Annu. Rev. Astron. Astrophys. 1994. 32: 115-52 Copyright © 1994 by Annual Reviews. All rights reserved

The Limitations of Galaxy Catalogs

All catalogs have limitations because of the adopted inclusion criteria and their degree of completeness, and conclusions drawn from the examination of any catalog (or dataset) must consider those limitations. Of particular importance, as noted by many authors, are biases against low surface brightness objects or the lack of homogenous sky coverage or survey depth. All of these issues are relevant to the current analysis. Given the catalogs from which we have drawn the sample analyzed herein, even before beginning, we want to emphasize the following failings of our analysis:

1.	Low surface brightness galaxies, including both the low mass, low luminosity dwarfs and the low surface brightness giants such as Malin 1, are not included in our results.
2.	Likewise, compact high surface brightness objects are also excluded. These include both compact BCD's and the smaller galaxies in clusters.
3.	Malmquist bias certainly is present in these data. Nearby samples include a predominance of low luminosity, low mass objects that are absent in samples at larger distances. Non-detections of HI and IRAS flux provide meaningful upper limits only for nearby samples. This is discussed further below.
4.	Types earlier than Sa suffer from a large fraction of non-detections in the 21 cm line measurements. While estimates of the HI mass and surface density can incorporate upper limits to the detected flux (as discussed below), total mass calculations that use the 21 cm line width as a measurement of the rotational velocity cannot be made for non-detections. Hence all discussions of such properties as total mass, mass surface density, and mass-to-light ratio are limited to S0/a-Im. For the Im objects, only the brighter, higher surface brightness members of the class are included, for the S0's only those with HI detections are included.

MALMQUIST BIAS The RC3 is to some degree magnitude-limited, and in most instances, the Malmquist bias seriously effects the availability of data for the present analysis. Figure 1 demonstrates that the Malmquist bias seriously affects the sampling of optical luminosities L_B. The upper panel shows clearly that the low luminosity galaxies are only sampled nearby, becoming increasing absent in samples at larger distances. The figure also demonstrates that there is a maximum luminosity near ~ 10^11.5 L.

Figure 1. (Upper). Blue luminosity versus corrected velocity. The data representing 7930 galaxies in the RC3-UGC sample clearly illustrate the Malmquist bias. (Lower). The total mass-to-luminosity ratio for 2864 galaxies in the RC3-UGC sample versus velocity. This ratio has a distance term but the data show essentially no distance or Malmquist bias.