445 ACCURATE TWO-COLOUR PHOTOMETRY AND ASTROMETRYFOR HIPPARCOS DOUBLE STARS: FIRST STATISTICAL RESULTS FOR A SAMPLE OF INTERMEDIATE BINARIES 1 2 3 3 1 2 E. Oblak , J. Cuyp ers ,P. Lamp ens , W. Seggewiss , M. Chareton , D. Duval 1 Observatoire de Besancon, France 2 Koninklijke Sterrenwachtvan Belgie, Belgium 3 Hoher List Observatorium, Germany 2. OBSERVATIONS, REDUCTION AND ABSTRACT CALIBRATION A large numb er of new astrometric and photometric CCD measurements of intermediate visual binaries The observations were gathered at ESO, La Silla angular separation b etween one and fteen arcsec, in Octob er 1991 J. Cuyp ers, in February 1992 literature values and Hipparcos data are compared. W. Seggewiss, in August 1992, Decemb er 1993 and The claimed mean relative p osition errors 0.02 arc- August 1994 P. Lamp ens, in November 1992 and sec for b oth the CCD and Hipparcos are con rmed. Novemb er 1994 E. Oblak. The 91-cm Dutch tele- For 8 p er cent of the systems in the sample we had scop e was used with di erent CCD camera's with no Hipparcos solution and in ab out 6 p er centofthe pixel sizes in the range 0.3 to 0.5 arcsec. cases the CCD and Hipparcos solutions were to o dif- ferent to be used in the comparison. The disp er- sion between literature values and CCD results for The raw CCD images were treated in a standard way, the joint magnitudes of the systems was 0.03 mag. but a sp ecially develop ed pro le tting pro cedure A comparable disp ersion is found only for the Hip- Cuyp ers 1997 was used for extracting the relative parcos data when colour corrections for b oth comp o- p ositions and di erential magnitudes when the pro- nents supplied by the CCD observations are taken les of the comp onents overlap. into account. In that case the mean errors on in- dividual comp onents from Hipparcos data are ab out Photometric observations were carried out with 0.04 mag for the primaries and 0.08 mag for the sec- Bessel V and Gunn i lters. Observations of stan- ondaries. dard stars from a list compiled by Grenon 1991, Key words: double stars, photometry, astrometry. based on lists of Landolt 1983 and Menzies 1989, were used to correct for extinction and to transform the data into the standard V R I system. 1. INTRODUCTION To calibrate the relative astrometry a set of wider pairs taken from the lists of Brosche and Sinachop ou- From 1991 on, a photometric observational pro- los 1988, 1989 were measured. Stars were trailed gramme of intermediate visual binaries angular sep- over the frames for a rst estimate of the orienta- aration b etween one and fteen arcseconds, di erence tion of the CCD detector. In some campaigns op en in magnitude less than 4 mag has b een carried out clusters could already be used to calibrate the as- in the framework of a Europ ean Network of Lab ora- trometry Sinachop oulos al. 1993. This resulted in tories Oblak et al. 1992a. The stars selected for the a very accurate scale and orientation calculation as programme had no complete comp onent photometry shown b elow. listed in the data base compiled at Besancon Kun- dera et al. 1997. Observations were made at di erent observatories La Palma, Teide and Calar Alto, Spain and Haute- Provence, France, but here we rep ort on data ob- 3. EVALUATION OF THE CCD DATA tained solely at the Europ ean Southern Observatory Chile, where several campaigns and a Key Pro- gramme Oblak et al. 1992b for this purp ose were carried out. Since each CCD observation consists of a rep etitive sequence of frames up to 15 for large m and b e- Based on observations obtained at the Europ ean Southern Observatory, Chile and on data from the Hipparcos astrometry cause some double stars were measured in di erent satellite. runs, an internal evaluation is p ossible. 446 3.1. Astrometry Toevaluate the quality of the photometric observa- tions a comparison was made b etween the measured and calibrated global V magnitudes of the systems Internal p osition errors are smaller than 0.015 arcsec, joint magnitude of comp onent A and B and the while the errors on the means are 0.004 arcsec in gen- values available in the catalogues Mermillio d et al. eral. An increase of the error with larger di erential 1996, Kundera et al. 1997. Of 133 double stars of magnitudes is obvious. There is also a degradation of our sample a global V magnitude was available, ei- the consistency for separations smaller than 3 arcsec ther from Stromgren 96 or from Geneva photome- due to seeing limitation and for