Astron. Astrophys. 338, 599-611 (1998)

3. Observational arguments

3.1. The objects

Here we summarize basic properties of the objects under consideration important for understanding the investigated variation in their LCs: (i) Fundamental parameters, (ii) main characteristics of the active and quiescent phases, (iii) the ephemeris for the orbital motion and, finally, we give (iv) a description of our observational results.

BF Cyg (=757.3 days):
Recent study suggested that the system consists of an M bright giant (, , ) probably close to its Roche lobe, and a hot, compact object (, , (in quiescence) K) with an orbital inclination (Skopal et al. 1997).

During the 1989 active phase, an optically thick shell was ejected, the hot star temperature decreased to 18 000 - 12 000 K and the optical continuum was restricted to the hot star (pseudo) photosphere. We observed a deep minimum caused by an eclipse of the active component by its giant companion (cf. Skopal et al. 1997; Cassatella et al. 1992). On the other hand, during quiescent phase, the energy distribution in the spectrum shows a strong nebular component dominating the near-UV/optical and a hot stellar source in the far-UV region (e.g. Fernandez-Castro et al. 1990). A variable attenuation of the far-UV continuum due to Rayleigh scattering of the hot star light by the neutral hydrogen atoms of the cool giant wind was indicated by the IUE observations (Gonzalez-Riestra et al. 1990). During the orbital cycle, 1986-89, the Rayleigh attenuation was seen from the orbital phase 0.9 to 1.6.

A very strong hot continuum masks absorption lines of the red giant in the optical and thus the orbital motions of the components are very poorly known (e.g. Mikolajewska et al. 1989). Therefore, for the purpose of this paper, we use the ephemeris derived from the primary minima (such that 116 days) in the historical LC (Table 1) as

The period of 757.3 days is the same as that already derived by Pucinskas (1970) who supported this value also by periodic variations of other spectrophotometric parameters. We adopt this period as the orbital period of the binary motion.

Table 1. Minima in the light curve of BF Cyg
References: 1 - Jacchia (1941), 2 - from data referred in Skopal et al. (1997), 3 - from the AFOEV data on CDS
* = 2 411 268.6 + 757.3E

Fig. 1 shows the historical LC with the diagram. Positions of the minima are given in Table 1. A systematic variation in the residuals is clearly seen. This behaviour was already noted by Jacchia (1941). An increase before the 1920 bright stage (E=1 to 11) corresponds to a period of 770 days, while the subsequent decrease (E=12 to 24) indicates a shorter period of 747.5 days. Skopal (1992) found that epochs of minima depend on the star's brightness, and generally ascribed this relation to an interaction of the circumstellar matter in the system. The same type of variability appeared again during the recent, 1989 active phase. Observed changes in both the position and the shape of the minima are illustrated in Fig. 2. During the transition from the active phase to quiescence (the transition), a systematic change in the minima positions at E=49 to 51 corresponds to a period of 730 days. During the transition from the quiescent to the active phase (), a significant change in the values by jump of +130 days was observed. The minima at E=45 to E=49 indicate an apparent period of 794 days (Table 1, Fig. 2).

CI Cyg (=855.25 days):
The recent detailed study of CI Cyg made by Kenyon at al. (1991) suggests that this system contains an M5 II giant (, ) and a low mass main-sequence star (, , K), with a high inclination of the orbit of .

Fig. 1. Top: The historical LC of BF Cyg. Bottom: The diagram for the minima listed in Table 1. Full and open squares represent the primary and secondary minima, respectively

Fig. 2. Top: The U, V and visual (smoothed in 30-day bins) LCs of BF Cyg covering its recent active phase. The change in the shape and position of the minima during the transition is well illustrated. Bottom: Corresponding residuals indicate an apparent period of 730 days. The dotted vertical lines give the position of the calculated minimum.

