The variable character of BN Ori was detected by Cannon (1919) and since then the star has been monitored extensively both visually and photographically. Tsesevich made nearly 4000 estimates of BN Ori from sky patrol photographic plate collections of the Harvard, Odessa, Sonneberg and Dushanbe observatories and constructed a light curve of over the period 1898- 1964 (Dragomiretskaya 1965; Tsesevich & Dragomiretskaya 1973).
Figure 1: Lightcurve of for BN Ori over the period 1898- 1964, constructed by Tsesevich & Dragomiretskaya (1973), and extended up to 1995
Table 2: Photometric Data of BN Ori on the Walraven (WULBV) system
Table 3: Photometric Data of BN Ori on the Johnson/Cousins system. The last column (N, O) gives the number of observations per night and the observatory: DASA(D), ESO(E) or Kitt Peak(K)
This light curve of BN Ori shows quiet periods with maximum brightness 97, during which the brightness frequently decreases by at most 05, separated by episodes of roughly 12.5 years, during which the mean brightness dropped with 1 or 2 over 100- 200 days or with 3 over 20- 50 days. Superposed on this long-term variability are rapid fluctuations with amplitude of 15 during about 10 days. It was already noted by Beyer (1937) and confirmed by Hoffmeister (1949) and Payne-Gaposhkina (1952) that the behaviour of BN Ori is reminiscent of that of the RW Aur variables T Ori and XY Per and we can extend the similarity now to other Herbig Ae stars with Algol-type minima such as BF Ori, UX Ori, WW Vul and HR 5999. In a discussion of the photometric variability of BN Ori, Dragomiretskaya (1965) concluded that the large amplitude-variations cannot be due to temperature differences on the stellar surface, since a dimming
Table 4: NIR photometric data of BN Ori
by 15 would already require a full coverage by cool spots of the stellar surface. Dragomiretskaya also used the visual photometry by Esch (1925), Beyer (1937) and Martinov (1951), to derive the colour indices for a number of dates and the resulting colour-magnitude diagram shows that the strong variations of BN Ori follow a reddening-line similar to those followed by many Herbig Ae stars. Such a type of variation is usually ascribed to obscuration by variable amounts of circumstellar dust (Wenzel 1968; Wenzel et al. 1971). For BN Ori the slope of the reddening-line was about 3, which is close to that of interstellar dust. Tsesevich & Lange observed that BN Ori showed Algol-like variations with a period of 27 days (Martinov 1951). From an analysis of quasi-periodic features during high-activity episodes in the light curve of BN Ori, Ishchenko (1982) detected a quasi-cyclic 18- 22 day periodicity superimposed on a stochastical behaviour and attributed this periodicity to the rotation of obscuring circumstellar dust clouds.
Figure 1 (click here) shows that the large-scale structure of the light curve changed around 1940: the amplitude of the strong variations decreased and the brightness maximium increased to 92 in Dec'46. Since then the variations were gradually quenched from an amplitude of 04 to about 01 in 1966, while the maximum of the brightness-variations slowly decreased to 101. In Sect. 6 (click here) we shall compare this behaviour with those of classical FUORs. After 1965 all observations have been made photoelectrically. Between 1965 and 1975 Zajtseva (1968, 1970), Kolotilov & Zajtseva (1976) and Kolotilov et al. (1977) have made several observations of BN Ori on the UBV and JHKL photometric systems, but they did not detect a pronounced variation.
The star has been observed regularly by various observers on several photometric systems (see Table 1 (click here)).
Figure 2: Observed (squares) and extinction-corrected (circles) SED of BN Ori based on the data of Feb'86. The L-band observation of Cohen (1973) is indicated by the "upper" triangle (with error-bar). The L-band observation from Table 4 (click here) is indicated by the "lower" triangle (with upperlimit). The solid line is the Kurucz model of = 7150K, = 3.8 and solar abundance, normalised to the extinction-corrected V-band flux of BN Ori. The dotted lines are the observed- and extinction-corrected UV-fluxes
After transformation of the data on the various systems to the Johnson magnitude and colour we can hardly find any change in brightness and colour over these years. The average values are: = 965 003 and = 047 002. Only on one occasion (Oct.29- Nov.2, 1991), the star seemed to dim by 05 (Kovalchuk 1991). Because there are no data available for the 34 nights preceding Oct.29 and for the 12 nights following Nov.2, we can only state that the variation was limited to at most 51 nights. The variation is along the reddening-line with slope 3, so that this event could have been due to infalling circumstellar dust which evaporated in the neighbourhood of the star.
Since 1976 we have obtained only two observations in the NIR (Table 4 (click here)), and in Fig. 2 (click here) we give a SED of BN Ori based on these data of Feb'86. After transformation of the magnitudes to fluxes and correction of these fluxes for foreground extinction with the law of Savage & Mathis (1979) and E(B-V) = 007 (Sect. 4.3 (click here)), the corrected SED could be fitted with a Kurucz (1991) model for = 7150K and , corresponding to a spectral type of F0III - IV (Sect. 5 (click here)) and an E(B-V) = 017. From Fig. 2 (click here) we note that we get a good agreement in the visual and NIR wavelength ranges, but that there remains an excess in the UV. This excess will be explained (Sect. 5 (click here)) in terms of the contribution from a (hot) boundary layer between the star and accretion-disc. If we compare our NIR data of 1986 with those obtained before 1977 (Glass & Penston 1974; Kolotilov et al. 1977) we note no difference in the H and K magnitudes. However, our J-band flux is somewhat larger than that of Kolotilov et al. and our L-band flux is significantly smaller than the one Cohen (1973; 76) measured in 1971. The earlier attempts to match the SED (Kolotilov & Zajtseva 1976; Brown et al. 1986) were based on this higher L-band flux. No IRAS point-source was detected at the position of BN Ori, probably because the far-IR fluxes of the star are not higher than the background.
From the Walraven photometry (which has two passbands in the UV: W at 3236Å and U at 3623Å), we can get a measure of the magnitude of the Balmer discontinuity (). The mean of the 9 observations of Dec'85- Jan'86 gives . This is a deficiency of 048 with respect to the prediction of the Kurucz (1991) model for = 7200K and which we take for the classification F0III (Sect. 5 (click here)). If this is due to the contribution of recombination of H free-bound transitions near the star, we can derive (Garrison 1978) that for a temperature of 10K in this region the emission measure is 1.86 cm. If we assume that this ionosphere is in hydrostatic equilibrium with a scale-height of 4 we find within 2 an electron density = 3.34 cm.
For HR 5999 (A5- 7III) we used the Walraven photometry to derive a equal to 123 which results in a deficiency of 020 when compared to a Kurucz (1991) model with = 8000K. This gives an emission measure of 6.0 cm, which corresponds to = 5.6 cm close to the star, so that is 2.8 times the value found for BN Ori.
Table 5: Log of spectroscopic observations of BN Ori. Columns 8 to 10 denote the dispersion () the resolution () and the exposure time (min). The IUE imagenumbers are given in Col. 5 and the first digit denotes the used camera (1 = LWP, 3 = SWP) and LAP means large aperture. Observers: AB - A. Brown, DW - D. de Winter, HC - H. Cuypers, HT - H.R.E. Tjin A Djie, MA - M.E. van den Ancker, VK - V. Kotyshev, VS - V.S. Shevchenko