We have determined the orbit of the pre-main sequence double-lined spectroscopic binary RX J1603.8-3938. Although RX J1603.8-3938 is a young binary system, the orbit is circular, despite the fact that the period is 7.5d. This is quite surprising, because pre-main sequence binaries with periods longer than 5d usually have eccentric orbits (Mathieu 1992). Only for main-sequence stars, this boundary shifts to about 11 days (Duquennoy & Mayor 1991). With values of 0.41 and 0.43 Å for the equivalent width of the LiI 6708 line, the strength of this feature in both components is substantially larger than is typical for stars of the same spectral type in the Pleiades (Soderblom et al. 1993). This and other evidence clearly establishes the system as a pre-main sequence object. The results for RX J1603.8-3938 imply that circularization certainly operates during the pre-main sequence phase of evolution. To our knowledge RX J1603.8-3938 is the binary with the longest period that has a circular orbit which is known for sure to be pre-main sequence. During the refereeing process we learned that an independent orbit for this system was derived by other authors (Melo et al. 2000), giving very similar elements.

From the orbit, we also conclude that the masses of the two companions are almost identical. In contrast to this, we find that the equivalent width of the photospheric lines are a factor of 1.7 smaller in the secondary than in the primary. As already outlined above, the small equivalent width of the secondary cannot be explained by the presence of a veiling continuum, as the two components are weak-line TTauri stars. For a binary with orbital period of 7.5d, extinction in front of one component is also not an option. Another possibility would be the presence of very large spots. For example, the long-term changes in the apparent magnitude of the young star P1724 of 0.2 mag are presumably caused by large spots. The spots also produce a short-term photometric modulation with an amplitude of 0.4 mag, and their shape and location were reconstructed by Doppler tomography (Neuhäuser et al. 1998). If spots play an important role in the case of RX J1603.8-3938, we would also expect periodic variations in its photometry and radial velocity. For example, Saar & Donahue (1997) find for late-type dwarfs a RV variation (semi-amplitude) of $A_{\rm s}= 0.0065\,f_{\rm s}^{0.9}\,v\,\sin\,i$ km s^-1, where $f_{\rm s}$ is the filling factor in percent. If the same holds for young stars, and most of the 0.5 $\rm km\,s^{-1}$ scatter in our RV curves were caused by spots, we would estimate the filling factor to be of the order of 5 to 10%. As mentioned before, the brightness in the V-band given in the TYCHO-2 catalogue of $11.01\pm0.17$ is the same, within the errors, as the value 11.02 given by Wichmann et al. (1997), making large amplitude variations unlikely. While this does not rule out the presence of spots completely, it seems somewhat unlikely that the difference in the equivalent width between the A and the B component is caused by spots. Since the spectral types are essentially identical, we are led to the conclusion that the secondary is intrinsically fainter than the primary by $0.55\pm0.05$ mag. Using the photometry by Wichmann et al. (1997) and the TYCHO-2 catalogue we can place the two stars into the HR diagram. From our measurements we thus conclude that the A and B components have bolometric luminosities of $5.32\pm0.05$ , and $5.87\pm0.05$ mag assuming a distance of 140 pc, or $6.05\pm0.05$ and $6.60\pm0.05$ assuming a distance of 100 pc.

Figure 4 shows the position of the two components in the HR digram, together with the evolutionary tracks by D'Antona & Mazzitelli (1994). From the position of the objects in the HR diagram, one would conclude that the two stars have different ages, and different masses. For example, if we assume a distance of 140 pc, we would estimate the mass of the primary to be slightly larger than $1.0~M_{\odot}$ , and that of the secondary as $\sim$ 0.85 $M_\odot$ . However, the true mass ratio is $0.9266\pm0.0063$ . Even more worrying is that the ages of the two components come out different. From the Fig. 4 we would estimate the age of the primary as 1 to $2\,10^7$ yrs, and the age of the secondary as $\sim$ $3\,10^7$ yrs. Such a large difference in age is also in contradiction to the fact that the equivalent widths of the LiI 6708 lines of the two stars are the same. Even if the determination of the spectral-types were grossly wrong, we are unable to shift the secondary far enough to be in agreement with the true mass-ratio. In order to demonstrate this, we also marked the position of the B component if the secondary were a K5 star (B3), and if the secondary were a K2 star (B1). However, as pointed out before, the B1-case is highly hypothetical. If the secondary were a K5 star, the mass derived from the evolutionary tracks would be $\sim$ 0.75 $M_\odot$ . Thus, the conflict with the true mass-ratio would be even larger.