Evolution of the LC from its major eruption in 1975 provides striking constraints on the structure and nature of the source of the optical light. The narrow minima during the first four cycles from the maximum suggest the source of the optical continuum to be formed in a small volume (in the binary dimensions) centered on the hot star. Also the HeII and HI emissions were restricted to the region eclipsed by the giant (Kenyon et al. 1991). This implies that the nebular contribution was negligible at the maximum. Contrary to this behaviour, during subsequent cycles, the broad minima in both the continuum and the HI lines developed. The near-UV/optical spectrum was dominated by the nebular emission and the lines of highly ionized elements (Kenyon et al. 1991). Such evolution resembles that observed for BF Cyg, although it was slower and exhibited a much higher stage of excitation.

Whitney (see Aller 1954) derived the ephemeris for the times of minima as

Belyakina (1979, 1984, 1992) confirmed this period with photoelectric photometry and clearly showed that CI Cyg is an eclipsing system. Positions of the deep eclipse at JD 2 442 692 and the last one, observed at JD 2 450 388, agree perfectly with their predictions by Eq. (2). Therefore we adopt the Whitney's ephemeris for the purpose of this paper, considering that this also represents timing of the inferior conjunction of the giant star in CI Cyg.

Fig. 3 shows the LC of CI Cyg in the U and V/visual bands covering the recent transition period. The residuals are listed in Table 2. Also here, more attempts to improve the photometric elements have been noted in the literature. Mikolajewska & Mikolajewski (1983) and Mikolajewska (1997) marginally modified Whitney's ephemeris, while Kenyon et al. (1991) suggested the orbital period to be 1-2 days shorter than 855.25 days, as the minima at E=37 to 39 (in our notation) occurred 5 days prior to the predictions of Whitney's ephemeris. Our results demonstrate qualitatively the same behaviour throughout the transition as in the case of BF Cyg. Minima observed during the nebular stage (from E=40 to 42) indicate a shorter period of 849 days. However, the last minimum (E=45) appeared to be more rectangular in shape and its position returned to that predicted by the ephemeris (2). Maybe CI Cyg is just preparing for a new activity.

Fig. 3. The U, V and visual LCs of CI Cyg from its main outburst in 1975. Variation in the shape of the minima during the transition (top) is followed by a variation in their positions (bottom)

Table 2. Minima in the light curve of CI Cyg.
References: 1 - Belyakina (1979, 1984, 1992), Hric et al. (1991, 1993), Skopal et al. (1992), 2 - from the AFOEV data on CDS
* = 2 411 902 + 855.25E

V1329 Cyg (=958 days):
The fundamental parameters of this system are not well established at present. The binary consists of an M giant and a white dwarf (, 145 000 K) with a high inclination of the orbit of (cf. Nussbaumer et al. 1986; Mürset et al. 1991; Schild & Schmid 1997).

The symbiotic phenomenon of V1329 Cyg developed in the 1964 outburst. The post-outburst LC shows large, 1.5 mag deep, periodic wave-like variations connected with the binary motion. The IUE observations revealed the presence of a strong nebulosity in the near-UV spectrum. It also varies along the orbit. Around the optical minima the UV continuum is flat and very faint, while at the maxima the near-UV fluxes are a factor of 5 stronger (cf. Fig. 2 and Table 4 of Nussbaumer & Vogel 1991).

Before the 1964 outburst, V1329 Cyg was an inactive star of about 15th magnitude displaying 2 mag deep eclipses. Photometric variability from the pre-outburst period is well documented by Stienon et al. (1974), who suggested an orbital period of 959 days. Later re-analyses of Stienon's et al. (1974) data set argued for shorter periods between about 950 and 954 days (Grygar et al. 1979; Munari et al. 1988; Hric et al. 1993). The most recent careful re-analysis of the pre-outburst data made by Schild & Schmid (1997) revealed the ephemeris

which we adopt for the purpose of this paper. In addition, as the data are not affected by any activity from that time, we also assume that this ephemeris is identical to that for the inferior conjunction of the giant star.

