 are -35, and for FU Ori -100 (Bastian & Mundt 1985; Reipurth 1990). The circumstellar origin of the NaI D lines in the spectrum of HR 5999 is clear from the variability of their EWs and the absence of NaI D lines in the spectrum of the common proper motion companion HR 6000 (Tjin A Djie et al. 1989).
Figure 6: High-resolution heliocentric-corrected NaI D spectra obtained at ESO in 1994 for BN Ori (top) and HR 5999 (bottom, see also Grady et al. 1996). The dotted lines indicate the continuum level. The spectrum of HR 5999 was used for the annotation and the difference in radial velocity between the two galactic directions of observation introduces a wavelength-shift in the figure for the spectrum of BN Ori
The KI resonance lines of BN Ori observed at ESO (Jan.20, 1995) were very weak. The 7699Å line is hardly distinguishable from the noise, while the 7665Å line is in the wing of a rotational line of the telluric A band of O. The lines have a heliocentric-velocity of +20 and are of interstellar origin. The EWs of these lines are 8.5mÅ ( 30%) and 4mÅ ( 50%) for the 7665 and 7699Å line, respectively. If the 7665Å line is optically thin we derive with the oscillator strength of the line, a column density of 2.3 cm. The ratio of the column densities of NaI to KI in the direction of BN Ori is then 54, the same value as for the direction of AB Aur (Felenbok et al. 1983).
A peculiar feature of BN Ori is the presence of broad, shallow and symmetric components of the NaI D lines, with a FW of about 400. The corresponding is close to the rotation rate (220) of the photosphere of BN Ori, which has = 7150K. Because of the low ionization potential of NaI ( = 5.14eV), we do not expect to observe these lines from photospheres with > 5500K. However, the broad components can originate from the atmosphere of the inner part of the accretion-disc, which has a temperature below 5890K (Table 7 (click here)) and a which is close to the photospheric-rotation rate. A similar broad component can also be found in the high-resolution profiles of the NaI D lines of HR 5999. However, in this case its width and EW are more difficult to estimate because of the circumstellar emission-component in both NaI D lines. The values for HR 5999 in Table 6 (click here) are therefore probably lower limits. The large width of the circumstellar components in the NaID profiles of FU Ori (Reipurth 1990) does not enable us to detect the presence of such shallow components for this star.
The H-profile of BN Ori has been observed by us at three dates in high-resolution: in Nov'91 at the SAO, and twice in Jan'95 at ESO. An additional profile obtained on Nov.10, 1989 by Fernández et al. (1995) is very similar to the one of Nov'91. The resulting profiles are also similar to those given by Kolotilov & Zajtseva (1976), but show more details. In order to estimate the chromospheric profile we divided the normalised fluxes in the observed profiles by the corresponding normalised fluxes of the profiles of 21 Vul (A7- F0IV, 200), which has a very weak shell. The profile of 21 Vul was taken from Doazan et al. (1991) and a contribution from the shell to H is only expected in the minimum of the profile. After this procedure, the SAO H-profile shows a single emission peak with a single, broad absorption-feature with -650, which agrees well with the velocities found by Zajtseva & Kolotilov (1973). One of the ESO H-profiles has an almost symmetrical shape, with a relative central part which is almost equal to that of the SAO H-profile. The other profile is less strong in its central part and has a depression near +240. The three profiles and their "chromospheric'' shapes are shown in Fig. 7 (click here). The shapes are different from those of HR 5999 (Tjin A Djie et al. 1989; Baade & Stahl 1989; Praderie et al. 1991), which show no high-velocity absorption-features. The H-emission flux of BN Ori can be derived from Fig. 7 (click here) by integration and normalisation with the continuum flux near 6600Å, obtained from the SED with the help of the distance to radius ratio and after correction for extinction (Sects. 2 (click here) and 4.3 (click here)). In this way we find emission fluxes of 1.56 for the observations of Nov.10, 1989 and Nov.21, 1991, and 1.22 for the observations of Jan.20, and Jan.17, 1995. In the same way we may derive the H emission flux of HR 5999 in its phase of maximum brightness = 68 (Tjin A Djie et al. 1989), which after an estimated correction for the central absorption, results to 2.3- 3.4 . For =76 it is 5.8 . The flux of HR 5999 is therefore larger than that of BN Ori by a factor which is in the range 22- 28.
Figure 7: Left: high-resolution H-profiles of BN Ori obtained at SAO in 1991 (top) and at ESO on Jan.17 1995 (middle) and Jan.20 1995 (bottom). Right: corresponding profiles after re-binning and division by the normalised H-profile of 21 Vul (from Doazan et al. 1991). Top-axis gives the velocity () relative to 6562.82Å
The H-profiles of FU Ori (Bastian & Mundt 1985; Reipurth 1990) have P-Cygni shapes, with a very broad and deep absorption trough. Outflow velocities as high as -350 can be estimated from these profiles. The observed profile has been corrected for photospheric contributions with the use of calculated H-profiles (Kurucz 1979) of a F0- F2I b photosphere ( = 7500K, g = 2.0). The extinction-corrected surface-flux of the symmetrised H-profile is then 3.8 . The H-emission fluxes are collected in Table 9 (click here).