This section studies the M and A of each SET with a linear model (Eq. 1), where and are from Table 3.
We found no earlier photometric flare detections for V 1794 Cyg, nor reported any in Jetsu et al. (1990a, 1990b). All UBVRI light curves were analysed with the flare detection procedure applied earlier for FK Com (Jetsu et al. 1993). It relies on the light curve residuals, and on the regular linear dependence between nonflare measurements in two arbitrary UBVRI magnitudes (Jetsu et al. 1993: Figs. 1, 2 and 3). Flares are above the light curve and violate the aforementioned linearity, especially in UB. Flares were identified in 19 subsets, including 4 in Jetsu et al. (1990a, 1990b). Our "Flare" and "Flare?" notations are as in Jetsu et al. (1993): U-band data were available only for the former ones, and thus the latter identifications are uncertain, the flares deviating less in BVRI (compare Figs. 1 and 2). These deviations were excluded before the light curve modelling and the determination of , regardless of whether they represent real flares or observational errors. If flares, these are most probably of short duration, because two successive measurements possibly representing a single event were recorded only during SET=40. But except for a few subsets, V 1794 Cyg was observed only once or twice each night, which prevents an indisputable flare identification, such as e.g. in Jetsu et al. (1993: Fig. 1). Apart from flares or observational errors, rapid light curve changes might induce some of these deviations (e.g. Fig. 1: SET=112), because an adequate light curve phase coverage requires a subset length of about . Unlike FK Com (Jetsu et al. 1993: Fig. 6b), V 1794 Cyg seems to have no preferred flaring phase with respect to the light maximum (Figs. 1 and 2). Henry & Newsom (1996) detected only five photometric flares in 17207 measurements of 69 chromospherically active evolved stars. Those flare detections for UX Ari, II Peg, and AR Psc coincided with light curve maxima, as also for FK Com (Jetsu et al. 1993). Our flare analysis for V 1794 Cyg resembles that by Henry & Newsom (1996), except that the F=Flare? events were identified without U-band data, and are therefore less reliable.
Fig. 3 displays the , , , , and changes for subsets with . Table 1 gives all M and A in UBVR. Only four subsets had in I. These SET=6, 10, 22 and 32 had , , and combined with = , , and , respectively. For example, has varied between 7.10 (SET=23) and 7.26 (SET=10), and from 0.01 (SET=14) to 0.24 (SET=45). There has been a half a year interval of nearly constant brightness (Fig. 1: SET=67-79), as well as changes from high to nearly constant brightness during a few months (e.g. Fig. 1: SET=112-114). Contrary to , the changes are smoother, but with some scatter superimposed. This may be partly due to photometric transformation inaccuracies for the data from different observatories. Nevertheless, this short-term scatter is mostly intrinsic to the object, because consecutive light curves are continuous (Fig. 1). The surface temperature of V 1794 Cyg follows these long-term mean brightness changes, since decreases when the brightness increases (Fig. 3c).
A second order weighted TSPA was performed for the , , and changes (Paper i: in Eq. 1). The best periods were and . Correlation between and (Figs. 3ab) induced the same periodicities for the former. The first ( double sinusoid) represents the whole data time span without the "lonely" at 1975. The second ( sinusoid) is about . Both are not significant, because the 93 values of have (Eq. 10 in Paper i: ). Our in Table 1 are comparable to photometric internal accuracies . These periodicities would not be significant even if the external accuracy () were assumed to reduce the by , i.e. from to . The short-term changes are therefore real, because represents the whole observing interval without the value at 1975. A significant second order model would require a much more conservative than our . The and changes are neither periodic. The best for has for . And here does not influence , because the are nearly independent of external accuracy. Thus we can not reject the "null hypothesis" that the variations are pure noise (Paper i: in Sect. 3.3). In conclusion, V 1794 Cyg has no regular M or A cycles.
The earlier M and A periodicities were detected with the power spectrum method from data between 1975 and 1989 (Jetsu et al. 1990a, 1990b). Those studies relied on a constant P ephemeris with a sinusoidal model. Without the first two subsets in Jetsu et al. (1990a, 1990b), the others were between 1980 and 1989, which explains this earlier cycle detection for M. The significance of the other cycle in A was not estimated in Jetsu et al. (1990a, 1990b), but here for must certainly be rejected.
© European Southern Observatory (ESO) 1999
Online publication: November 2, 1999