Fig. 4 shows the compiled photographic/visual LC from 1920. The times of minima are listed in Table 3. A sudden change in the star's brightness due to the eruption causes a sudden change in the position of the subsequent minimum (E=31) by about -80 days. All minima occurred prior to the time of spectroscopic conjunction. The residuals systematically increase along the decrease of the star's brightness (see Figs. 4, 5), indicating a larger period than the orbital one. A linear regression of the minima in Table 3 yields their ephemeris as

Fig. 4. The historical LC of V1329 Cyg with the diagram for the minima which developed during the symbiotic activity

Fig. 5. Top: A part of the LC of V1329 Cyg covering its active phase. Bottom: A systematic increase in the values indicates an apparet period of 963.2 days

Table 3. Minima in the light curve of V1329 Cyg.
References: 1 - Stienon et al. (1974), Arkhipova & Mandel (1973), Hric et al. (1993), 2 - Arkhipova (1977), Arkhipova & Ikonnikova (1989), 3 - from the AFOEV data on CDS
* = 2 410 443 + 958.0E

which, according to Eq. (3), sets the initial epoch of the post-outburst minima at the orbital phase . Practically the same result, 963.3() days, was obtained by Hric et al. (1993), who applied a sine wave fit to the photographic data throughout 7 cycles. The two different periods in the pre- and post-outburst LCs, respectively, are also clearly identified by the Stellingwerf's (1978) method of phase minimization (cf. Hric et al. 1993). The shift between the position of the pre-outburst eclipses and the post-outburst minima is directly seen in Fig. 5.

AG Peg (=812.6 days):
This symbiotic star is composed of a normal M3 giant (, , ) and a hot component (, , K) embedded in a dense ionized nebula (Kenyon et al. 1993). The star is classified as a symbiotic nova. In mid-1850's it rose in brightness from 9 to 6 mag and afterwards followed a gradual decline to the present brightness of 8.6 mag in the visual. The spectral energy distribution of the continuum shows a strong nebular component in the optical/near-UV region (Contini 1997). Also the Balmer jump in emission (see Fig. 1 of Kenyon et al. 1993) demonstrates well the recombination process in AG Peg.

We re-analyzed all available radial velocity data covering more than 20 orbital cycles during 1945 - 1992 (cf. Merrill 1951, 1959; Cowley & Stencel 1973; Hutchings et al. 1975; Kenyon et al. 1993). The data set of 82 radial velocity measurements (one point, [41 618.74, -19.2], was omitted as a very strong continuum was superposed on the comparison spectrum; cf. Hutchings et al. 1975) yielded practically the same elements as already derived by Kenyon et al. (1993), but the time of spectroscopic conjunction was shifted by about 60 days (probably a misprint in the Kenyon's et al. paper). Therefore we adopt the ephemeris

as the best timing of the inferior conjunction of the cool giant in AG Peg from our solution for a circular orbit.

From 1940 the LC developed a periodic wave-like variation (Meinunger 1983). The periodic variability has been very intensively studied. The real shifts of the observed minima from the predicted positions were often noted (Belyakina 1985; Luthardt 1984, 1989). As a result, many different periods ranging from 730 to 830 days were suggested (Kenyon et al. 1993 and references therein). Generally, the older data gave a longer period (e.g. Meinunger 1983) than the more recent observations (e.g. Fernie 1985).

Fig. 6 shows the photographic and visual LCs from 1935. Characteristic points (mostly maxima and minima) were taken from Meinunger's (1983) photographic measurements, while the visual data represent smoothed AFOEV estimates available from CDS. Positions of the observed minima are listed in Table 4. Their ephemeris is

Table 4. Minima in the light curve of AG Peg
References: 1 - Meinunger (1983), 2 - determined from the published data (see text), 3 - from the AFOEV data on CDS
*   = 2 427 664.2 + 812.6E

Fig. 6. Top: Compiled photographic/B and visual LCs of AG Peg as recorded from 1935. Bottom: The diagram for the observed minima. A gradual increase in the values indicates an apparent period of 820.3 days

