Annu. Rev. Astron. Astrophys. 1984. 22: 445-70 Copyright © 1984 by Annual Reviews. All rights reserved

4.1 Comparison Samples and Definition of H I Deficiency

As a rule, the parameters that describe the optical appearance of a galaxy are used to estimate its expected H I content, with the assumption that it does not deviate from the standards of normalcy set by some control sample. The ratio between that expectation and the actual observed value of a given galaxy's H I content is labeled the ``deficiency factor'' or the ``H I deficiency.'' It is usually expressed as a logarithmic quantity, positive for H I-deficient galaxies. The reliability of this parameter, which depends not only on the quality of the observed quantities but also on the choice of the control sample and of the criteria utilized to estimate the H I content, has been questioned (11). When rigorous criteria are applied, however, it yields quite satisfactory results (46). Earlier cluster work used as comparison samples those of normal galaxies compiled by Roberts (89) and by Balkowski (6). These samples include mainly nearby galaxies, which are sometimes members of groups or pairs, a circumstance frequently seen to favor disruptive phenomena (as discussed in Section 3). They are also vulnerable to the appearance of Malmquist-type biases in the analysis of distant clusters (17, 43). Larger compilations of H I properties of individual galaxies now exist (18, 69), but the largest comparison sample, rigorously selected to include objects affected as little as possible by interactions with neighbors or violent environments, consists of approximately 300 isolated galaxies observed at Arecibo and Green Bank (56, 58, 63). Of course, most cluster galaxy samples include objects found in different regimes of local density, thus providing a measure of built-in relative comparison.

The various observers have adopted different criteria for the definition of a galaxy's H I content, as illustrated in Section 2.1. Although the most frequently used quantity is log (M_H / L), it is operationally the least accurate indicator of H I content. For the sample of isolated galaxies, the rms dispersions of log (M_H / L) values around mean estimates are high within all morphological types, ranging between 0.3 and 0.4. By comparison, in commensurable units, that scatter is reduced to about 0.21 (when averaged over all types) if one analyzes H I content in terms of the hybrid surface density M_H / D², as outlined in Section 2.1. An added advantage of the second approach is its almost negligible dependence on morphological type; because the estimate of type, especially in faint small angular diameter galaxies, is accompanied by a large margin of error, the minimization of the possible effect of misclassification on the inferences of H I deficiency is a welcome result.