2 Selection of new candidates and other targets

2.1 Origin

The main result of Paper I was a set of criteria to be met by a binary system, so that one can expect to determine the mass of its components from the Hipparcos astrometric measurements. These criteria are still applicable in the present study (an orbital period $P\leq 30$ years and the semi-major axis a''> a few tenths of arcsecond), even if the application to real data has shown that our expectations for pairs with periods larger than 20 years were slightly too optimistic. Two different sources, again largely redundant, were used to select the new candidate systems:

1.

The most recent version of the catalog of orbits of visual binary stars compiled by Worley in mid 1997 (not yet published), containing 1403 orbits for 1038 systems.

2.

The compilation file of orbits of Scardia (June 1997), also unpublished, containing 1744 orbits for 709 systems.

After elimination of all the objects already studied in our previous paper, and elimination of the systems presenting one or more additional disturbing companions, we are left with 37 systems matching the above criteria in the first source and 5 additional systems in the second one. These 42 new candidates are listed in Table 1.

In addition, we have checked the orbital elements of the 145 systems already processed in Paper II, allowing us in some cases to estimate new masses based on revised orbits, or even get masses for systems where the previous treatment failed.

These 28 systems are listed in Table 2, followed by six systems whose masses were successfully computed in Paper II. For these six binaries flagged "C" in the Hipparcos Catalog (component solution), the current processing is based on the Hipparcos magnitude differences, when they are of comparable or better quality than the ground-based $\Delta m$ that we used before (but the orbital elements remain unchanged). Eventually, the application of our selection rules yields a total of 99 orbits for 76 systems (42+34).

  

Table 1: HIP numbers of the 42 new preselected systems

  

Table 2: HIP numbers of the 34 reprocessed systems. For the last six systems, masses were already derived in the previous paper

2.2 Elimination of objects

The same rules as in Paper II apply here; no significant results can be obtained for multiple systems if a third (or more) bright components is within 25'' of the pair under consideration. This restriction is directly linked to the Hipparcos observing field. In some cases, when the magnitude of the parasitic source was close to or beyond the sensibility threshold, a solution could however be derived. Among the new candidates, 5 such systems were removed in the present study (18 in Paper II).

2.3 Description of the systems

The description of the sample shown in Fig. 1 aims to distinguish between two categories of systems:

1.

the so called "Type I" stars for which the mass fraction B=M₂/(M₁+M₂) can be derived solely from the Hipparcos data (pairs with semi-major axis typically larger than 0$.\!\!^{\prime\prime}$3),

2.

the "Type II" stars, with a possible determination of the scale factor between the relative and photocentric orbits, namely the difference $\beta-B$ between the intensity and mass fractions.

Although there are five likely candidates of the first kind in this study (exception made of the last 6 systems in Table 2), HIP 7918 was the only object for which a separate computation of the mass fraction was possible. This is primarily due to the limitations caused by the large period characterizing this type of object.

  

Figure 1: Distribution of the 70 candidate systems (42 new systems + 34 reprocessed, minus the last 6 systems of Table 2) in period-semi-major axis space. The filled symbols correspond to the stars whose processing has been successful. Only one orbit per system was considered

  

Table 3: The 13 "new" astrometric binaries whose processing yields satisfactory results. The columns give the Hipparcos, ADS and HD identifiers when available, the usual name, the seven orbital elements and their reference

¹ When available, the more common name of the star, or the discoverer designation.
² The three first characters of the author's name followed by the two last digits of the publication's year.
³ Ground-based measurements yield the size of the photocentric orbit (0$.\!\!^{\prime\prime}$108). The value 0$.\!\!^{\prime\prime}$33, which refers to the relative orbit, is just an assumption (see Kamper 1987).

More interesting are the systems with periods smaller than 10 years, making 40% of the stars selected so far. Two thirds of the accepted results fall in this group (see Table 6). There are two good reasons accounting for this fact: first, the odds that the orbital motion shows up in the Hipparcos data are more favorable for short periods, and secondly most of these systems have quasi-definitive orbits, allowing a good matching between the set of observations and the model.

  

Table 4: The 9 astrometric binaries unpublished in the previous paper or reprocessed with different orbital elements, followed by 6 revised systems whose masses have already been presented in the previous work

¹ When available, the more common name of the star, or the discoverer designation.
² The three first characters of the author's name followed by the two last digits of the publication's year.