The ionizing cluster of 30 Doradus. I.

A method to obtain an unbiased estimate of the slope is discussed. Astrometric calibration (from B. Skiff): The source images were taken with the ESO NTT 3.6-meter telescope, with a CCD detector at an image scale of 0.13arcsec/pixel in about 0.8arcsec seeing. Positions were determined via a reference net of 26 relatively bright and isolated stars in the region appearing in either UCAC2 (Cat.

I/289) or 2MASS (Cat. II/246). The x,y coordinates were transformed using a quadratic least-squares fit; the rms residuals using all the reference stars were 0.50 pixels in RA, and 0.55 pixels in Dec, i.e. 0.097arcsec total.

A similar re-reduction of the survey by Parker (1993AJ....106..560P), which overlaps this one, shows that the Selman linkages with Parker stars (column 'P93') are incomplete and contain some errors. Some of these result from the relatively small image scale (0.49arcsec/pixel) and poor seeing (1.2-2.0arcsec) of Parker's data, such that a single Parker star is sometimes resolved into multiple components by Selman. Among bright stars (V <~ 15) that have been observed for spectral type there are few such errors in the match-up.

But among Parker's fainter entries (V >~ 17) are numerous false detections (clumps in nebulosity) that are incorrectly linked by Selman with true stars a couple of arcseconds from Parker's non-stellar objects.