Table 2: Optical photometry of AS 78 at the Tien-Shan Observatory

JD 2400000+ V U-B B-V V-R R-I JD 2400000+ V U-B B-V V-R R-I

$^{\rm a}$ 11.26 -0.07 0.71 - - 50097.19 11.26 -0.06 0.67 0.86 0.77

48997.15 11.25 0.12 0.86 0.80 0.70 50340.47 11.18 -0.13 0.69 0.85 0.80

48998.11 11.39 -0.12 0.88 0.80 0.65 50348.48 11.30 -0.07 0.66 0.86 0.78

49000.10 11.21 0.15 0.79 0.78 0.65 50353.48 11.23 -0.09 0.70 0.81 0.76

49001.14 11.18 0.18 0.78 0.77 0.65 50415.27 11.39 0.00 0.70 0.87 0.77

49002.12 11.26 0.25 0.74 0.73 0.67 50416.23 11.39 0.03 0.67 0.88 0.77

49005.13 11.26 0.04 0.77 0.79 0.68 50417.23 11.33 -0.16 0.71 0.82 0.82

49232.47 11.20 0.10 0.79 0.96 0.95 50424.21 11.27 -0.03 0.73 0.81 0.75

49409.08 11.26 -0.09 0.80 0.76 0.70 50426.27 11.21 0.10 0.71 0.81 0.80

49423.11 11.21 0.01 0.70 0.76 0.70 50429.26 11.33 -0.06 0.67 0.86 0.80

49605.38 11.34 0.10 0.76 0.89 0.74 50439.21 11.30 -0.18 0.85 0.86 0.75

49610.39 11.16 -0.03 0.75 0.84 0.70 51047.46 11.19 -0.09 0.69 0.88 0.77

49612.39 11.14 -0.07 0.81 0.78 0.67 51051.44 11.25 -0.06 0.66 0.86 0.80

49729.10 11.23 -0.07 0.69 0.88 0.83 51154.23 11.26 -0.03 0.70 0.87 0.80

49735.06 11.18 -0.03 0.71 0.84 0.76 51156.25 11.20 -0.20 0.70 0.89 0.82

49738.11 11.19 -0.03 0.71 0.85 0.73 51157.24 11.25 -0.11 0.70 0.86 0.79

49740.09 11.22 0.00 0.69 0.90 0.80 51159.24 11.23 -0.10 0.68 0.85 0.76

49741.09 11.23 -0.02 0.73 0.85 0.80 51171.29 11.24 -0.06 0.67 0.91 0.83

49742.13 11.15 -0.01 0.70 0.86 0.80 51247.15 11.25 0.00 0.67 0.86 0.75

50096.15 11.28 -0.01 0.62 0.82 0.74

**Table 2:** Optical photometry of AS 78 at the Tien-Shan Observatory
JD 2400000+	V	U-B	B-V	V-R	R-I	JD 2400000+	V	U-B	B-V	V-R	R-I
$^{\rm a}$	11.26	-0.07	0.71	-	-	50097.19	11.26	-0.06	0.67	0.86	0.77
48997.15	11.25	0.12	0.86	0.80	0.70	50340.47	11.18	-0.13	0.69	0.85	0.80
48998.11	11.39	-0.12	0.88	0.80	0.65	50348.48	11.30	-0.07	0.66	0.86	0.78
49000.10	11.21	0.15	0.79	0.78	0.65	50353.48	11.23	-0.09	0.70	0.81	0.76
49001.14	11.18	0.18	0.78	0.77	0.65	50415.27	11.39	0.00	0.70	0.87	0.77
49002.12	11.26	0.25	0.74	0.73	0.67	50416.23	11.39	0.03	0.67	0.88	0.77
49005.13	11.26	0.04	0.77	0.79	0.68	50417.23	11.33	-0.16	0.71	0.82	0.82
49232.47	11.20	0.10	0.79	0.96	0.95	50424.21	11.27	-0.03	0.73	0.81	0.75
49409.08	11.26	-0.09	0.80	0.76	0.70	50426.27	11.21	0.10	0.71	0.81	0.80
49423.11	11.21	0.01	0.70	0.76	0.70	50429.26	11.33	-0.06	0.67	0.86	0.80
49605.38	11.34	0.10	0.76	0.89	0.74	50439.21	11.30	-0.18	0.85	0.86	0.75
49610.39	11.16	-0.03	0.75	0.84	0.70	51047.46	11.19	-0.09	0.69	0.88	0.77
49612.39	11.14	-0.07	0.81	0.78	0.67	51051.44	11.25	-0.06	0.66	0.86	0.80
49729.10	11.23	-0.07	0.69	0.88	0.83	51154.23	11.26	-0.03	0.70	0.87	0.80
49735.06	11.18	-0.03	0.71	0.84	0.76	51156.25	11.20	-0.20	0.70	0.89	0.82
49738.11	11.19	-0.03	0.71	0.85	0.73	51157.24	11.25	-0.11	0.70	0.86	0.79
49740.09	11.22	0.00	0.69	0.90	0.80	51159.24	11.23	-0.10	0.68	0.85	0.76
49741.09	11.23	-0.02	0.73	0.85	0.80	51171.29	11.24	-0.06	0.67	0.91	0.83
49742.13	11.15	-0.01	0.70	0.86	0.80	51247.15	11.25	0.00	0.67	0.86	0.75
50096.15	11.28	-0.01	0.62	0.82	0.74

$^{\rm a}$

AS 78 (Merrill & Burwell 1933) = LS I $+56\hbox{$^\circ$ }$ 96 is an 11-mag object in Camelopardalis which was identified by Miller & Merrill (1951) as a Be star. Only one other UBV observation is found in the literature (Haug 1970). Dong & Hu (1991) identified AS 78 with the IRAS PSC source 03549+5602, with fluxes at 12, 25, and 60 $\mu$ m of 3.75, 3.13, and 0.57 Jy respectively. These fluxes indicate a strong IR excess with respect to the radiation of a hot star (see Fig. 11), that might be due the presence of circumstellar dust or a cool companion. The absence of spectral features of a late-type star in the spectrum of AS 78 and the large near-IR colour-indices (see Sect. 3.2.1) suggest that if there is a secondary component in the system, it is not responsible for the bulk of the visible and IR flux. Furthermore, the IRAS data are in agreement with the presence of circumstellar dust in the system (see Sect. 3.3).

