3 Reduction and calibration

All frames with double stars have been reduced with a profile-fitting reduction method developed by JC. A special method was necessary since the majority of our frames showed only the overlapping profiles of the double stars, but no further (single) stars on the frames to define the point spread function (for details: see Cuypers [1997]). The astrometric standard stars and a few preliminary HIPPARCOS results were used for a first scale calibration (0.378 $^{\prime\prime}$ pixel^-1). The final calibration of separations and position angles was performed on the basis of all available corresponding HIPPARCOS data (ESA, [1997]) by minimizing the sum of the squares of all the differences in relative positions. Stars with large discrepancies in position (3$\sigma$-level) between the CCD and HIPPARCOS measurements were not included in the final calculation of the transformation, but they can be found in the tables of results.

The final scale correction was 0.9969 for 1991 and 0.9968 for 1992 and the position angle correction 1.280$^\circ$ in 1991 and 3.063$^\circ$ in 1992.

Most nights were of sufficient photometric quality. All instrumental magnitudes taken during these nights through the Bessel V and Gunn ifilters were, after correction for atmospheric extinction, transformed into the Cousins V and I standard system with the aid of the observed standard stars. The differences between both pairs of standard bands (V,i) vs. (V,I) are not negligible but they could be treated safely by the transformation equations.