4 Errors

The total internal errors of the proper motions presented in Table 2 were obtained by means of the relations given in [7, Wroblewski & Torres (1989)]. The errors in the measurement of the target and of the background reference stars used in the above relations were average values previously determined (this because only one setting was made on each reference or target star, which prevented a more direct calculation of the measurement error). The values adopted varied- depending on the magnitude of the star- between $\pm0.10\hbox{$^{\prime\prime}$}$ and $\pm0.21\hbox{$^{\prime\prime}$}$ for the X coordinate, and between $\pm0.10\hbox{$^{\prime\prime}$}$ and $\pm0.23\hbox{$^{\prime\prime}$}$ for the Y coordinate. Precision levels between 7 and 13 mas/year were achieved for the proper motions.

To check the reality of the errors estimated as described above, many pairs of plates with similar epochs have been reduced throughout our survey. For timebases of roughly one year, background reference stars used in the final calculations, known to have virtually null proper motions in a timebase of about 20 years, show however calculated proper motions of the order of 0.2 arcsec/year in the mean. These "motions" are simply a reflection of the overall error of the procedure for $T\sim1$. In a timespan of 24 years- which is the mean timebase of the results given in this paper- this error is reduced to about 0.008 arcsec/year, figure which is in good agreement with those given in Table 2 and in Sect. 5.

Although we did not determine individual total internal errors for the positions of the fast moving stars, based on extensive experience with a similar observational and reduction set-up (see e.g. [2, Costa & Loyola 1989]; [3, Costa & Loyola 1998)], we estimate that these errors are in in the range 0.15$\hbox{$^{\prime\prime}$}$ to 0.25$\hbox{$^{\prime\prime}$}$.