The data processing of the Hipparcos observations was conducted by keeping the number of hypotheses regarding the structure of the light sources to minimum. In particular the signal recorded behind the grid was compressed in such a way that no information linked to a possible multiplicity was lost during this process, allowing at a latter stage to recover the individual sources and their relative position and brightness. A similar care was applied to the astrometric model used to describe the varying position of the sources on the sky, keeping all the way through the analysis the possibility to decide whether the detected motion was linear or accelerated, or even more complex in order to select the model accordingly.
When a Hipparcos target was a single object, the standard astrometric model in the data reduction assumed a uniform rectilinear motion in space relative to the barycentre of the solar system. In this case only five parameters were necessary to describe the full astrometric information embedded in the Hipparcos measurements: the two angular coordinates to specify the coordinate direction at the Catalogue epoch ( =J1991.25), the two components of the proper motion used to propagate this position at any other past of future epochs and the parallax, which, besides its obvious astrophysical interest, allows to determine the proper direction from the Earth.
In the case of a non single object, known before the mission or recognized by Hipparcos, the situation was more complex and additional parameters were needed to model properly the observations, the number of which was variable with the type of multiple system. The treatment was much influenced by the separation and the period of the orbital motion, and corresponds more or less to the usual distinction made for the ground-based observations, between the visual and the astrometric binaries.
For double stars with an orbital period much longer than the duration of the Hipparcos mission, the data processing was adapted to cope with this difficulty and eventually this led to a good decoupling between the relative and absolute astrometry for more than binary systems included in the Catalogue. The various principles applied to the recognition of these systems and the properties of the relative and absolute astrometric solution have been presented in detail in the literature (Mignard et al, 1992, Mignard et al, 1995) and will not be repeated here. The definitive and most detailed information is now available directly in the Hipparcos documentation (ESA, 1997).
In this work, we are more concerned with the close pairs with an orbital period less than 10 years which could exhibit a significant non-linear motion of their photocentre over the observation timespan. The amplitude of this motion is related to the size of the relative orbit of the two components and to the difference between the mass and intensity ratio.
For very close binaries with a maximum separation less than few mas, the satellite could not show any photocentric displacement beyond the linear proper motion of the barycentre and the entries were processed in the same way as the single stars. The most interesting cases, as far as the masses are concerned, are to be found in the orbital pairs which fulfilled the following two criteria :
Obviously these two constraints imply that the interesting systems are located not very far from the solar system, a circumstance favorable to obtain the parallax with a good relative precision and consequently to determine the total mass of the system, but at the same time a very demanding constraint which considerably narrows the number of good candidates.
In these cases the curved or wavy displacement of the photocentre was significant and could be separated from the proper motion, provided provision for such a motion was made in the astrometric modeling. Although it was practically impossible to determine the full set of orbital elements only from the Hipparcos observations, it is well known that the photocentric orbit and the relative orbit are similar in shape, differing by a scale factor which depends only on the mass of the components and on the magnitude difference. Therefore, including this scale factor as an additional unknown to the standard astrometric model, permits to derive the masses of the components alongside the positions, parallax and proper motion. Martin et al., 1997, have shown how to extend this basic property to the one dimensional Hipparcos observations and that, in some cases, both the mass ratio and the intensity ratio could be derived from the absolute motion of the Hippacentre, when the latter could be decoupled from that of the photocentre.
The first paper (Martin et al., 1997) dealt primarily with the methodology. The capabilities and limitations of the method were primarily assessed from a monte-Carlo simulation and the paper ended with a short list of likely good systems worth testing. In this second paper we apply the method to more than 40 real systems of the Hipparcos program, found to have the appropriate characteristics. The first section sums up the sources of the orbital data and closes with the selection of the 140 systems for which a solution should be attempted. The processing is presented in the second section and the astrometric results are shown and discussed in the following section. The masses are derived in the fourth section together with the discussion of their reliability and the comparison to published results.
© European Southern Observatory (ESO) 1998
Online publication: January 16, 1998