The precision of spectroscopic luminosity in A-F type stars and the classification of AM stars

We analyze the work of three teams of classifiers providing spectral classifications in the range of late B-type to early F-type stars. With the help of a statistical model we derive an accuracy of +/- 0.7 luminosity classes for the spectroscopic luminosity class provided by classifiers. This is sligthly better than one luminosity class, and corresponds, in the region of the A-type stars, to a precision of about +/- 0.4 (absolute) magnitudes. We do not find differences in the precision of three series analyzed. We analyze also the case of Am stars and conclude that the probability that two independent classifiers agree that a given object is Am, is of the order of 60%.