On the basis of the identifications given in Table 2 we shall discuss the elements present in the spectrum of the star, ordered by atomic number.

The Balmer series is present between 3736 and 5120, represented by H 12 - H 4 in absorption. The intensities from H 12 to H 8 follow a regular pattern, decreasing afterwards. In 1991, in H 6 one observes a wide absorption line accompanied by a strong emission redshifted by about 250 km/s from the line center. A similar structure, but much fainter, is also seen in H 7. In H 5 the redshifted emission is much stronger; the shift being of 220 km/s.

On the spectrum from 1993 the redshifted emission component (250 km/s from the line center) is well visible in H 6, wheras in H 7, 8 and 9 only a deformation in the red wing is seen.

Figure 1: Reproduction of CCD spectra of HD 45677. Abscissae: wavelengths in A. Ordnates: Intensities. The continuum level is set to unity. Important lines are identified. The spectra reproduced correspond to the following dates: 3900 - 4336, 23-10-1991 4292 - 4721, 25-10-91 4702 - 5135, 25-10-1991 7573 - 7991, 07-01-1990 8369 - 8779, 21-2-1992 9838 - 10225. 03-01-1990

In H 4 (observed only in 1991) there exists a wide (3900 km/s) and shallow absorption line upon which a strong emission (W=5.96 Å) is placed. A blue and fainter emission component is also visible at

from the emission line center. A sharp and intense blue shifted absorption at

from the emission line center is also present. We are thus in presence of a rather complicated structure.

The Paschen lines observable in the wavelength range covered by us run from 25 to 7 (except 8), but the lines 25-22 were only observed in 1990. From P 21 to P 12, the lines follow a regular pattern. Whereas in 1990 and 1993 the equivalent widths were rather similar (differences less than 10%), in 1992 the W's are systematically weaker by about 20%. The lines appear symmetric on all spectra, except in 1992 when double structures can be appreciated in P13 - P19. The average separation between peaks is of the order of 110 km/s. It should be noticed that P 14 is double even in the years when all other Paschen lines are simple, and this is due to NI 8595.51.

Neutral helium is well represented by absorption lines from the

series and perhaps by two lines from

. However in the latter series 5047 is only doubtfully present, so that we regard the presence of this series as dubious.

In general the helium lines are simple pure absorption profiles, but on one spectrum from 1991 we find two lines with abnormal profiles. In this year, both 4920 and 5017 have a P Cyg profile, with a displacement of 190 km/s between the absorption and the emission peaks. There exists further a violet absorption component, displaced 250 km/s from the central absorption.

C I is represented by lines from M.3, as well as by one line (9849) of [CI], all in emission. C II is represented by two weak features from M.4 and 6, both in absorption.

Neutral nitrogen is represented by many lines from M.1, 7, 8, 9, 15, 18, 19 and 174, all in emission. We do not observe N II, but in our wavelength regions no strong lines are expected.

Neutral oxygen is represented by emission lines from M.1, 4, 8 and 54. The strength of M.4 is surely due to fluorescence effects from Lyman alpha. Its intensity has a minimum in 1992 (1.8 Å), as compared to 24 (1990) and 27.3 (1993) Ionized oxygen is represented by emission lines from M.50 and 97.

Neutral magnesium is represented by a weak line of M.1, in emission. Ionized magnesium is represented by emission lines of M.1 and 8, whereas the strong M.4 (4481) is seen in absorption. The latter line is however flanked by two symmetrical emission peaks.

Ionized silicon is represented by M.3 (4128 - 30) in absorption and by the high excitation lines at 4200.6 and 4200.8, in emission.

Ionized calcium is present with a weak absorption line 3933 from M.1. The other line, 3968 is lost in the structure of H 7. Because of the lack of an enhancement of P 13, 15 and 16 with regard to other lines of the Paschen series, we infer that no infrared triplet Ca II lines are present.

Ionized scandium is represented by two weak emissions from M.15 and by one forbidden line from M.1, 8650. The element seems thus to be present.

Neutral titanium is represented by two lines from M.149 and by two weak lines from M.13 of [Ti 1]. Ionized titanium is present under the form of [Ti II] from M.5, 6, 10, 23 and 25.

Ionized vanadium is present with the two strongest lines from M.56. [V II] is represented by lines from M.4 and 11.

Ionized chromium is represented by the strongest lines from M.44 and [Cr II] by lines from M.4, 10 and 15.

Ionized manganese is represented by only one line from an infrared multiplet. (Thackeray & Velasco 1976) The other lines of the multiplet are blended.

This is the element which has the largest number of lines both in absolute as well as in relative number (36%). Ionized iron is represented by lines from M.27, 28, 37, 38 and 43. [Fe II] is represented by lines from M.3, 4, 5, 6, 7, 8, 13, 14, 19, 20, 21, 23, 30, 35, 37 and 41. [Fe III] is represented by one weak line 5059, from M.1.

Ionized nickel is present in the form of [Ni II], with two weak lines from M.2 and 3.

Ionized copper is present in the form of [Cu II], with one weak line from M.1. (4375).

Ionized zirconium is represented by the strongest line from M.88 and, under the form of [Zr II] by 10083 from M.13.

The most extended line identification work on this star was carried out by Swings (1973), based upon material covering the region

with a variety of dispersions, obtained in the years 1962-72. He lists about 200 features, number to which iron contributes with about 45% of the lines.

In our work we list about 235 lines in a shorter interval (3300 Å), and 35% of these correspond to iron. The percentage of unidentified features is about the same in both cases.

**Table 3:** Elements identified by different authors
	PW	Swings	Sw+Str	Gr	Isr

		1973	1943	1993	1996
H	x	x	x
He I	x	x	x
C I	x
[I]	x
II	x		x	x	x
IV				x
N I	x	x
II			x		x
[II]		x
O I	x	x
[I]		x	x
II	x		x
Na I					x
Mg I	x
II	x	x	x
Al III				x
Si II	x			x
IV				x
S II					x
[II]	x	x
Ca II	x	x	x
Sc II	x
[II]	x
Ti I	x
[I]	x
II	x	x
V II	x
Cr II	x	x
[II]	x	x
Mn II	x	x
Fe II	x	x	x	x
[II]	x	x	x
III				x
[III]	x
Co II		?
Ni II		x
[II]	x	x
Zn II		x
Cu [II]	x	x
Zr II	x
[II]	x

The results of a comparison of the elements identified by different authors is given in Table 3 (click here).

Some comments upon the differences between the two most extended identifications - the present work and Swings (1973) - seem to be relevant. The main differences are due to our lack of coverage of the 5100-7000 region which prevents us from observing the presence of Na I, [N II] and [O I]. Our non observation of Mn II and Ni II is mostly due to our lack of coverage of the region below 3750, where Swings found most of the lines. The elements which we found and he did not, are most likely due to genuine variability of the shell. For instance he did not observe C II 4267 although this line was present in other epochs. A similar case can be made for O II and Si II. This implies that the lines of some elements appear and disappear over the years. We should recall that this has also been found in other B[e] stars investigated in this series, so that the phenomenon seems to be a general characteristics of B[e] stars and similar objects.

4. Elements present