Recently AS 78 was detected by the MSX satellite in the course of a mid-IR galactic plane survey (MSX5C-G146.9227+02.3139, Egan et al. 1999). Its fluxes were measured in three bands centered at 8.28, 12.13, and 14.65 $\mu$ m (3.69 $\pm$ 0.18, 3.06 $\pm$ 0.23, and 2.87 $\pm$ 0.18 Jy respectively). The 12- $\mu$ m IRAS, MSX, and IRTF fluxes agree with each other within a 3 $\sigma$ range. The smooth flux decrease with wavelength in this region suggests that the 10 $\mu$ m silicate feature is probably weak or not present in the spectrum (see Fig. 11). Furthermore, a steep decrease in the IR flux toward longer wavelengths implies a large density gradient in the dust envelope and/or a small size. It might indicate that the dust formation process started recently and the dust is concentrated mainly close to the star.

No previous near-IR observations for this star have been reported in the literature. Our measurements (see Tables 3 and 4) show that a significant excess radiation has been reliably detected in this spectral range. Note that the CST K-magnitude is nearly 1 mag fainter than those of the other dates. This cannot be due to clouds, which were noticed nearly an hour after the observation of AS 78, because the measurements in all three bands were repeated several times in different sequences, and the J- and H-band CST magnitudes are in good agreement with those obtained at Lick 3 weeks earlier. Nor do we believe that this discrepancy is due to the small diaphragm size used at CST, because the near-IR emission is due to hot dust which is located close to the star; 15 $\hbox{$^{\prime\prime}$ }$ corresponds to 0.2 pc at the distance of AS 78 (see Sect. 3.2.1) and certainly includes all the hot dust and even most of the cooler part of the dusty envelope. Our IRTF measurements through a 10 $\hbox{$^{\prime\prime}$ }$ aperture confirm this suggestion. In any case, our near-IR data indicate that there is a noticeable variability at least in the H- and K-bands.

In order to estimate the strength and shape of the near-IR excess we adopt $\overline{V-K}$ = 4.8 mag, $\overline{J-H}$ = 1.1 mag, and $\overline{H-K}$ = 1.2 mag (from Table 3), which correspond to the dereddened values of 2.4, 0.8, and 1.0 mag, respectively, using E_B-V = 0.9 mag derived below. This is in agreement with the above suggestion about the presence of the circumstellar dust (see Fig. 11).

Our optical photometry (38 observations from more than 6 years) shows that the star's brightness varies with an amplitude of 0.25 mag (see Table 2). The U-B and B-V colour-indices display a positive correlation with each other and a negative correlation with V-R and V-I (Fig. 1). The temporal trend direction in the U-B vs. B-V plane (see Fig. 1b) is closer to the line of the stellar intrinsic colour-indices (Strajzhys 1977) than to the interstellar reddening vector. This suggests that these colour variations are caused by changes in the star's parameters rather than by those of the circumstellar matter. The object's strong emission-line spectrum implies a certain contribution of the latter to the brightness and colour-indices. However, this contribution is probably small, since our results for the interstellar reddening derived from the photometry and spectroscopy (Sect. 3.2.1) are very close to each other.

The colour-indices (U-B, B-V) change between (-0.10, 0.67) and (0.15, 0.77). These extreme values correspond to the spectral types B2 and B4 respectively (if we assume luminosity types V or III) with essentially the same reddening, E_B-V = 0.90 $\pm$ 0.05 mag. The spectral type interval would be B7 to B9.5, assuming that the star is a supergiant. These estimates were obtained using the ratio of total to selective extinction R = 3.1, and E_U-B/ E_B-V = 0.84, which was derived from the data listed in Table 5. A small ( $\sim$ 0.1 mag) excess radiation in the R and I bands is seen for all suggested luminosity types after dereddening. As the spectral type becomes earlier, V-R and V-I increase from 0.75 and 1.4 to 0.9 and 1.7 mag, respectively, yielding an increase in the red-excess radiation. Temporal evolution of the colour-indices shows that both U-B and B-Vdecrease with time, while both V-R and V-I increase (see Fig. 1d). This indicates that both the star's temperature and the red-excess radiation, which is most likely due to free-free radiation, increase with time. The latter fact might imply strengthening of the stellar wind.

A similar behaviour of U-B and B-V is a frequently observed phenomenon in classical Be stars (e.g., Harmanec 2000), which have significantly weaker emission-line strengths than those of AS 78, and very flattened circumstellar gaseous envelopes. A positive correlation at the same brightness level (as is seen in AS 78) is usually observed in the systems seen nearly edge-on and can be explained by a change of the Balmer discontinuity (Divan & Zorec 1982). The latter is due to a change in the amount of circumstellar matter close to the star. Balmer line profiles of AS 78 suggest an insignificant flattening of its envelope (see Sect. 3.2.1) that may have the same impact on the color-indices. Therefore, while the stellar wind strengthening mentioned above may account for all the four color-index changes, since we do not have more detailed spectrophotometric information, we cannot rule out temperature changes as a possible contributor to the observed photometric behaviour.

Periodogram analysis of the optical brightness variations shows that the power spectrum contains a number of peaks of almost equal strength between 60 and 500 days. However, none of the possible periods displays a convincing phase curve.

Table 3: Near-IR photometry of AS 78

JD 2400000+ J H K Obs.

50439.21 6.4 $\pm$ 0.2 TS

51047.46 6.57 $\pm$ 0.07 TS

51051.44 7.1 $\pm$ 0.1 6.16 $\pm$ 0.07 TS

51247.15 6.51 $\pm$ 0.07 TS

51419.03 8.8 $\pm$ 0.1 7.7 $\pm$ 0.1 6.3 $\pm$ 0.1 L

51439.12 8.76 $\pm$ 0.03 7.75 $\pm$ 0.02 7.45 $\pm$ 0.02 T

**Table 3:** Near-IR photometry of AS 78
JD 2400000+	J	H	K	Obs.
50439.21			6.4 $\pm$ 0.2	TS
51047.46			6.57 $\pm$ 0.07	TS
51051.44		7.1 $\pm$ 0.1	6.16 $\pm$ 0.07	TS
51247.15			6.51 $\pm$ 0.07	TS
51419.03	8.8 $\pm$ 0.1	7.7 $\pm$ 0.1	6.3 $\pm$ 0.1	L
51439.12	8.76 $\pm$ 0.03	7.75 $\pm$ 0.02	7.45 $\pm$ 0.02	T

The observatory name is listed in Col. 5 (TS - Tien-Shan, L - Lick, T - del Teide).

