For the Potsdam programme of linking the proper motions of the provisional Hipparcos catalogue H37cr to an extragalactic reference system we used exclusively deep Schmidt plates (Table 1 (click here)). Most of these plates were taken with the Tautenburg Schmidt telescope (134 cm aperture/200 cm diameter of the mirror/400 cm focal length). In addition, some plates from the Palomar (122/183/307) and UK (120/180/306) Schmidt telescopes were combined with Tautenburg plates (see Col. 4 of Table 1 (click here)). However, this combination of plates taken with different Schmidt telescopes and with different plate centres caused problems in the determination of accurate proper motions of the relatively bright Hipparcos link stars (see Sect. 3 (click here)). For each field, at least two first- and two second-epoch Schmidt plates were used. All plates of the same field were measured with the same measuring device.

Whereas the initial aim was to determine absolute proper motions of Hipparcos stars in 10 Tautenburg proper motion fields (cf. Dick et al. 1987), later on other fields from different Potsdam proper motion projects were included. The final sample of Hipparcos link fields consists of 7 fields from the MEGA programme (Schilbach 1988), 7 fields with globular clusters and 9 other fields with a sufficient number of galaxies as reference points. The results for one field (3C345) were kindly provided by Tucholke (1995) who also used deep Tautenburg plates. Two fields with dwarf spheroidal satellites (dSph) of the Galaxy (Scholz & Irwin 1994) were later excluded from the link due to large proper motion errors of the Hipparcos stars (see Sect. 3 (click here)). Altogether, the Hipparcos link fields with absolute proper motions based on deep Schmidt plates are well distributed over the northern sky (see Fig. 1 (click here)). Each field covers about 10 square degrees. The fields M3-Z (centered on the globular cluster) and M3-IV (with the cluster in the plate corner) overlap by about 25 per cent.

Among all available Hipparcos stars in each field only those with reliable measurements were selected for the proper motion determination (see Fig. 2 (click here)). The reliability of the measurements was estimated from the image parameters used in the classification of objects (APM, MAMA, PDS) and/or from the internal accuracy of the determined proper motions (comparison of independent results from different plate pairs). The number of the reference galaxies used per field varied in dependence on the Galactic latitude as well as on the measuring and reduction process (see Table 1 (click here)).

field	R.A.	Dec.	number of	measuring	baseline	number of	all	reliably	finally
			plates	device	[years]	galaxies	Hipparcos	measured	used stars
	2J1991.25					stars	H37cr stars	()
M31	10.7	40.9	8T	PARSEK	25	70	36	32	20
A193	21.4	7.9	4T	PDS	21	150	23	13	11
M33	23.1	30.4	4T	ASCORECORD	18	60	26	23	10
SNHoffm	37.2	31.6	4T	PDS	29	30	23	15	4
MEGA01	62.0	30.9	4T	MAMA	25	270	23	10	7
MEGA02	67.5	2.0	4T	MAMA	19	120	22	12	11
MEGA04	78.5	5.1	4T	MAMA	25	70	33	20	10
MEGA07	90.0	51.3	4T	MAMA	25	130	30	20	14
Mrk4	100.4	74.9	4T	PARSEK	22	100	33	17	10
MEGA10	116.6	26.1	4T	MAMA	24	30	15	10	5
MEGA11	144.9	49.7	4T	MAMA	19	190	32	15	13
Virgo2	186.9	12.0	4T	APM	25	540	21	11	8
Virgo3	190.2	12.9	4T	APM	24	680	32	10	10
MEGA14	191.3	29.9	4T	MAMA	25	600	30	18	14
M3-Z	205.5	28.6	53T	APM	30	1600	34	14	14
M3-IV	206.8	27.4	4T	APM	22	1050	32	6	6
UMi	226.4	67.2	4P, 3T	APM	35	1280	24	0	0
M5	229.7	2.0	2P, 1UK, 5T	APM	38	1160	38	15	14
M13	250.4	36.6	4T	ASCORECORD	21	50	36	25	20
3C345	250.6	39.8	11T	PDS	20	120	28	17	16
M12	252.3	- 1.4	2P, 1UK, 2T	APM	43	1380	24	7	6
A2255	258.8	63.9	4T	PARSEK	18	190	25	14	7
M92	259.4	43.3	10T	APM	26	2300	27	7	7
Draco	260.2	58.1	4P, 3T	APM	37	820	18	0	0
M15	322.3	12.2	2P, 6T	APM	24	510	28	11	11
Mrk319	351.2	24.4	4T	PDS	21	50	28	21	11
total							706⁺	360^*	256^*

T - Tautenburg, P - Palomar, UK - AAO-Schmidt plates.
⁺ there are 15 common stars in the overlapping field of M3-Z and M3-IV, ^* 3 stars were measured in M3-Z and M3-IV.
Table 1: Link fields

Figure 2: Number of Hipparcos stars in 26 link fields. Solid line: all available Hipparcos stars, dotted line: stars of the provisional Hipparcos solution H37cr provided for the link, dashed line: selected stars from H37cr with reliable measurements. Only reliably measured stars with H37cr data and

(on the right side of the dot-dashed line) were used in the final solution

The first three fields included in the Hipparcos extragalactic link programme were M33, M13 and M31 (Scholz 1991). The first two of them had been measured with the semi-automated measuring device ASCORECORD at the Lohrmann-Observatorium of the Technical University Dresden (Schilbach 1984), whereas the M31 field was measured with the automated measuring machine PARSEK at the Main Astronomical Observatory of the Ukrainian Academy of Sciences at Kiev (Sergeev & Schornikov 1984). All available Hipparcos stars, but only a limited number of visually classified galaxies could be measured with the ASCORECORD. The same was due for the PARSEK measuring machine which needs rough input coordinates of all objects to be measured. In later PARSEK measurements (fields Mrk 4 and A 2255) we used the image classification on APM-measured POSS1 plates to provide input coordinates of reference galaxies in these two fields. This image classification was visually checked during the automated PARSEK measurements.

For the majority of link fields the Schmidt plates were fully scanned by the automated measuring machines APM (Kibblewhite et al. 1984) at Cambridge, MAMA (Guibert 1983) at Paris and PDS (Horstmann et al. 1989) at Münster. In dependence on the available reduction software, the images and/or image parameters of the Hipparcos stars and of potential reference galaxies were investigated and a selection of well measured objects was made.

So we used standard APM software for the classification of all measured objects into stars, nonstellar objects, noise images and merged objects. In order to exclude those Hipparcos stars which were unreliably measured, only those images classified as stars on all plates were used in the proper motion determination. The reference galaxies were selected according to their classification on the deepest plates. The measurements were corrected for the known periodic errors of the APM (Evans 1988; Evans & Irwin 1995).

APM and MAMA measurements of the same two plates were used for a comparison of the accuracy of both machines (Schilbach & Scholz 1992). A somewhat better accuracy for bright stars (in the magnitude range of the Hipparcos catalogue) was found for the MAMA measurements. On the other hand the faint objects (i.e. in the magnitude interval of the reference galaxies) were more accurately measured by the APM.

In the case of PDS measurements we took advantage of the specific output data in the Münster software (Horstmann et al. 1989), i.e. both image parameters and pixel data for each measured object. Due to the special output data format, both automatic image classification and visual check of the measured images are possible. In the proper motion determination we preferred to use those reference galaxies classified by both procedures. Among the measured Hipparcos stars we selected for further reduction only those fitted by a Gaussian profile on all plates.

2. Observations and measurements