3. Membership in the moving group
3.1. The kinematic criterion
Based on the positions, parallaxes, and proper motions provided by the Hipparcos database, and the radial velocities (SIMBAD database and references therein), we have computed the kinematic properties of our sample of stars. Table 3 shows the numbers in the Gliese Catalog, the positions, parallaxes, proper motions, radial velocities, and the Galactic velocities (U, V and W), with their associated errors for all these quantities. Galactic velocities and errors were computed following the description provided by Johnson & Soderblom (1987) and follow the right-handed coordinate system (positive toward the Galactic center, Galactic rotation and North Galactic Pole). Since the time span for the Hipparcos mission was short, the proper motions based on Hipparcos data might be not accurate. For this reason, we have also computed the Galactic velocities using the proper motion values from the PPM (Roeser & Bastian 1988; Bastian & Roeser 1993; Roeser & Bastian 1994), and the Hipparcos positions and parallaxes. Except in very few cases (e.g. GL351, GL521.2, Castor itself), the discrepancies do not affect the conclusions. However, we note that the differences between the Hipparcos and the PPM proper motions differ by more than the quoted uncertainties for all stars in the sample. Narayanan & Gould (1998) have shown that Hipparcos proper motions are consistent with its parallaxes (at least in the direction toward the Hyades), an indication that there are not systematics errors due to the short baseline. The derived V component of the Galactic velocities also differ by more than the nominal uncertainties in 30% of the sample. This fact indicates that some stars catalogued as members of the moving group could be spurious members.
Table 3. Coordinates and velocities for the Castor moving group.
We have made a special effort to use the most accurate radial velocities. Primarily, we selected the values from the WEB catalog (Duflot et al. 1995), a compilation of three different works (Wilson 1963; Evans 1978; Batten et al. 1989). We also provide other values quoted in the literature. Only in three cases (GL426, GL564.1 and GL842.2), there are disagreements between these values. For GL426 and GL842.2, the kinematics and other information seem to indicate that they are not associated with the moving group. The case of GL564.1 is discussed below.
Figs. 2a and 2b allow us to identify possible members based on their kinematic properties (V against U and W, respectively, from Hipparcos data). Soderblom & Mayor (1993) pointed out that the components of a stellar kinematic group must have the same motion in the direction of the Galactic rotation (the V component). Visual inspection indicates that the velocities are clustered around (-10,-8,-10). The dispersions of these velocities are rather small, in particular for the the 7 A spectral type stars (note that Castor is itself composed of two pairs of spectroscopic binaries, but we have counted them as one star). However, there are several stars which have velocities in disagreement with these average values.
In particular, the Galactic velocities (specifically, using the V component as a criterion) seem to indicate that GL426AB, GL466 and GL696 are probable non-members of the moving group, and that GL803 and GL842.2 are possible non-members (indicated with the labels "N" and "N?" in the second column of Table 4). The table also indicates if the stars are probable or possible members ("Y" and "Y?", respectively).
Table 4. Membership in the Castor moving group.
GL564.1 (2 Lib) deserves special attention. It is physically associated to GL563.4 (Poveda et al. 1994). The WEB value of the radial velocities of both stars leads to a a similar value of the V component of the Galactic velocity, but Poveda et al. (1994) provide a radial velocity in large disagreement with the WEB value. We have preferred to keep this last value. On the other hand, the U component of the velocity of GL563.4 is quite different from the average of the group. The only relevant difference between both stars is the radial velocity and this could be due to the orbital motion as well as to uncertainties in the measured values. Therefore, we do not rule out the possibility that it indeed belongs to the moving group.