Table 4: IRTF photometry of AS 78

JD 2400000+ K L M N 8.7 9.8 11.6 12.5

51599.77 6.57 5.14 4.35

51600.75 6.57 5.08 4.29 2.71 2.74 2.49 2.43 2.24

**Table 4:** IRTF photometry of AS 78
JD 2400000+	K	L	M	N	8.7	9.8	11.6	12.5
51599.77	6.57	5.14	4.35
51600.75	6.57	5.08	4.29	2.71	2.74	2.49	2.43	2.24

$\mu$

To decide which luminosity type is more realistic for AS 78 we searched for photometric data of nearby stars in the sky. Several B-type supergiants found within an angular distance of 1 $\hbox{$^\circ$ }$ (from the object) are listed in Table 5. Despite the fact they have colour-indices close to those of AS 78 and similar reddenings, they are 1-3 mag brighter than AS 78. This fact might imply that our object is a low- or an intermediate-luminosity star rather than a high-luminosity supergiant. The photometric data from Table 5 indicate that the interstellar extinction (A_V) varies between 2.5 and 3.0 mag beyond $\sim$ 2.5 kpc from the Sun in this direction. The extinction for AS 78 we derived from the averaged B-V is A_V = 2.7 $\pm$ 0.2 mag. The above data on the nearby stars and the derived A_V imply that distance of AS 78 is at least 2.5 kpc.

Table 5: Photometry of stars in the vicinity of AS 78
Name V U-B B-V Sp.T. A_V D Ang.

kpc $\hbox{$^\circ$ }$

BD $+55\hbox {$^\circ $ }837$ 9.58 -0.26 0.70 B2 Ib 2.69 3.5 0.3

BD $+55\hbox{$^\circ$ }838$ 9.29 -0.10 0.82 B3 Ib 2.97 2.5 0.5

BD $+56\hbox{$^\circ$ }873$ 8.98 -0.51 0.49 O9.5 Ib 2.50 3.4 0.6

BD $+56\hbox{$^\circ$ }868$ 9.14 -0.57 0.31 B0.5 V 1.63 1.7 0.7

HDE 237213 8.72 -0.06 0.77 B3 Ia 2.82 4.0 0.7

LS I $+56\hbox{$^\circ$ }92$ 10.26 -0.09 0.69 B6 I 2.36 5.0 1.0

**Table 5:** Photometry of stars in the vicinity of AS 78
Name	V	U-B	B-V	Sp.T.	A_V	D	Ang.
						kpc	$\hbox{$^\circ$ }$
BD $+55\hbox {$^\circ $ }837$	9.58	-0.26	0.70	B2 Ib	2.69	3.5	0.3
BD $+55\hbox{$^\circ$ }838$	9.29	-0.10	0.82	B3 Ib	2.97	2.5	0.5
BD $+56\hbox{$^\circ$ }873$	8.98	-0.51	0.49	O9.5 Ib	2.50	3.4	0.6
BD $+56\hbox{$^\circ$ }868$	9.14	-0.57	0.31	B0.5 V	1.63	1.7	0.7
HDE 237213	8.72	-0.06	0.77	B3 Ia	2.82	4.0	0.7
LS I $+56\hbox{$^\circ$ }92$	10.26	-0.09	0.69	B6 I	2.36	5.0	1.0

1981

3.1.2 MWC 657

The results of a photometric study of MWC 657 have recently been presented by Miroshnichenko et al. (1997), who concluded that it is an early B-type star which is located above the main sequence and is surrounded by an optically thin dusty envelope. Ten UBVRI observations obtained by these authors revealed a variability of about 0.3 mag in all bands. Our new optical photometric data (see Table 6) show that these variations are even larger and reach $\sim$ 0.6 mag for the whole data set (20 observations during 2.2 years). The near-IR excess in MWC 657 has been detected by Miroshnichenko et al. (1997) on the basis of 2 K-band measurements obtained on 1996 December 4 and 10 (6.67 $\pm$ 0.15 and 6.31 $\pm$ 0.15, respectively). Recently we obtained new measurements on 1999 September 17 (JD 2451439.043) with the following results: J = 9.14 $\pm$ 0.04, H = 8.05 $\pm$ 0.02, K = 6.84 $\pm$ 0.02. The 1999 K-band brightness is lower than that of 1996 which we explain below. Nevertheless, the new data confirm the existence of the near-IR excess in the object's SED. Furthermore, MWC 657 was detected by the MSX satellite (MSX5-G107.6722+01.4002) in four bands centered at 8.28, 12.13, 14.65, and 21.34 $\mu$ m with the fluxes 5.64 $\pm$ 0.28, 5.33 $\pm$ 0.22, 4.43 $\pm$ 0.21, and 4.33 $\pm$ 0.45 Jy respectively. The 12- $\mu$ m MSX flux of MWC 657 is nearly 10 per cent smaller than that of IRAS, which is about a 2.5 $\sigma$ difference. The MSX fluxes indicate that the presence of the 10- $\mu$ m silicate feature in the spectrum of MWC 657 is questionable, as in the case of AS 78.

Despite the small amount of photometric data obtained, we attempted the Lomb-Scargle periodogram analysis of the optical light curves (Scargle 1982). The strongest peak in the power spectrum corresponds to a period of 86.2 days. The phase curves folded onto this period are presented in Fig. 2. It turned out that the 1996 and 1998 optical photometry was obtained in non-overlapping phase ranges. However, since the same equipment was used in both data sets, this fact shows that the variations were stable during at least 10 cycles. It is seen in Fig. 2 that the object generally becomes bluer when it brightens, although there is a short period of reddening around phase 0.5 which is best seen in the B-V colour. The K-band brightness seems to increase along with the optical brightness. At the same time, we cannot rule out the possibility that the photometric variations detected reflect different activity states of the object and are not periodic.

A search for photographic plates containing images of MWC 657 in the Sternberg Astronomical Institute (Moscow, Russia) archive revealed 85 such plates. The object's brightness was estimated by N.E. Kurochkin (priv. commun.) by eye. He found that it varied between $m_{\rm pg}$ 13.5 and 14.5 in 1901-1968. Unfortunately, it turned out to be close to the plates' threshold, and the estimates have large errors. As a result, they do not show any variations with a period close to that of the photoelectric data. Nevertheless, the photographic data show no dramatic variations in the brightness of MWC 657.

