The selection of objects included in this book is of considerable importance and we shall make some comments on this point.

To trace the curves of equivalent widths for objects of different spectral types and luminosity classes we have relied only upon MK standard stars. This was done for three reasons. First of all, because MK standards have well-established classifications. Second, they have also been closely scrutinized for spectrum anomalies (both for general line weakening and/or for abnormal line strengths of individual elements). Third, they have been observed so often that spectrum variability can be excluded. If, in the papers from which the equivalent widths were taken, the authors have observed other stars besides the MK standards, then these were not taken into account. When nothing else was available, in a few explicitly mentioned cases, we have accepted stars that are not MK standards.

The use of spectral types and luminosity classes, rather than temperatures and log g values, is justified by the fact that the former are observable characteristics of stellar spectra and are thus stable, whereas temperatures and gravities are fitted parameters and are thus subject to changes in calibration.

The restriction to MK standards has been applied consistently with regard to the behavior of atomic lines. They could, however, not be applied in the chapter on molecules, because the number of stars in which molecules have been measured is too small to allow such stringency in the selection of stars. So we have taken whatever data were available, provided that the classification of the stars was in the MK system.