To confirm the real difference of the photometric period from the orbital one given by Eq. (4), we divided the data set of radial velocities into two parts: (i) The old data from 1945.8 to 1973.8 (24 measurements), and (ii) the more recent data from 1978.5 to 1992.0 (58 measurements). Then we solved circular orbits for each data set separately with the fixed period of 820.3 days. Both solutions differ from each other only in the time of spectroscopic conjunction corresponding to an average shift of 0.15 between them (Fig. 7 top). However, the phase diagram of all radial velocities constructed for the elements in Eq. (4) does not display any systematic shift (Fig. 7 bottom). This means that the photometric period is inconsistent with the orbital period, and represents an apparent period in the system. The residuals display a systematic increase along a gradual decrease of the star's brightness. Such behaviour reflects a longer apparent period than the orbital one as in the case of V1329 Cyg.

Fig. 7. Phase diagrams for the radial velocity measurements of AG Peg. Top: The 1945.8 - 1973.8 data (open circles) and the more recent, 1978.8 - 1992.0, measurements (full circles). The data sets are shifted by 0.15 relative to each other when folded with the photometric period of 820.3 days. Bottom: No systematic shift can be seen when the data are folded with the orbital period of 812.6 days (Eq. 4)

Z And (=758.8 days): This system consists of an M3-M4 giant (, , ) and a hot component (, , K) surrounded by an ionized nebula (cf. Mikolajewska & Kenyon 1996; Fernandez-Castro et al. 1995; Nussbaumer & Vogel 1989). The inclination of the orbital plane is 47 (Schmid & Schild 1997b).

The nebular, hydrogen-recombination radiation dominates the near-UV/optical spectrum during both the quiescent and active stages (see Figs. 6 and 8 of Fernandez-Castro et al. 1988 and 1995, respectively). A broadening of all the emission line profiles in the recent, 1984-85, activity indicates an ejection of a shell at moderately high velocities of 200-300 km s^-1. The attenuation of the far-UV continuum due to Rayleigh scattering has only been detected during bright stages (Fernandez-Castro et al. 1995).

Based on all available M-type radial velocities, Mikolajewska & Kenyon (1996) derived the ephemeris for the time of inferior conjunction of the giant as

which we also use in our paper, assuming that the period of 758.8 days represents the orbital one.

A complex wave-like modulation of the LC is indicated in quiescence as well as active stages throughout the whole observational period of Z And. As a result, more different periods have been suggested. For example, Payne-Gaposchkin (1946) obtained a mean period of 694 days, Belyakina (1985) suggested a period of 700 days and Kenyon & Webbink (1984) derived a 756.85-day period. Recently, Formiggini & Leibowitz (1994) analyzed all available photometric observations in the last 98 years. They found a sample of periods between 610 and 865 days, among which that of 759 days was suggested to be the orbital period and that of 656 days was present during high stages of activity. Fig. 8 shows the compiled photographic/visual LC as recorded from the beginning of this century. Timing of the observed minima are listed in Table 5. The values depend on the activity of the system. An abrupt change in the star's brightness, as in 1915, 1940 and 1984 ( transitions), causes a significant change by a jump in the residuals to positive values, i.e. a larger period is indicated, while the transition is followed by a gradual decrease in the quantities indicating a shoter period than the orbital one. Fig. 9 illustrates such behaviour for the recent active phases. Apparent periods of 732.7 and 737.8 days correspond to the minima in the transitions, while the separation between the minima just preceding (E=41) and following (E=43) the activity in 1984-85 corresponds to a period of 818 days. However, all the primary minima from Table 3 (such that 182 days) yield the ephemeris as

in which the average period is identical (within its uncertainty) with the orbital one in Eq. (6).
AG Dra (=550.8 days): This symbiotic system is composed of a K-type (bright) giant () and a hot compact star, which could be a white dwarf ( K) embedded in a dense nebula (Mikolajewska et al. 1995). There are no signs neither in the optical nor far-UV regions of eclipses. Recently Schmid & Schild (1997a), based on spectropolarimetric observations, derived the orbital inclination .

