After the first paper of Swings & Struve (1943), many authors have shown that the lines vary in strength and in structure. We shall not recapitulate all facts discovered, except to summarize saying that up to now no period or cycle has been discovered. This is true for the total light, as well as for line structures and line strengths (intensities or equivalent widths). For more details, see for instance Swings et al. (1980) and Israelian et al. (1996).

In what follows we shall report on the variations of the equivalent widths of the lines in our material. We stress the point that despite the large amount of descriptions available, our measures represent the first attempt to collect objective measures of line strengths.

The lines observed on different occasions are listed in Table 4 (click here) and displayed in Figs. 2 (click here) and 3 (click here). Table 4 (click here)a and Fig. 2 (click here) show that between 1991 and 1993 no large changes have occurred, although He I is slightly weaker in 1993. The same is true for Si II, whereas Ca II is stronger in 1993. Let us observe that both Si II and Ca II have shown large variations in the past. Si II 4130 was not seen by Swings (1973) and Ca II has undergone changes from very strong and neat in 1943 to the weak line we see in 1991.

Figure 2: Equivalent widths of different lines in 1991 and 1993, in the region 3896 - 4189. Empty circles stand for absorption lines, full circles for emission lines

If we examine the results of Table 4 (click here)b and Fig. 3 (click here) we see immediately that the Paschen lines had a minimum in 1992 and came back to previous strength in 1993. O I 8446 also had a minimum in 1992, a fact which is in line with the weakening of the hydrogen lines. For the lines of N I, [Fe II] and [V II] there might be a minimum, but it is rather uncertain. What is remarkable is that the Paschen lines seem to be much less sensitive to the profile changes which are observed in the Balmer lines. In fact we cannot compare the equivalent widths of H7 and H 6 from 1991 and 1993 because of the changes in profile, whereas we see no obvious structure in the Paschen lines. We add that the two spectra of 1990 were taken at a three days interval, so that they can be used to obtain an impression of the precision of the equivalent width measures.

Figure 3: Equivalent widths of different lines over the years. Abscissa: time in years Ordinate: Equivalent width in A. Please observe the scales, which differ according to the line

**Table 4:** Variations in equivalent width at different epochs
Part a
Lambda	Id	a/e	1991	1993

3933.7	Ca II 1	a	0.08	0.20
4009.3	He I 7¹D	a	0.39	0.26
26.2	He I 5³D	a	0.87	0.81
68.6	[S II] 1	e	0.45	0.54
76.2	[S II] 1	e	0.06	0.06
4130.9	Si II 3	a	0.29	0.07
43.8	He I 6¹D	a	0.47	0.42
73.4	Fe II 27	e	0.06	0.06
78.9	Fe II*	e	0.30	0.31

Part b
Lambda	Id	1989	1990	1990	1992	1993

8374.5	P 21	p	2.03	p	1.52	1.40
92.4	P 20	2.29	2.27	2.31	1.92	1.91
8412.0	P 19	2.93	3.60	3.43	p	2.83
38.0	P 18	p	2.90	3.23	p	2.87
46.5	O I 4	24.22	23.82	24.10	17.87	27.34
67.3	P 17	4.15	4.60	4.50	2.30	4.35
90.4	Fe II J A	0.88	0.96	0.83	0.88	0.99
8502.5	P 16	5.01	5.74	5.47	3.07	5.26
45.4	P 15	5.74	6.44	6.30	3.82	6.00
67.7	N I 8	0.27	0.25	0.30	0.24	0.35
79.2	[V II] 11	0.23	0.16	0.23	0.18	0.27
98.4	P 14	6.45	6.80	6.92	5.78	7.24
8617.0	[Fe II] 13	0.57	0.50	0.53	0.41	0.53
29.2	N I 8	1.95	2.13	2.24	1.89	1.97
49.1	[Sc II] 1	p	0.14	0.21	0.19	0.26
55.9	N I 8	p	0.63	0.75	p	0.76
65.0	P 13	6.85	6.88	6.92	6.33	6.91
80.2	N I 1	4.06	3.53	4.36	3.46	4.33
83.4	N I 1	p	2.89	2.81	2.21	p
95.1	Fe II B	n	0.37	0.43	0.35	0.51
8703.2	N I 1	0.82	1.04	0.82	0.60	1.01
11.7	N I 1	0.86	0.87	0.67	0.73	0.97
18.8	N I 1	0.51	0.67	*	0.64	0.89
28.9	N I 1	p	0.18		0.10	0.30
50.5	P 12	6.63	7.05		5.23	7.67

5. Line variability