2. Observations and data reductions
The data were collected on October 20, 1997 at the ESO Danish 1.54m telescope equipped with DFOSC. The camera employed a pixels Loral CCD, with a pixel size of 0:0040 on the sky, for a total field of view of .
Table 1 lists the complete log of the observations. The weather conditions were good during the night, which was stable and photometric, and the seeing was .
Table 1. Journal of the Pal 12 observations for Oct 20, 1995
The image processing was carried out within the IRAF environment. First, a map of the bad features of the chip was created and they were removed from the raw images. Then, the bias stability was checked by comparing frames taken at different times during the entire run, and no significant discrepancies were found. A 0.4 % spatial gradient was found along the x direction, thus a master bias image was created by taking the median of all the bias images. This master bias image was subtracted from all the remaining frames.
Sky flats were used to create master flat fields as medians of the single frames.
In order to avoid the fall of quantum efficiency (QE) all around the border of the Loral CCD, we cut our images outside the limit where the QE was 90% of the central value. From an inspection of the flats this limit imposed an effective area of pixels (i.e. ; Saviane & Held 1998, hereafter SH98, give further details). The effective field is schematically represented in Fig. 1.
Observations of Landolt's (1992) standard stars were used to calibrate the photometry. In addition, the shutter delay time was measured with a sequence of images taken with increasing exposure times. A value of s (where the error is the standard deviation) was found. The raw magnitudes were first normalized according to the following equation
where are the instrumental magnitudes measured in a circular aperture of radius (SH98), is the shutter delay and X is the airmass. For the extinction coefficients we adopted and (from the Geneva Observatory Photometric Group data).
The normalized instrumental magnitudes were then compared to the Landolt's (1992) values, and the following relations were found:
where the uncertainties represent the 90% confidence ranges of the fit for one interesting parameter. The standard deviations of the residuals are 0.013 mag in V and 0.022 mag in I, respectively.
In order to transform the PSF magnitudes into aperture magnitudes we assumed that (Stetson 1987). For each individual frame a sample of bright isolated objects were then found, and all their neighbors were subtracted. The `cleaned' images were used to measure aperture magnitudes for the selected stars, and for each star we computed the difference with respect to the PSF magnitude obtained on the averaged frames. The same aperture used for the standard stars was employed. The internal uncertainty of the calibration of the order of 0.01 mag for each filter (cf. SH98).
Our photometric catalogs are compared with those of Harris & Canterna 1980 (HC80), GO88, S89 and DA90 in Fig. 2. Noticeable differences are found for the V band, where a mag is present between our values and those of HC80, S89 and DA90, in the sense that our magnitudes are fainter, and an even larger difference is found between our data and GO88 ( mag, cf Fig. 2). For the I band, only the DA90 data allow a comparison, and we find that the two calibrations match within the errors.
We tried to sort out the possible reason for the observed discrepancies in the V band, while no significant differences are found for the I band. From a comparison with existing photometry of the Fornax dwarf galaxy, SH98 conclude that their V band calibration is consistent with the previous works. A possible source of uncertainty could be a problem with the V exposure times: however, the (small) shutter delay has been included in Eq. 1. Moreover, when the individual zero points of the 7 available V frames are compared, no differences larger than 0.01 mag are found, which furtherly confirms that there is no shutter delay problem. In principle, thin cirrus could have been present at the beginning of the night, although it should have blocked a remarkably constant percentage of light during the min time span of the cluster observations, which seems unlikely. The above arguments lead us to trust our calibration, although further checks are needed in order to settle this issue.
In any case, the global zero-point difference in V between our Pal 12 photometry and that of the previous works will not affect our conclusions on the relative age of this cluster.
© European Southern Observatory (ESO) 1998
Online publication: September 30, 1998