$\begin{figure} \par\includegraphics[angle=180,width=12.7cm,clip]{dantona_mazzitelli.ps}\end{figure}$

Figure 4: Shown is the position of the two companions of RX J1603.8-3938 in the HR diagram. The filled symbols are for distance of 140 pc. Marked with A is the position of primary component. With B1, B2, B3 we denote the positions of the secondary component, where B1 is the position if the spectral type of the secondary is K2. B2 is for a spectral-type of K3, and B3 for a spectral type K5. Since the total luminosity of the system has to be kept constant, the point A would also move. For example, if the secondary gets fainter, the primary would get brighter. However, we decided not to show the A1, A2, A3 because all these points would be close to A, and the figure would become confusing. The open symbols are for a distance of the system of 100 pc. In this case, we assume that both companions have a spectral type of K3. Also shown are the evolutionary tracks calculated by D'Antona & Mazzitelli (1994). According to these tracks the two components would have different ages, and different masses, in contradiction to the mass ratio derived from the orbit. As can clearly be seen, even if the determination of the spectral-types were grossly wrong, we are unable to shift the secondary far enough to be in agreement with the true mass-ratio

We can also turn round this argument: let us assume that the primary has a mass of one solar mass and a spectral type of K3. Then the secondary would have a mass of $0.93\,M_{\odot}$ according to the mass-ratio derived from the orbit. In this case we would derive from the evolutionary tracks that its effective temperature would have to be 4470 K, corresponding to a spectral type of about K4, and the secondary would be about 0.2 mag fainter than the primary. However, if the secondary had this effective temperature, the brightness difference derived from the measurements of the equivalent width would be almost one magnitude. So, there is again the same discrepancy between the evolutionary tracks and the measured large difference in equivalent width of the lines.

Figure 5 shows the position of the two components of RX J1603.8-3938 in the colour-magnitude diagram (using the brightness and $V-I_{\rm c}$ -colour published by Wichmann et al. 1997). Also shown is the average position of the stars in the clusters IC2391 and IC2602 taken from Stauffer et al. (1997). These two clusters have an estimated age of 25 Myr. For a distance of 100pc, the primary falls slightly above and the secondary slightly below the average position of stars in IC2391 and IC2602 in the colour-magnitude diagram. For a distance of 140 pc, both components are above the average position of stars in the two clusters.

$\begin{figure} \par\includegraphics[width=7.2cm,clip]{colour-magnitude}\end{figure}$

Figure 5: The position of RX J1603.8-3938 in the colour-magnitude diagram of IC 2391 and IC 2602. The two clusters have an age of 25 Myr. For a distance of 100pc, RX J1603.8-3938 fits in to the colour-magnitude diagram of these two clusters. For 140 pc, both stars are above the curve

In fact, due to the limited photometric accuracy, intrinsic variability, slight difference in distance, and possible differences in age and chemical composition between the two clusters, the scatter of the members of these two clusters is comparable to the distance of the two components of RX J1603.8-3938. From an observational point of view, the positions of the two components of RX J1603.8-3938 are by no means special. The problem in placing RX J1603.8-3938 onto the HR diagram is not so much the specific evolutionary tracks used but the fact that two pre-main sequence stars of almost identical mass have such a differences in brightness. If a cluster of stars is observed, this difference in brightness would not be recognised because of the large scatter in the colour-magnitude diagram. One possible explanation for the problem in comparing RX J1603.8-3938 with the tracks is that the evolutionary tracks that are computed for isolated stars, are not valid for very close binaries that may have interacted during the formation. Another problem of course is that the observed colour, and magnitudes have to be converted in to $T_{\rm eff}$ and $L/L_{\odot}$ before they can be compared with the evolutionary tracks.

5 Discussion