The plates used for the proper motion determination (see Scholz et al. 1996) were taken with the Tautenburg, Palomar and UK Schmidt telescopes and are listed in Table 1 (click here). All plates were measured with the APM measuring machine in Cambridge/UK (Kibblewhite et al. 1984). For more details of the measuring process we refer the reader to our earlier globular cluster papers (Scholz et al. 1993, 1994). A comparison of the quality of the different Schmidt plates, concerning their astrometric use is given in Scholz et al. (1996). Whereas on the Palomar and UKST plates M 5 was close to the plate edge the Tautenburg plates were taken centered on the cluster.
Figure 1: Distribution of the proper motion errors for
images with
stellar (solid line) and non-stellar (dotted line)
classification
a) all objects excluded in the membership
determination, i.e. all faint objects (B > 19)
and all objects within 5 arcmin from the cluster
centre
b) the objects used in the determination of
membership probabilities (in the solution of (3))
i.e. all objects with 5 arcmin < r < 50 arcmin
and with B < 19.
Note the difference between the proper motion errors
shown in this diagram and the
proper motion dispersions 
obtained
in the membership analysis
All measured objects were classified into stars, nonstellar objects, noise images and merged objects using the standard APM software. In the case of M 3 (Scholz & Kharchenko 1994) we included only the objects classified on all plates as stellar in the membership determination. Here we are interested in a more complete sample of cluster members, i.e. including the large number of objects in the cluster region misclassified as non-stellar (galaxies or merged objects) due to crowding effects. Of course we have to consider the much lower accuracy in the proper motions and in the two-colour photometry of these objects.