Fig. 8. Top: The historical LC of Z And. It is compiled from photographic data (Payne-Gaposchkin 1946), visual AAVSO estimates (Mattei 1978) and smoothed visual AFOEV data. Bottom: The diagram for the minima in Table 5

Fig. 9. Top: The U and visual LCs of Z And during the recent active phases. Bottom: The residuals indicate apparent periods of 732.7 and 737.8 days

Table 5. Minima in the light curve of Z And
References: 1 - Payne-Gaposchkin (1946), 2 - Mattei (1978), 3 - Romano (1960), 4 - Mjalkovskij (1977), 5 - Belyakina (1985), 6 - from the AFOEV data on CDS. 
* = 2 445 703 + 758.8E

The system undergoes occasional eruptions, lasting about 1-2 orbital cycles. The star's brightness abruptly increases (1 mag) showing multiple maxima separated approximately by 400 days (e.g. Luthardt 1983; Skopal 1995, 1998; Bastian 1998). During eruptions the hot component develops a fast wind at a few km s^-1 (e.g. Viotti et al. 1994a, 1994b). The spectral energy distribution of the continuum shows a strong nebular component dominating the UV/optical region mainly during the activity (Fig. 5 of Greiner et al. 1997). The quiescent phase of AG Dra is characterized by a periodic wave-like variation in the optical continuum, which is more pronounced at short wavelengths (e.g. Skopal 1994).

Also in this case we re-analyzed radial velocity data as published by Mikolajewska et al. (1995), Smith et al. (1996) and Tomov & Tomova (1997). The dataset of 73 radial velocity measurements (one point, [49 376.6, -138.6], from the last dataset was omitted as it is deviated from other data by more than the semiamplitude of the whole variation) covering nearly 9 orbital cycles gave the elements very close (within their uncertainties) to those previously derived ( days, ). As a result we adopted the ephemeris

as the best timing of the inferior conjunction of the cool component in AG Dra. Fig. 10 shows the /U LC from 1940 when AG Dra began to be active (cf. Friedjung 1988). The photographic LC was constructed from the data published by Luthardt (1983) in the same way as for AG Peg. The photoelectric measurements in the U/u bands represent those summarized by Skopal (1994). Positions of the observed minima are listed in Table 6. They obey the ephemeris

which is very close to that derived from radial velocities. Thus the orbital period can be considered to be 550.8 days. The residuals display rather short-term fluctuations, lasting a few periods. We can recognize apparent periods of 515 (E=5-7), 536 (E=21-24) and 538 days (E=30-32).
YY Her (=592.8 days): There is no detailed study on this object. Previous investigations showed that YY Her consists of a normal M3 giant (Kenyon & Fernandez-Castro 1987) and a hot compact star, K (Mürset et al. 1991). Spectroscopic observations revealed a strong emission line spectrum of HI, HeI, HeII and [OIII] (Herbig 1950). Munari (1997) showed that during the 1993 active phase the total integrated flux of all emissions dramatically increased. At that time the hot continuum dominated the optical spectrum (up to 7 500 Å) displaying a prominent Balmer jump in emission. Munari et al. (1997) summarized the historical LC from 1890 to 1996, showing that YY Her underwent several outbursts and/or bright phases. During quiescence, the continuum exhibits a complex wave-like modulation with a periodicity of 590 days (Munari et al. 1997).

Fig. 10. Compiled photographic/U LC of AG Dra from 1940 (top). A fast transition produces a rapid change in the residuals (bottom)

Table 6. Minima in the light curve of AG Dra
References: 1 - Sarov (1960), 2 - Luthardt (1983), 3 - Skopal (1994).
*   = 2 430 974 + 550.8E

Fig. 11 shows a part of the LC covering the period from its bright stage in 1971-74. Positions of the observed minima are compiled in Table 7. They determine the ephemeris as

It is identical (within uncertainty) in the period with that of Munari et al. (1997), but differs from it in the initial epoch by 48 days. Thus we adopted elements in Eq. (8) for the purpose of this paper. The residuals display a large scatter, which, however, represents real variation in the minima position (see Fig. 11). For example, their systematic decrease from the 1971-74 brightening to the 1981-82 outburst, indicate an apparent period of 578 days. Also Munari et al. (1997) found the period to be 20 days shorter than 590 days prior to the 1981-82 active phase.

