1 Introduction

X-ray observations of the Sun have shown that almost all the coronal emission comes from plasma confined in closed magnetic structures in the form of arcs (loops). In this sense, instead of saying that loops are in the corona we should rather say that they are the corona (Vaiana & Rosner 1978). The theory of the dynamo developed by Parker (1979) explains how the differential rotation generates a toroidal field in the interior of the star from an initial dipole field and how the convection bringing this field up to the surface gives rise to the emergence of the loops. In its emergence on the surface, one loop may intrude into other already-established loops; this occurrence of two fields of opposite directions pushed one against the other is one of the assumed mechanisms of large flares. The release of energy produces relativistic particles emitting bremsstrahlung in hard X-rays and gyro-synchrotron radiation at radio wavelengths (Parkes 1979; Bastian et al. 1998). Images of the Sun with the Yohkoh telescope and the Nobeyama Radio Interferometer (Nishio et al. 1996) have actually confirmed this scenario of colliding loops.

The RS Canum Venaticorum (RS CVn) stars are close binary systems with flares in radio and X, orders of magnitudes stronger than solar flares. The higher degree of magnetic activity is due to an increased efficiency of the dynamo because of the deeper convective zone of those stars and because of their higher rotational velocity, compared to the Sun. In fact, the more active star of the system is generally a sub-giant of spectral type G or K rotating in less than one month (Owen et al. 1976; Dulk 1985; Elias et al. 1995; Beasley & Güdel 2000).

Figure 1: Observations of V773 Tau with the Effelsberg 100-m telescope and the VLA. The VLA observations are those at 2.0 cm centered at Julian Day 11730. The system V773 Tau is composed of two stars orbiting with a period of 51.075 days; the bars in the figure show the periastron passages assuming as initial epoch t0= 2449330.94 JD (Welty 1995).

One of the most active RS CVn stars, UX Arietis, produces strong flares following two clearly alternating ``regimes'': an active regime (with flares above 250 mJy) and a quiescent one alternating with a period of 158.7 days (Massi et al. 1998). During the active phase the activity does not take place randomly, but shows a period trend of 25.5 days. Moreover the sign of the circular polarization seems to reverse within the cycle of 25 days, returning to its initial value after $56 \pm 4$ days (Massi et al. 1998). What is the origin of this periodicity? UX Arietis is a binary system with an orbital period of 6.44 days (Carlos & Popper 1971; Elias et al. 1995). Before speculating on possible mechanisms responsible for the periodic emergence of the loops, one must first establish whether the flares could be the result of collisions of loops anchored on the two stars of the system. VLBI observations of UX Arietis have been interpreted by a model of two intruding loops, emerging from the same star and pushed one against the other (Franciosini et al. 1999).

Figure 2: Radio observations of V773 Tau folded with the orbital period of 51.075 days. Phase 0 (1, 2) refers to the passage at periastron.

However, in close binary systems the magneto-spheres of the two stars may interact, giving rise to a joint magnetosphere (Uchida & Sakurai 1983). In order to study the mechanism of the flares and their relationship with the dynamo we selected a new source (avoiding the complication that may be implied by a modified topology of the magnetic field because of the small distance of the stars). V773 Tau (HD 283447) also is a binary system, but with the two stars separated by tens of stellar radii (Welty 1995). V773 Tau belongs to the class of T Tauri stars (Neuhäuser 1997; Guenther et al. 2000). The fact that they are fully convective objects, together with their fast rotation, makes these objects very similar to the RSCVn-type of stars from the point of view of their magnetic activity with radio flares of three to six orders of magnitude brighter than solar flares (Feigelson & Montmerle 1999).

Figure 3: Fourier power spectrum of the time series shown in Fig. 1. The dominant peak is at 52 $\pm$ 5 days. The second peak (corresponding to 27.4 $\pm$ 1.4 and 25.9 $\pm$ 1.3 days) could be one harmonic of the dominant peak. The third peak at 17.9 $\pm$ 0.6 days is a mode interaction feature: ${1 \over {1 \over 52}+{1 \over 27.4}}=17.9.$ and disappears when the Phase Dispersion Minimization method (Stellingwerf 1978) is used instead of the Fourier analysis.