Table 6: Optical photometry of MWC 657

JD 2400000+ V U-B B-V V-R R-I

51053.39 12.92 -0.14 1.34 1.57 0.87

51054.26 12.82 1.42 1.54 0.94

51060.45 12.74 1.44 1.54 0.86

51073.27 12.93 -0.3: 1.47 1.63 0.94

51083.31 12.82 -0.09 1.47 1.56 0.93

51100.27 12.67 0.02 1.41 1.51 0.93

51102.30 12.66 -0.1: 1.36 1.53 0.91

51154.11 13.07 1.48 1.80 0.77

51155.06 12.86 0.08 1.41 1.58 0.92

51157.06 12.92 0.0: 1.34 1.51 0.80

**Table 6:** Optical photometry of MWC 657
JD 2400000+	V	U-B	B-V	V-R	R-I
51053.39	12.92	-0.14	1.34	1.57	0.87
51054.26	12.82		1.42	1.54	0.94
51060.45	12.74		1.44	1.54	0.86
51073.27	12.93	-0.3:	1.47	1.63	0.94
51083.31	12.82	-0.09	1.47	1.56	0.93
51100.27	12.67	0.02	1.41	1.51	0.93
51102.30	12.66	-0.1:	1.36	1.53	0.91
51154.11	13.07		1.48	1.80	0.77
51155.06	12.86	0.08	1.41	1.58	0.92
51157.06	12.92	0.0:	1.34	1.51	0.80

3.2 Spectroscopic data

The optical spectra of AS 78 and MWC 657 look very similar. However, the emission lines are stronger in MWC 657. Both objects show Balmer emission lines with very steep decrements, and a large number of Fe II lines in emission. The strongest Fe II lines (4923, 5018, and 5169 Å) have P Cyg-type profiles in AS 78 and in the SAO spectrum of MWC 657. Forbidden emission lines of O I are found in the spectra of both objects, while no [Fe II] lines are detected. The absorption features are mainly diffuse interstellar bands (hereafter DIBs). The only features which can be assigned to the photosphere are He I lines (and the Ne I 6402 Å line in the spectrum of AS 78). However, in the OHP 1995 and SAO spectra of MWC 657 these are seen partly in emission. A catalog by Coluzzi (1993) was used to identify the spectral lines. Characteristics of the identified lines in the spectra of both objects are listed in Table 7 (Balmer lines only), Table 8 (lines in the near-IR spectrum of AS 78), and Table 9 (other lines in the optical region). For MWC 657 the information from the OHP 1994 and SAO spectra is given in Table 9, because the OHP 1995 spectrum was obtained with a lower signal-to-noise ratio. The most informative parts of the low-resolution spectra of both objects are shown in Fig. 3 (AS 78) and 4 (MWC 657), while those of the high-resolution spectra are shown in Fig. 5 (both objects). An important spectral region containing the He I 5876 Å line and the Na I D_1,2 lines is presented in Fig. 6. The variations of the He I 5876 Å line in the spectrum of MWC 657 are shown in Fig. 7. The theoretical He I line 5876 Å profiles presented in Figs. 6 and 7 were calculated using the code Synspec (Hubeny et al. 1995) on the basis of the Kurucz (1994) model atmospheres and solar abundances.

$\begin{figure} \par\resizebox{7cm}{!}{\includegraphics{1872_f06.eps}}\par\end{figure}$

Figure 6: The He I 5876 Å and Na I D_1,2 lines in the high-resolution SAO spectra of AS 78 and MWC 657. The theoretical profile for $T_{\rm eff}$ = 17 000 K, log g = 2.5, and vsin i = 70 kms^-1 is shown by a dotted line for AS 78. The absorption components of the Na I lines in the spectrum of AS 78 are marked by short solid lines, while the telluric water vapour absorption lines are marked by dots. Units of the wavelengths and intensities are the same as in Fig. 3

$\begin{figure} \par\resizebox{7cm}{!}{\includegraphics{1872_f07.eps}}\par\end{figure}$

Figure 7: The He I 5876 Å line profile in the high-resolution spectra of MWC 657. The theoretical profile for $T_{\rm eff} =15\, 000$ K, log g=3.0, and vsin i = 100 kms^-1 is shown by a dashed line. The velocities are given in kms^-1. The intensities are normalized to the continuum. The spectra are shifted with respect to each other by 0.35 $I_{\rm cont}$

$\begin{figure} \par\resizebox{7cm}{!}{\includegraphics{1872_f08.eps}}\par\end{figure}$

Figure 8: The H $\alpha$ line profiles in the spectrum of AS 78 obtained with different resolutions. The spectrum of 1994 January 5 is shown by a dashed line, that of 1994 January 21 by a solid line, while that of 1999 July 28 by a dotted line. Units of the wavelengths and intensities are the same as in Fig. 3

Despite the above-mentioned similarities, the objects' spectra are different in details. Thus we will discuss the spectral features of these objects separately.

Table 7: Balmer lines in the spectra of AS 78 and MWC 657

Name Line $V_{\rm blue}$ $I_{\rm blue}$ $V_{\rm abs}$ $I_{\rm abs}$ $V_{\rm red}$ $I_{\rm red}$ EW Rem.

MWC 657 H $\alpha$ -203 6.48 -115 1.83 -35 37.28 172.7 OHP 1994

-249 8.22 -118 1.46 -36 46.68 198.0 OHP 1995

-208 22.00 -130 8.30 -30: $\sim$ 80 SAO 1999 $^{\rm a}$

H $\beta$ -283 1.90 -135 0.42 +6 9.83 38.5 OHP 1994

-238 1.46 -125 0.46 -21 8.96 33.3 OHP 1995

-220 3.45 -150: 1.30: -25: 12.00: 48.6 SAO 1999

H $\gamma$ -315 1.38 -141 0.13 +16 4.26 16.9 OHP 1994

AS 78 H $\alpha$ 6.4 77 1994 Jan. 5

H $\alpha$ 11.9 74 1994 Jan. 21

H $\beta$ 2.8 10 1994 Jan. 21

H $\gamma$ 1.4 5 1994 Jan. 21

H $\alpha$ -570 1.18 -350 0.06 -23 29.00 117.5 1999 Jul. 28

H $\beta$ -330 0.05 -12 7.10 28.8 1999 Jul. 28

**Table 7:** Balmer lines in the spectra of AS 78 and MWC 657
Name	Line	$V_{\rm blue}$	$I_{\rm blue}$	$V_{\rm abs}$	$I_{\rm abs}$	$V_{\rm red}$	$I_{\rm red}$	EW	Rem.
MWC 657	H $\alpha$	-203	6.48	-115	1.83	-35	37.28	172.7	OHP 1994
		-249	8.22	-118	1.46	-36	46.68	198.0	OHP 1995
		-208	22.00	-130	8.30	-30:	$\sim$ 80		SAO 1999 $^{\rm a}$

	H $\beta$	-283	1.90	-135	0.42	+6	9.83	38.5	OHP 1994
		-238	1.46	-125	0.46	-21	8.96	33.3	OHP 1995
		-220	3.45	-150:	1.30:	-25:	12.00:	48.6	SAO 1999

	H $\gamma$	-315	1.38	-141	0.13	+16	4.26	16.9	OHP 1994

AS 78	H $\alpha$						6.4	77	1994 Jan. 5

	H $\alpha$						11.9	74	1994 Jan. 21
	H $\beta$						2.8	10	1994 Jan. 21
	H $\gamma$						1.4	5	1994 Jan. 21

	H $\alpha$	-570	1.18	-350	0.06	-23	29.00	117.5	1999 Jul. 28
	H $\beta$			-330	0.05	-12	7.10	28.8	1999 Jul. 28

$^{\rm a}$ The red peak of the H $\alpha$ line is saturated.