Fig. 3c displays the velocities computed with the PPM data. The agreement in the kinematics of these stars also suggest a common origin. However, when computed from the PPM values, Castor has a V component of the velocity quite different from the average of the moving group, with a difference much larger than the quoted uncertainties. Therefore, it would be possible that these system would not be a member of the moving group (and the association should be called Vega moving group). In fact, if a very restricted definition would be used, this figure indicates that there could be 2 different groups; one including Vega, Fomalhaut and Phe and another with 14 Lep and Lib. Alderamin could be in either of them, due to its large error bars. However, the data based on Hipparcos and the evidences provided below seem to indicate that, indeed, all of them share a common origin.
3.2. Color-magnitude diagrams
Further information on the membership can be obtained from Color-Magnitude Diagrams. Fig. 3a displays the absolute visual magnitude against the (B-V) color, whereas Fig. 3b has (V-I)c in the x-axis. The initial sample of late spectral type stars are shown as circles. Filled symbols indicate final members and open symbols those stars rejected as physically associated (see Sect. 3.4). Both figures include the D'Antona and Mazzitelli (1994) isochrones for the ages 3, 10, 35 and 70 Myr (top to bottom, dashed lines). The solid line is the ZAMS. Details concerning the convertion between theoretical values ( and luminosity) and the observational plane can be found in Stauffer et al. (1995) and Barrado y Navascués et al. (1997a).
We have selected those stars located below the 35 Myr isochrone and above the ZAMS as possible members and indicated this in the third and fourth columns of Table 4.
3.3. Stellar activity
Fig. 4 compares the equivalent width of H for our sample of stars (filled circles) with the Hyades (open circles). This graph can be used to support rejection of several candidates or to accept them as members: GL696 presents a minimum value in its H absorption (it is a very inactive star), so it seems to be older than the Hyades and a non-member of the moving group. GL896A, GL896B and GL803 are very active stars. Their activity is similar to the most active Hyades stars, an indication that they could be younger. However, Barrado y Navascués et al. 1998) have shown that some dM Praesepe stars have larger activity than equivalent Hyades stars (both clusters are coeval). Therefore, these stars could be as old as 600-800 Myr. The spectroscopic binary YY Gem is a BY Dra system, whose rotational period is synchronized with its orbital one, phenomenon which produces an enhanced stellar activity, not related with the age of the system. Its location on the plot cannot provide any information about its membership to the moving group. Similar conclusions can be reached if the coronal activities are compared.
3.4. A list of members
We have combined the information provided by the kinematics and the CMD to select a list of members. We have also taken into account the lithium abundance and the H equivalent width when that information is available. We find that there is a good agreement between these criteria (if the kinematics are different from the average value, then the location in the CMD usually desagree with membership).
Column six in Table 4 provides our final classification for the membership. The hotter stars are classified as members based only on the kinematic. However, they are relatively young, and the probability of having several stars of A spectral type, leaving the Main Sequence, with the same spatial velocity, and not having been born at the same time, is quite low. We have a total of 16 probable members and another star which could be part of the group (1 Lib).
3.5. Are these stars physically associated?
A significant fraction of the stars studied here fulfill the Soderblom & Mayor (1993) definition of a stellar kinematic group. The most restrictive requirement is that of having identical V component of the Galactic velocity. In our case, there are 8 stars with V within 1 sigma of the average of the final members. This group includes Castor, Vega and Fomalhaut. Although we do not rule out the possibility of having a spurious kinematic group, we believe that the data presented here provide strong indications for a common origin of, at least, the majority of the sample. However, the membership of individual stars cannot be known without doubt. In fact, Whitmire et al. (1992) studied the kinematics of several A stars in the solar vicinity, including several stars in common with our study, namely, GL20, GL217.1, GL278, GL564.1, GL721, GL826 and GL881. Although they discussed the possibility of a common origin for several of these stars, they concluded that the small velocity dispersion is due to an observational selection. Their model implies an interaction with an interstellar cloud, a phenomenon which could enhance the circumstellar disks. However, not all A stars in our sample display IR excesses, a subproduct of the disks. In any case, some caveats about particular membership should be kept in mind.
© European Southern Observatory (ESO) 1998
Online publication: October 22, 1998