2. Observations and data reduction
Our data 1 were taken in April, 1993 (during 13 nights in Dutch time) with the m Dutch telescope (1st dataset). A second, very limited dataset was taken on the 6th February, 1996 with the m Danish telescope.
The Dutch telescope was equipped with a pix2 Tektronix CCD (ESO #33) with a scale factor of pix-1 ( pix = m) and a total field of view of , the Danish telescope with a pix2 Loral/Lesser CCD (W11-4 Chip) with a scale factor of pix-1 ( pix = m).
The 1st dataset contains 25 CCD fields (0-24, see Fig. 3) with an overlap of about 80 pix. They form two coherent strips, one pc strip through the OB superassociation LH 77 from east to west with 10 fields (i.e. CCD positions which cover an area of each) and an pc strip from the south to the north with 15 fields, reaching the rim of the supergiant shell LMC 4 (see Figs. 1 and 3 ). The whole 'J'-shaped area is about 298 in size (without the 16% of overlapping regions). For each position we have long exposed frames of 10 min in the B and 5 min in the V passband (ESO #419 and #420) reaching down to approximately mag (fields 0-4 corresponding to region a, see Sect. 4) or to mag (fields 5-24 corresponding to regions b-e) and short exposed frames of min in B and min in V, all with a seeing of 1.3 -2.6 . The 2nd dataset contains 3 CCD fields (, and , see Fig. 3) at the beginning and the end of the area covered by the 1st dataset. This set was obtained to get a better calibration of the main dataset, so we took short exposed frames in the B and V passbands (ESO #450 and #451) reaching down to mag with a seeing of 1.8 -2.7 .
The data reduction was carried out with MIDAS and the built-in photometry package DAOPHOT II of Stetson (1987). Additionally some special steps (e.g. correction of bad double columns and reading the time information of the header) was done with IRAF.
To calibrate the frames we used the following standard fields of Landolt (1992): Rubin 149, PG 0918 029 and PG 1633 099. Because of wrong time information in the file headers we were only able to use the standard fields of night 8 and 9 of the 1st dataset (i.e. fields 11-16) and to calibrate the beginning (i.e. fields 0-3) and the end (i.e. fields 23-24) of the 'J'-shaped region by the 2nd dataset. To get a homogeneous dataset (see Table 1) we adjusted the magnitude levels of the fields 4-10 and 17-22 by their overlapping areas.
Table 1. Part of the table of photometric results around Sk 67 198 = MACS J0534-670#003 (Magellanic Catalogue of Stars: Tucholke et al. 1996, (see Table 3). The position of the stars in the (rough) general coordinate system and on the long exposed B frames are given with the calibrated B, V magnitudes and total errors (statistic and systematic) and the dereddened and values. The table with 20 812 stars is available electronically (see the footnote to Sect. 2)
with index n indicating the normalization to exposure time s and airmass 0 and of the shift from point spread function (PSF) to aperture magnitudes. The airmass correction was made by means of the atmospheric extinction coefficients measured on La Silla by the Geneva group (Burki et al. 1995a, b).
The mean errors of our photometry are given in Table 2. The total error contains all statistical and systematical errors (including PSF fit, calibration and PSF to aperture shift as their main part) and is rather an overestimation (see Fig. 2), while the statistical error from DAOPHOT is a clear overestimation of the reached accuracy.
Table 2. Mean total and statistical errors with standard deviation of stellar V magnitudes and colours. The number of stars in the corresponding magnitude range is given in parentheses
There is no CCD photometry inside LMC 4 in the literature overlapping with our data, so we compared the brightest stars of spectral type B0 to A9 and luminosity class I/Ia marked by Sanduleak (1969) with the B, V magnitudes of Rousseau et al. (1978).
This leads to an unexpected high deviation of the 1st dataset, which concerns only up to five of the brightest (evolved) stars of a CCD field, so it won't effect the results of the isochrone fit. For all other stars the values of the 1st and 2nd dataset agree within the scope of the errors.
To calculate absolute magnitudes we used the distance modulus of mag (see Westerlund 1990, and refs. therein) corresponding to a distance of kpc.
For identification purpose we looked for stars having MACS entries (Magellanic Catalogue of Stars) to get good positions;
Table 3. Cross identification of the stars in our analysed area with the MACS (Magellanic Catalogue of Stars: Tucholke et al. 1996), which may serve as an astrometric reference grid. As an example we give the data of 5 stars around Sk 67 198 = MACS J0534-670#003. The field sequence number gives the CCD field (here 2) and the sequence number in the original DAOPHOT table, the x and y coordinates show the position on the long exposed B frames. The entire table is available electronically (see the footnote to Sect. 2)
Examples of the resulting colour-magnitude diagrams (CMDs) are shown in Fig. 2.
© European Southern Observatory (ESO) 1997
Online publication: March 24, 1998