separations larger try 72. The histogram of the di erences is given in than 9 arcsec mayb e due to loss of isoplanicity. Figure 1. We do not know whether the slight o set 0.006 mag b etween the two sets is signi cant. The disp ersions of the di erences are 0.028 mag for the The mean value of the di erences in p ositions for the Stromgren and 0.030 mag for the Geneva data. In 28 systems observed more than once was 0.024 arc- this disp ersion the errors on the literature values are, sec, but the median was as small as 0.013 arcsec, since of course, also included, so 0.03 mag is undoubtedly 4 systems have di erences greater than 0.07 arcsec. avery conservative upp er limit of the mean error on The errors due to the calibration of the scale and ori- the CCD magnitudes. No signi cant dep endence of entation of the CCD for each campaign are included the di erences on the system parameters was found. here and some of the systems were remeasured be- A more detailed error analysis is given in Lamp ens cause the rst observation was uncomplete or not et al. 1997. satisfactory. 0.15 3.2. Photometry Hipparcos error Hipparcos − CCD CCD errors (internal) Internal errors on the di erential and derived com- p onent magnitudes are very small <0.005 mag. There is some degradation for separations smaller 3 arcsec. The standard stars measurements than 0.10 show deviations smaller than 0.03 mag after calibra- tion and transformation to the standard system. The blue V I < 0:2 mag and, as exp ected, the very red stars V I>1:2 mag sometimes deviate more. Only a limited numb er of double stars 14 was mea- Arcseconds The median deviation of the sured more then once. 0.05 magnitudes was only 0.007 mag for the primaries, 0.013 mag for the secondaries and 0.012 mag for the di erence in magnitude b etween A and B. The mean and standard deviation of the di erences were 0.02 mag, but it is not excluded that some comp onents are variables. 0.00 0 5 10 15 Separation (Arcseconds) Figure2. The claimed internal errors on CCD and Hip- parcos relative astrometry and the position di erences be- tween CCD and Hipparcos results for 416 double stars. 4. COMPARISON WITH HIPPARCOS DATA 4.1. Astrometry A total of 488 double star systems were used for a detailed comparison b etween our CCD data and the Hipparcos results the 28 systems measured twice were considered only once. For 41 systems 8 per cent we had no Hipparcos solution, while for 31 systems 6 per cent the p ositional di erences were larger than 0.15 arcsec. The errors on the Hipparcos solutions for the latter cases were well ab oveaverage. Some of these systems had small separations and, as a consequence, larger errors on the CCD measure- Figure1. The distribution of the di erences between the ments. In other cases a di erent, wide, usually faint, global V magnitudes in literature and the CCD results. comp onentwas considered as the secondary. In two 447 cases Hipparcos detected a very close companion, not colour information is necessary to compare the Hip- recognized in the CCD images. For 15 other systems parcos magnitudes to the Johnson V magnitudes, wehave no explanation for the large di erences yet. otherwise errors could be as large as 0.2 mag Fig- ures 4a and 5a. We used our measured V I colour For the sample of 416 double stars the b est matchbe- index for the transformation from Hp to V magni- tween the CCD and Hipparcos p ositions was searched tudes following the indications in the Hipparcos cat- in the least-squares sense for each observing cam- alogue ESA 1997. The errors after correction are paign separately. For some campaigns no orientation given in Figures 4b and 5b and the distribution of change was necessary, since it was found to b e accu- these errors is also shown Figures 4c and 5c. The rate the 0.1 level, due to the well de ned star trails rms of the errors for the A comp onents is ab out or the go o d calibration with the aid of the astromet- 0.05 mag and ab out 0.09 mag for the B comp onents. ric standard elds. A scale correction was usually Since we estimated the error of the CCD V magni- necessary with a nal accuracy at the 0.1 per cent tudes to b e 0.03 mag, in agreement with the compar- level. ison to the literature values, a go o d estimate of the The di erences in p ositions are given in Figure 2, errors on V magnitudes derived from Hipparcos Hp is together with the estimated Hipparcos errors and in- ab out 0.04 mag on average for the A comp onents and ternal CCD errors.