3.2.1 AS 78

The Balmer lines (H $\alpha$ - H $\delta$ ) have P Cyg-type profiles in both high- and low-resolution spectra of AS 78, indicating the presence of a strong stellar wind. The weak blueshifted peak and broad wings seen in the high-resolution H $\alpha$ profile suggest a significant amount of electron scattering. The emission components of the higher members of the series are seen together with the broad photospheric lines (see Fig. 3). The equivalent width of the H $\alpha$ line was nearly 75 Å in 1994 and 117.5 Å in 1999. The central intensity of the Balmer lines strongly depends on the resolution (see Table 7 and Fig. 8). This is partly due to the narrow emission component of the line, but so remarkable a change in the line strength between 1994 and 1999 can also be explained by variations of the stellar wind. The change in the Balmer decrement (log $I_{{\rm H}\alpha}/I_{{\rm H}\beta}$ = 0.95 in 1994 and 0.69 in 1999) confirms the latter suggestion.

The sodium D_1,2 lines consist of one emission and three absorption components (see Fig. 6). The strengths of the two deepest components, which are barely resolved at our resolution and certainly have an interstellar origin, are consistent with the high reddening evident from the photometry. Their radial velocities (hereafter RV), -4 and -40: kms^-1, suggest that the star belongs to the Perseus arm. Munch (1957) has shown that in this direction of the galactic plane such components are seen in the spectra of stars located farther than 2 kpc from the Sun. The emission components of the D-lines seem to have P Cyg-type profiles, which are very similar to those of the strongest Fe II lines.

Nearly 10 DIBs are found in the spectrum of AS 78. Since their fine structure is not resolved in our spectrum, we are only able to roughly estimate where they are formed. The mean DIBs RV of about -12 kms^-1 is consistent with their formation between the Sun and Perseus arm. The strength of some DIBs in stellar spectra display a good correlation with the star's reddening (e.g. Herbig 1993). We use this property to obtain an independent estimate of E_B-V. The equivalent widths of the DIBs at 5780 and 5797 Å, 0.43 and 0.11 Å, respectively, correspond to E_B-V = 0.9 (Herbig 1993), which is in good agreement with the dereddened optical colour-indices.

Only a few absorption lines of non-interstellar origin are seen in the spectrum of AS 78. The He I lines (4713, 5876, 6678, 7065, and 10830 Å) are rather weak, while all except for the IR line, which is not resolved, show emission components superimposed on the underlying absorption line (see Fig. 6). Thus these lines are most likely affected by the envelope emission and cannot be used to derive the star's spectral type. At the same time, the line wings suggest a rotational velocity of nearly 70 kms^-1. Another photospheric line, Ne I 6402 Å, is also very weak (EW = 0.03 Å, see Fig. 5). It is much weaker than those of B-type supergiants, reviewed in Miroshnichenko et al. (1998), irrespective of their sub-type and the presence of emission lines. This fact may indicate that the luminosity of AS 78 is not very high; however, this line may also be contaminated with the envelope emission.

The systemic velocity may be estimated by measuring the RV of the Fe II lines (e.g. Humphreys et al. 1989). The mean RV, determined by using 21 Fe II lines, is -41 $\pm$ 1 km s^-1 and corresponds to D = 2.9 $\pm$ 0.1 kpc from the Sun, which is based on the galactic rotation curve (Dubath et al. 1988). The mean RV of the He I lines, -35 $\pm$ 2 kms^-1, which may be less affected by the envelope velocity field, is close to the result obtained for the Fe II lines. However, since RV were measured for only three He I lines (see Table 9), the distance estimate based on them may introduce a larger uncertainty. In any case, the spectroscopic result does not contradict the photometric distance estimate (see Sect. 3.1.1). Note that the distance (2.9 kpc) towards BD $+55\hbox {$^\circ $ }837$ , the closest star to AS 78 in Table 5 (and which also has RV=-40 kms^-1, Rubin 1965), determined from the galactic rotation, is in good agreement with the spectroscopic estimate (3.5 kpc). Assuming D=2.9 kpc for AS 78, V = 11.2 mag, A_V = 2.6 mag and spectral type B3 (see Sect. 3.1.1), one can derive M_V = -3.7 mag and log $L_{\rm bol}/L_{\odot}=4.0$ . Thus, the luminosity of AS 78 turns out to be intermediate between those of luminosity classes II and III (Strajzhys & Kurilene 1981), unless we significantly underestimated the distance towards it.

The low-resolution near-IR spectrum of AS 78 contains emission lines of neutral hydrogen, singly ionized nitrogen, calcium, oxygen, and iron (see Table 8 and Fig. 9). The only line with a strong absorption component identified in this spectrum is that of He I 1.083 $\mu$ m, which also has a weak, but noticeable, emission component, fairly unusual at this resolution. The IR triplet of Ca II has an averaged intensity of 1.25 of the continuum flux. A weak Br $\gamma$ and a few higher members of this series are detected in emission. The flux ratio of the O I lines at 8446, 11 287, and 13 164 Å can be used for an independent interstellar reddening estimate by means of the method suggested by Rudy et al. (1991). This method assumes that an equal number of photons are generated in the 8446 and 11 287 Å lines by the Ly $_{\beta}$ fluorescence mechanism, but corrects this ratio, by using the 13 164 Å line, for the additional photons generated in the 8446 Å line by continuum fluorescence. The flux ratio observed in the spectrum of AS 78 (0.344/0.322/0.039) gives E_B-V=0.70 $\pm$ 0.15, which agrees with the above photometric estimate.