Fig. 11. Visual LC of YY Her from the 1971 bright phase. The AAVSO data set was taken from Munari et al. (1997), and the AFOEV data represent means in 30-day bins with a step width of 15 days. The residuals reflect a larger scatter as can be seen directly from changes in the positions of minima

Table 7. Minima in the light curve of YY Her
References: 1 - from data published by Munari et al. (1997) 2 - from the AFOEV data on CDS
*   = 2 440 637 + 592.8E

3.2. Common properties

There are two main characteristics in the variation of the residuals:

1. Systematic variation in positions of the minima is connected with a variation in the energy distribution of the hot component radiation.

The effect is very striking when also the nature of the hot continuum changes. In the case of eclipsing systems, the broad minima (i.e. those appearing during the nebular phase when the recombination radiation dominates the optical spectrum) occur prior to the time of spectroscopic conjunction. Contrary, when the ionized region is disrupted by eruption, the position of the minimum returns to the conjunction time. For example: (i) In BF Cyg the time of inferior conjunction of the giant star is indicated by a deep narrow minimum (the eclipse) at the epoch E=49. Both the subsequent and the preceding minima occurred before the time of spectroscopic conjunction (cf. Fig. 1, Table 1). A significant change in the nature of the optical continuum during the transition (E=49 to 51) - from a black-body to a nebular radiation - is well documented by the spectroscopic observations (Sect. 3.1; Fernandez-Castro et al. 1990; Skopal et al. 1997). (ii) The case of CI Cyg is qualitatively the same as that of BF Cyg. During the nebular phase of the transition (E=40-42) the broad minima also appeared prior to the spectroscopic conjunction of the giant (Fig. 3, Table 2). The variation in the hot component radiation in the optical during this transition is of the same nature as in BF Cyg (see Sect. 3.1). The return of the recent minima (E=44, 45) to the position of the conjunction predicted by Eq. (2) is again followed by a narrowing of their profile (Fig. 3). (iii) The nebular phase of V1329 Cyg has developed immediately after the 1964 eruption. The first well observed minimum occurred 80 days prior to the time of conjunction predicted by Eq. (3). All the subsequent minima also preceded the conjunction time, gradually approaching to it along a decline in the star's brightness, and thus causing a larger apparent period then the orbital one. (iv) Similar behaviour is observed in AG Peg, although we do not have a direct sign of eclipses in this system. At the beginning of its nebular phase, when the wave-like variation developed in the LC, from 1940, the minima were shifted by about -100 to -200 days from their prediction by Eq. (5), and were appearing systematically closer to the time of conjunction resulting in a larger apparent period. Also here such behaviour in the residuals is followed by a gradual decline of the star's brightness.
(v) In the case of non-eclipsing systems (Z And, AG Dra) the minima occur on both sides of the position of spectroscopic conjunction. The residuals generally move from positive to negative values during transitions.

2. Separation between the minima correlates with the velocity of the brightness variation.

Generally, a fast and large change in the period is indicated during transitions, when the brightness changes rapidly. Contrary, transitions, during which the star's brightness declines slowly, produce a smaller change in the period. To demonstrate this better, we determined a relation between parameters and characterizing the velocity of the change in the star's brightness and the corresponding phase shift in the period, respectively, for different transition epochs. The result is compiled in Table 8 and plotted in Fig. 12. The empirical relation

fits the data quite well. The parameter is in (see Table 8).

Table 8. Parameters of selected transition epochs. Quantities in the last two columns are factor of 10² and 10⁴, respectively, enlarged
Notes:
a - corresponds to the nebular phase (E=50-51)

Fig. 12. Correlation between the velocity of the change in the star's brightness () and the corresponding phase shift in the period () for different transition epochs. The solid line represents a best linear fit given by Eq. (9)

© European Southern Observatory (ESO) 1998

Online publication: September 14, 1998
helpdesk@link.springer.de