Astron. Astrophys. 330, 585-599 (1998)

2. The selection of systems and orbits

2.1. The sources of data

According to the results of the simulation presented in Martin et al., 1997, the search of astrometric binaries likely candidates for a mass determination from the Hipparcos observations, must be restricted to pairs with orbital periods smaller than or equal to 30 years. Another selection based on the separation will be considered later. Four main sources of orbits were used to identify the potential systems :

1. A file of orbits available at the Observatory of Côte d'Azur from P. Morel and P. Couteau (identified by 'OCA' in Tables 4 - 5). It is indeed a compilation of orbits coming from several sources, and particularly from the Catalogue of Worley. It includes 905 orbits of 800 systems and contains additional information like the magnitude at 550 nm. It happens that several orbits may be proposed for the several pairs or for hierarchical systems containing more than one pair, which explains that the number of orbits is larger than the number of systems.
2. The file of orbits from the Royal Observatory of Belgium kindly provided by J. Dommanget, containing 864 orbits of 838 systems ('ORB' in Tables 4 - 5).
3. The fourth catalog of orbits of visual binary stars (Worley & Heintz, 1983), including 928 orbits for 847 systems (a triple star being identified as two systems).
4. Finally, searches in the published literature were necessary to update the information contained in the previous files and provided several new orbits recently computed.

The above sources are in practice largely redundant. The intersection of these four sources and that of the Hipparcos observing program combined with the limitation in period, yields at the end a set of 302 orbits for 191 different entries of the Hipparcos Catalogue. This constitutes the basic data set to be investigated.

2.2. Elimination of objects

Among these 191 objects, there are several systems which cannot be processed by the method of Paper I, because of the parasitic effect of a third component in the vicinity () of the central pair. This led to the rejection of 16 systems from the initial set. In addition, two more systems were eliminated as they were finally not successfully observed by the satellite (HIP 21088 and HIP 116191). After this step, we were left with 277 orbits for 173 objects.

There were in this sample 28 astrometric pairs for which the published orbits referred to the absolute motion of the photocentre on the sky, instead of the relative motion of the two components. In this case, nothing more could be done with the Hipparcos data and these stars had to be removed from the sample. Eventually, the useful sample numbers 242 orbits associated to 145 objects listed in 1by their HIP identifier.

2.3. Description of the subset of binaries

2.3.1. Statistical description

The first description proposed here aims to recognize the pairs for which the independent determination of the mass and intensity ratios seems possible, from those for which the values of the period and the separation will not allow this distinction. Fig. 1a reveals a non negligible population of interesting pairs with semi-major axes larger than 0:0025, but also with periods generally larger than 15 years. It shows also the difficulty, expected indeed, to find binaries with both large separations and small periods. Fig. 1b shows the distribution of the largest separation on the sky reached by every pair during a complete revolution (as long as the Hipparcos observations are concerned, this quantity is more representative than the semi-major axis).

Fig. 1. Statistical description of the file of 145 orbital pairs with periods smaller than 30 years. The most recent orbit has been chosen for each object. In b the abscissa is the largest apparent separation over the orbital period.

2.3.2. An adapted characterization

As the sample of objects considered is not too large (compared to the whole set of about 12000 binaries for which an astrometric solution is published in the final Hipparcos Catalogue), it has been possible to define an 'identity card' of each selected star, allowing to illustrate the potential of Hipparcos in each case. In the following diagrams, each 'card' appears as a segment and four numbers, with the following meanings:

• The number in abscissae of each segment is the Hipparcos identifier of the star (HIP number).
• The ordinates at the top and bottom of the segment represent the maximum and the minimum apparent separations reached by the double star during an orbital period, computed from the orbital elements.
• An open box on each segment shows the range of apparent separations covered by the Hipparcos observations during the 3 years of the mission.
• The number of epochs of observations stands on the right of each box. An epoch corresponds to a combination of a few consecutive observations carried out within one or two days, while the interval between two epochs is typically of six weeks; the different epochs are generally well distributed in time over the mission duration.
• The approximate global Hipparcos magnitude of the object and the orbital period in years (from the most recent orbit determination) are indicated at the top of each segment.

This information is displayed in Fig. 2. Independently, on each diagram three horizontal lines have been drawn and represent the theoretical thresholds in apparent separation beyond which it becomes possible to distinguish the photocentre and the Hippacentre of the star, for a system of 2 mag (bottom line), 10 mag (middle line) and 12 mag (top line). In connection with the results of the simulation of Paper I, the previous representation allows to estimate quickly what can be expected from the processing of each star, assuming that the orbit is perfectly known. However only about 25% of the published orbits are considered as 'very good' or 'definitive' according to the criteria of the Worley's catalog; this is probably the main reason why significant results could not be obtained for all of the stars considered here.

Table 1. HIP numbers of the 145 preselected systems

Table 2. The set of astrometric binaries for which a ground-based orbit has been used to reprocess the Hipparcos observations for determining the mass of the components. The columns give the Hipparcos, ADS and HD identifiers, the usual name, the semi-major axis of the relative orbit in arcsec and the orbital period in years.

Table 3. Same set as in Table 2, showing the Hipparcos results from the standard processing. The labels C, O, G, X refer to the sections of the Double and Multiple Systems Annex in which the solution has been placed. A blank in this column indicates that a single star solution has been adopted.

Fig. 2. Description of the sample of short period binaries, as found in the diagrams of Fig. 3. If the orbit is perfectly known, the ability of Hipparcos to obtain the mass ratio directly is as important as the global magnitude and the period are small, as the number of observation's epochs is large, as the range in separation covered by Hipparcos is large compared to the total possible range on the sky (size of the segment), and as the separations observed are large.

Fig. 3. Individual description of some interesting binary systems as sketched out in Fig. 2. The top diagram contains stars for which a solution has been found, divided into 'Type I' and 'Type II' solutions. On the bottom diagram are represented some potentially interesting stars (the processing can be successful for these objects if the knowledge of the orbit is improved). The three horizontal lines, labelled by the global magnitude , are the theoretical thresholds in separation beyond which photocentre and hippacentre are no longer alike. Caution: the vertical scales are different in the two diagrams.

© European Southern Observatory (ESO) 1998

Online publication: January 16, 1998
helpdesk.link@springer.de