The continuum level in the blue part of the near-IR spectrum decreases with wavelength, while it increases in the red part. This fact implies a change in the continuum formation mechanism. Shortward of 1.2 $\mu$ m it is due to the radiation of the star and the stellar wind. Extrapolation of the blue continuum slope to the red part of the spectrum shows that the observed continuum is 1.7 times stronger in the H-band and 5 times stronger in the K-band. Such a sharp rise cannot be explained by the presence of a cool companion, but rather by radiation of the circumstellar dust.

Table 8: Lines in the near-IR spectrum of AS 78

$\lambda_{\rm lab}$ ID $I/I_{\rm c}$ $\lambda_{\rm lab}$ ID $I/I_{\rm c}$

8138.59 [Ti II] (7F) 1.07 10049.38 P7 1.45

8203.572 P $^{\rm a}$ 1.10 10459.79 N II (11) 1.09

8446.35 O I (8) 1.27 10494.00 [Cr II] (28F)? 1.07

8498.018 Ca II (2) 1.17 10501.50 Fe II 1.05

8542.089 Ca II (2) 1.30 10683.08 C I 1.20

8598.394 P14 1.10 10830.34 He I (1) $^{\rm b}$ 0.72

8662.140 Ca II (2)+P13 1.25 10862.64 Fe II 1.08

8750.00 P12 1.10 10938.09 P6 1.65

8862.787 P11 1.13 11125.44 Fe II 1.06

9017.911 P10 1.20 11287.0 O I 1.27

9061.33 Fe II (71) 1.10 11750.0 C I 1.12

9229.017 P9 1.25 12818.05 P5 1.95

9381.78 [Cr II] (23F) 1.17 13164.8 O I 1.05

9403.36 Fe II (71) 1.08 16109.31 Br13 1.05

9545.974 P8 1.14 16407.19 Br12 1.12

9590.94 [Cr II] (16F) 1.08 16806.52 Br11 1.10

9663.588 Ca II (55) 1.07 17362.11 Br10 1.11

9997.58 Fe II 1.08 21655.29 Br $\gamma$ 1.16

**Table 8:** Lines in the near-IR spectrum of AS 78
$\lambda_{\rm lab}$	ID	$I/I_{\rm c}$	$\lambda_{\rm lab}$	ID	$I/I_{\rm c}$
8138.59	[Ti II] (7F)	1.07	10049.38	P7	1.45
8203.572	P $^{\rm a}$	1.10	10459.79	N II (11)	1.09
8446.35	O I (8)	1.27	10494.00	[Cr II] (28F)?	1.07
8498.018	Ca II (2)	1.17	10501.50	Fe II	1.05
8542.089	Ca II (2)	1.30	10683.08	C I	1.20
8598.394	P14	1.10	10830.34	He I (1) $^{\rm b}$	0.72
8662.140	Ca II (2)+P13	1.25	10862.64	Fe II	1.08
8750.00	P12	1.10	10938.09	P6	1.65
8862.787	P11	1.13	11125.44	Fe II	1.06
9017.911	P10	1.20	11287.0	O I	1.27
9061.33	Fe II (71)	1.10	11750.0	C I	1.12
9229.017	P9	1.25	12818.05	P5	1.95
9381.78	[Cr II] (23F)	1.17	13164.8	O I	1.05
9403.36	Fe II (71)	1.08	16109.31	Br13	1.05
9545.974	P8	1.14	16407.19	Br12	1.12
9590.94	[Cr II] (16F)	1.08	16806.52	Br11	1.10
9663.588	Ca II (55)	1.07	17362.11	Br10	1.11
9997.58	Fe II	1.08	21655.29	Br $\gamma$	1.16

$^{\rm a}$

$^{\rm b}$ The line has a P Cyg-type profile with an emission component at 1.07 $I_{\rm cont}$ .

3.2.2 MWC 657

In contrast with those of AS 78, the Balmer lines of MWC 657 show a double-peaked structure with a deep central depression (see Fig. 10). Such behaviour suggests that the lines are formed in a dense circumstellar disk, which is viewed close to edge-on. The intensity ratio of the blue and red peaks (V/R) in the H $\beta$ line profile is nearly 1.5 times larger in the 1999 spectrum compared to those in 1994/5. The same effect is probably observed in the H $\alpha$ line as the blue peak strength in the 1999 spectrum is at least 3 times larger than that in 1994/5, while this number should not be larger than 1.5 for the red peak (a rough estimate from the unsaturated part of the profile). The variations of the H $\alpha$ and H $\beta$ profiles in the spectrum of MWC 657 are shown in Fig. 10. The low-resolution DSO spectra reveal the presence of a DIB at 4430 Å and a number of Fe II lines in the region blueward of 4735 Å, the blue edge of the high-resolution SAO spectrum (see Fig. 4). The Balmer lines up to H $\epsilon$ are seen in pure emission; while the shortest wavelength Balmer lines observed, H8 and H9, partially fill-in their absorption cores.

Since the emission-line spectrum of MWC 657 is much stronger than that of AS 78, no reliable information about the star's parameters can be derived from the few He I lines detected in absorption. Moreover, in the 1995 spectrum a part of the He I 5876 Å line is seen in emission, as well as both He I 5876 and 6678 Å in the 1999 spectrum (see Fig. 7). The fact that no He II 4686 Å line is present puts an upper limit of 26-2710³ K on the disk's excitation source temperature. On the other hand, the occasional emission in the He I lines implies that it is $\ge$ $15\, 000$ K.

The presence of the strong DIBs and deep interstellar components of the Na I D_1,2 lines is consistent with the high reddening estimated by Miroshnichenko et al. (1997). The equivalent widths of the DIBs at 5780 and 5797 Å, 0.51 and 0.15 Å respectively, lead to an estimate of the interstellar reddening of $E_{B-V} \sim$ 1.1 mag, which is noticeably smaller than the 1.6 mag derived from the observed and intrinsic (for an early B-type star) colour-indices. This discrepancy indicates that part of the reddening is circumstellar. Such circumstellar reddening is usually observed in classical Be stars (e.g. Waters et al. 1987). In the case of MWC 657, this points out that the circumstellar gas density distribution decreases more slowly with distance from the star than that of AS 78. The emission peak separation ( $\sim$ 200 kms^-1) is typical for classical Be stars, which may have Keplerian disks (e.g. Hummel 2000). On the other hand, the latter display significantly weaker emission-line spectra, indicating that MWC 657 has a higher average disk density.

The mean RV determined from the Fe II line measurements is -55 $\pm$ 5 kms^-1 in the OHP and SAO spectra. According to Dubath et al. (1988) this corresponds to a distance of 4.5 $\pm$ 0.5 kpc, while the distance to the neighboring OB-association Cep OB1, which has an averaged RV of -60 kms^-1, is 3.5 kpc (Humphreys 1978). The interstellar extinction law in the direction of MWC 657 suggests $D \sim$ 2-3 kpc for the interstellar part of the object's reddening (Miroshnichenko et al. 1997). Thus, D = 3.5 kpc seems to be a reasonable estimate for the distance towards MWC 657. The averaged optical brightness (V = 12.8 mag), the minimum star's temperature ( $15\, 000$ K), and the estimated interstellar reddening and distance yield the absolute visual luminosity $M_{V} \sim -$ 3.3 mag, assuming that MWC 657 is a single star and the circumstellar contribution to its optical brightness is small. According to Strajzhys & Kurilene (1981), the bolometric correction for such a star is BC $\sim$ -1.3 mag, leading to log $L_{\rm bol}$ / $L_{\odot} \le$ 3.7. The uncertainty of this estimate, due to the uncertainties in the distance and the optical brightness, is at least 50 per cent. In addition, a higher temperature would result in a higher luminosity, although this may be partially canceled by a reduction for the possible circumstellar continuum emission.

The spectroscopic data have been obtained at different phases (the OHP 1994 spectrum was taken at phase 0.84, the OHP 1995 at 0.89, the DSO 1998 at 0.95, the SAO 1999 at 0.33, while the DSO 1999 spectrum was obtained at phase 0.46), but mainly close to the photometric minimum. Nevertheless, the observed line profile changes seem to suggest that the periodicity is real. On the other hand, the last spectrum obtained at DSO nearly at the photometric maximum turned out to be almost identical to the first DSO spectrum. This result may indicate that the variations are difficult to detect at this resolution, or that the object's variability is more complicated than simple cyclicity.

3.3 Modelling

Since the circumstellar envelope of MWC 657 is most likely non-spherical and very dense, we did not model its Balmer line profiles, as the modelling of such structures is not yet well developed (e.g. Hummel 2000). The SED of MWC 657 has been modelled by Miroshnichenko et al. (1997). In this section we concentrate on modelling the SED and Balmer line profiles of AS 78.

The near- and far-IR excesses observed in the SED of AS 78 suggest that the object is surrounded by a dusty envelope. In order to model the dusty environments we constructed the SED using our photometry, the MSX, and the IRAS fluxes. The models were calculated with a code by Ivezic et al. (1999), which solves the radiative transfer problem in a spherical dusty envelope. The temperature structure of the envelope is calculated self-consistently taking into account absorption, emission, and scattering of the stellar radiation in the envelope.

$\begin{figure} \par\resizebox{14cm}{!}{\includegraphics{1872_f10.eps}}\par\end{figure}$

Figure 10: The H $\alpha$ (left panel) and H $\beta$ (right panel) line profile variations in the spectrum of MWC 657. The OHP 1994 spectrum is shown by a solid line, that of OHP 1995 by a dashed line, while that of SAO 1999 by a dotted line (the redshifted peak of H $\alpha$ is saturated). Units of the velocities and intensities are the same as in Fig. 7

The calculations were performed for stellar temperatures $T_{\rm eff}$ in the range from $16\, 000$ K to $22\, 000$ K, which corresponds to B2-B5 spectral types (Strajzhys & Kurilene 1981). The stellar radiation was modeled with Kurucz (1994) model atmospheres. The dust temperature at the inner envelope's boundary ( $T_{\rm in}$ ) was varied between 1000 and 1500 K. Since no reliable information about the 10- $\mu$ m silicate feature is available, we used the optical properties of two different grain chemical compositions: interstellar dust (Mathis et al. 1977, hereafter MRN) and amorphous carbon (Hanner 1988). The density distribution in the envelope was chosen to be a power-law $r^{-\alpha}$ , where r is the distance from the star. The steep decrease of the far-IR fluxes implies a relatively small ratio ( $Y_{\rm out}$ ) of the outer to inner radii and a small optical depth of the envelope ( $\tau_V$ ); these are two other parameters of the modelling. Since both the interstellar and circumstellar contributions to the overall extinction are not known a priori, we treated A_V as an additional free parameter and derived it by comparing the observed and modelled SED.

The optical part of the SED is dominated by the star's radiation, so that $T_{\rm eff}$ controls the A_V. The best fit with A_V = 2.86 mag, which is in accord with the above independent estimates, is found for $T_{\rm eff}$ = $17\, 000$ K, $\tau_V$ = 0.12, $Y_{\rm out}$ = 1000, $T_{\rm in}$ = 1200 K, a power-law density index $\alpha$ = 1.6, and amorphous carbon grains. The theoretical SEDs for the MRN mixture give a much poorer fit. Changes in the stellar surface gravity do not affect the quality of the fits. The observed SED, dereddened with A_V=2.86 mag, and the corresponding best fits for both types of dust are presented in Fig. 11 in units of the bolometric flux.

Despite the fact that the spherical envelope fit is close to the observed SED, such a geometry does not solve the near-IR variation problem. It is hard to imagine that the star lost a significant part of its dusty envelope in a few weeks (between 1999 August 28 and September 17). The distance between the star and the dust, which radiates the bulk of 2- $\mu$ m emission, is several hundreds of stellar radii. This requires a dramatic decrease in the star's luminosity or a matter outflow with a speed of about 0.01c to push the dust outwards. Both these suggestions seem to be unlikely, given the star's long-term stability in the optical region.

A possible solution of this problem is the suggestion that the dust is distributed non-spherically. Recently Miroshnichenko & Corporon (1999) showed that a SED, similar to what we see in AS 78, in the region longward of 1 $\mu$ m can be produced by an optically-thick and geometrically-thin disk. Such a disk would have an outer temperature of a few hundred Kelvin and would be much smaller than the spherical envelope producing the same SED, because the disk is much cooler due to its high optical depth. The latter property makes the disk spectrum flat in the 10- $\mu$ m region. However, a weak 10- $\mu$ m emission could exist, because some dust may still be present in the form of a spherical envelope. We might speculate about a suggested disk surrounding a secondary companion, the luminosity of which is not large enough to make a noticeable contribution to the observed radiation of the system. This might explain why we do not see brightness modulation in the optical region. Such a disk's contribution to the radiation in the J and H bands would be rather small, but would become dominant longward of approximately 2 $\mu$ m. Its effective radius for the 2 $\mu$ m radiation is a few radii of the secondary, so that it can be occulted easily by the primary. The photometric variations should be smaller at longer wavelengths, where the disk radiates at larger radii. Thus, the binary hypothesis is capable of explaining both the IR excess and the near-IR variations. However, current information about the possible secondary is insufficient to model the SED of such a binary.

In order to put further constraints on the physical characteristics of the star and to estimate the parameters of its wind, we tried to calculate the Balmer line profiles by means of the method described by Pogodin (1986) for spherically symmetric envelopes. The spherical model usage is justified by the P Cyg-type profiles, which can be produced either in a spherical or a geometrically-thick disk-like envelope. Some initial ideas about the wind kinematics can be obtained from comparing the object's line profiles with those of well-studied stars, which display similar emission-line spectra. Recently the results of the analysis of the Balmer and He I lines of P Cyg and HDE 316285 have been published (Najarro et al. 1997 and Hiller et al. 1998, respectively). The line profiles of both stars were well fitted with the same $\beta$ -velocity law ( $\beta$ = 2.5), with wind terminal velocities of 185 (P Cyg) and 410 km s^-1 (HDE 316285), and with the wind density parameters ( $\dot M\,R_{*}^{-1.5}$ ) of 4.610^-8 (P Cyg) and 3.710^-7 (HDE 316285) $M_{\odot }$ yr^-1 $R_{\odot}^{-1.5}$ . Comparison of the H $\alpha$ profiles of the three objects shows that their emission component cores have almost the same blue and red slope which suggests $\beta$ = 2.5 for AS 78 (see Fig. 12). The wind terminal velocity of AS 78 is nearly 500 kms^-1 based on the red border of the absorption feature. The full width at half maximum (FWHM) of the emission component of AS 78 is close to that of P Cyg, while the absorption component of AS 78 is much wider than those of both P Cyg and HDE 316285. The latter fact leads to the suggestion that the acceleration in the outer wind regions of AS 78 is significantly smaller than in those of the other two stars and that the $\beta$ -velocity law does not work there. The wind density parameter does not seem to be smaller than that of P Cyg, since both the H $\alpha$ and H $\beta$ lines in AS 78 are stronger than those of P Cyg. This suggestion leads to a rough mass loss rate estimate of $\dot M \ge$ 1.510^-6 $M_{\odot }$ yr^-1.

$\begin{figure}\par\resizebox{7cm}{!}{\includegraphics{1872_f11.eps}}\par\end{figure}$

Figure 11: Fitting the SED of AS 78. The dereddened SED of AS 78 normalized to the bolometric flux is shown by filled circles (averaged Tien-Shan, CST, and Lick data), filled triangles (IRTF data), open circles (MSX data), and open squares (IRAS data). The best fit for MRN dust is shown by a dashed line, and for amorphous carbon dust by a solid line. The SED of a star with $T_{\rm eff}$ = $17\, 000$ K and log g = 3.0 (Kurucz 1994) is shown by a dotted line

Here we try to check the above estimates on the low-resolution line profiles (the SAO 1994 data). The kinematics of the stellar wind was described through the $\beta$ -velocity law of Cohen & Barlow (1977): $w(x)=w_0+(1-w_0)\,(1-1/x)^{\beta}$ , where $w(x)={v(x)/v_{\infty}}$ , w₀=w(x=1), v(x) is the velocity at the dimensionless distance x from the stellar center, which is expressed here in stellar radii, and $v_{\infty}$ is the terminal velocity of the stellar wind. Stellar radiation and photospheric line profiles were introduced using Kurucz (1979) model atmospheres. The wind temperature was assumed constant throughout the envelope at a level of 80% of $T_{\rm eff}$ (Lamers & Waters 1984). The photospheric line profiles were broadened by the star's rotation of 70 kms^-1 (see Sect. 3.2.1). For each model the radial electron scattering optical depth ( $\tau_{\rm e}$ ) was calculated, and the emergent line profiles were corrected for electron scattering using the method by Castor et al. (1970). For comparison with the observations, we convolved the resulting profiles with an instrumental Gaussian profile in order to achieve a resolution of 2 Å.

With these assumptions we calculated a grid of models for different $T_{\rm eff}$ , $\dot M\,R_{*}^{-1.5}$ , log g, $\beta$ , w₀, and $v_{\infty}$ . The best result was obtained for the following parameter values: $T_{\rm eff}$ = $17\, 000$ K, log g = 3.0, $v_{\infty}$ = 350 kms^-1, w₀ = 0.15, $\dot M\,R_{*}^{-1.5}$ = $3.8\,10^{-8}\ M_{\odot}$ yr^-1 $R_{\odot}^{-1.5}$ , $\beta$ = 4.5, $\tau_{\rm e}$ = 0.5. These fits are shown in Fig. 13. The star's radius, calculated from the above bolometric luminosity estimate and $T_{\rm eff}$ , turns out to be 11.5 $R_{\odot}$ . The wind density parameter and R_* give the mass-loss rate estimate $\dot M$ = 1.510^-6 $M_{\odot }$ yr^-1, which is in good agreement with the lower level derived from the comparison with P Cyg.

The observed absorption components of the H $\gamma$ and H $\delta$ lines are wider than the theoretical ones at the blue side. This can be due to the poor resolution, simplicity of the model, and/or blending with nearby weak absorption lines (e.g., with those of O II Mult. 18 which are located at -400-600 kms^-1 from the center of the H $\delta$ line). Nevertheless, the agreement between the observations and model results is good especially for the H $\beta$ line. The models with larger $T_{\rm eff}$ have shallower absorption components in the H $\beta$ - H $\delta$ lines with respect to the observed ones. A similar effect is seen for the models with a larger wind density parameter $\dot M\,R_{*}^{-1.5}$ , which leads to a larger $\tau_{\rm e}$ . Smaller values of w₀also produce the same effect, while a smaller $\beta$ results in shallower Balmer decrements.

These modelling results show that in 1994 the wind of AS 78 probably had a lower acceleration than in 1999. The difference in the Balmer decrement noticed above and the results of the H $\alpha$ profile comparison with those of other stars (Sect. 3.2.1) indicate that the wind structure has changed during the period between our observations. Qualitatively these changes can be understood as follows: the wind acceleration has increased, and the denser regions near the star's surface were moved outwards resulting in a wider P Cyg-type absorption component. The mass-loss rate has most likely increased as well. This scenario requires a characteristic time of nearly a year, since this is roughly the time for the gas to travel all the way through the envelope outer boundary ( $\sim$ 100 R_*) at a velocity of 100 kms^-1. At this point, we will not attempt to model our high-resolution data until new spectra are obtained.

3 Results

3.1 Photometric data

3.1.1 AS 78

3.1.2 MWC 657

3.2 Spectroscopic data

3.2.1 AS 78

3.2.2 MWC 657

3.3 Modelling