4 Discussion

A great number of articles have been devoted to investigation of the gas properties of galaxies (e.g., review by Young & Scoville 1991). In particular, gas-to-luminosity relations of various types of galaxies have been studied extensively. The results of comparison between the gas content and luminosity reveal different effects. For example, in the FIR luminous objects ( $\log{L_{\rm fir}}~\ga 10.5$), the HI content is not correlated with the FIR luminosity, while for fainter objects, there exist a strong correlation between these quantities (Kandalyan et al. 1997). In nearby spirals, the HI content is unrelated to the $\element{H}_{2}$ content (Braine & Combes 1992), whereas in the IRAS-selected galaxies there is a strong correlation between them (see, e.g., present section; Young et al. 1989; Andreani et al. 1995). Thus, gas phase to gas phase and gas phase to luminosity relations in the galaxies are different and very complicated and depend on many internal and external factors such as the environment, luminosity, morphology, star formation history etc.

Up to now, the atomic and molecular hydrogen properties have been investigated on the basis of either optically- or IRAS-selected samples of galaxies. The different samples reveal different gas-to-luminosity relations in the galaxies. For instance, IRAS-selected samples are biased toward galaxies with recent or present star forming activity, while in the optically-selected samples, the past star forming galaxies dominate. In Table 4 we summarized the main results of the comparison of the HI, $\element{H}_{2}$ surface densities and surface brightness obtained from various samples of galaxies, where B, FIR and R denote the blue, far-infrared and radio continuum surface brightness, respectively. The first line in each box corresponds to $S_{\rm HI}$, and the second line to $S_{\rm H_{2}}$. The presence of either HI or $\element{H}_{2}$ indicates the existence of a correlation between the surface density and corresponding surface brightness. "n'' indicates the absence of a correlation between two quantities. An empty field means that the relation has not been considered. Bold letters indicate that the considered correlation is stronger than the relationship between the two other variables in the same box. When several articles were available in the literature for each sample, the results were combined and in the references the most recent papers are presented. Table 4 includes only five samples which, in our opinion, represent the main samples studied for the CO emission so far. We have used the following abbreviations: "Nearby'' - optically selected nearby galaxies; "Starburst'' - optically selected starburst galaxies; "UV-IRAS'' - Markarian galaxies detected by IRAS; "IRAS'' - IRAS selected galaxies; "Cluster'' - clusters' galaxies. These samples are comprised mainly of spiral galaxies. Several important conclusions can be drawn from Table 4.

   

Table 4: Comparison of star formation activities of different samples.

Sample	B	FIR	R	$S_{\rm HI}$	Reference
Nearby	HI	HI	n		1, 2, 3
	H2	H2	H2	n
Starburst	HI	HI			4
	H2	H2		H2
UV-IRAS	HI	HI	HI		5
	H2	H2	H2	H2
IRAS	HI	HI			6, 7, 8
	H2	H2		H2
Cluster	HI	HI	n		9
	H2	H2	H2	n

1. Braine & Combes (1992); 2. Sage (1993); 3. Elfhag et al. (1996); 4. Jackson et al. (1989); 5. Present work;
6. Young et al. (1989); 7. Sanders et al. (1991); 8. Andreani et al. (1995); 9. Casoli et al. (1996).

(a) For all samples, the FIR surface brightness is correlated with the molecular hydrogen surface density more strongly than with the neutral hydrogen surface density. This suggests that, during present or recent star formation, the molecular phase of the gas plays an immediate part in the star formation.
(b) The strong correlation between the HI surface density and the FIR surface brightness for all samples indicates that, in present or recent star formation, the neutral hydrogen phase is also important. Of course, in some galaxies, a part of the FIR emission could originate in the diffuse atomic medium and not be related to the star formation regions. Thus, we can conclude that both phases of the gas are important for star formation. Casoli et al. (1996) arrived at the same conclusion.
(c) It is clear from Table 4 that both phases of the gas are related to the past star forming indicator $S_{\rm B}$. But now the molecular phase does not dominate in this relation for all samples, as it was in the case of $S_{\rm fir}$. Only in the "UV-IRAS'' and "IRAS'' samples did $S_{\rm B}$ correlate with $S_{\rm H_{2}}$ more strongly than with $S_{\rm HI}$. Hence, both components of the gas are important for the past star formation activity.
(d) Unfortunately, the radio continuum data were not available for all samples of Table 4. Nevertheless one can see that there is a strong correlation between $S_{\rm R}$ and $S_{\rm H_{2}}$. In the "UV-IRAS'' sample, the radio continuum surface brightness is correlated also with the atomic hydrogen surface density. The universal correlation between the radio continuum and FIR emission, which is observed from normal galaxies to quasars, is due to the star formation activity (e.g. Kandalyan 1996). Therefore correlation between $S_{\rm R}$ and $S_{\rm H_{2}}$ has the same origin as the former relation. From this point of view the observed relation between $S_{\rm R}$ and $S_{\rm HI}$ for the "UV-IRAS'' sample may indicate that the FIR and HI emissions in these objects are basically due to the star formation while in the "Nearby'' and "Cluster'' samples a part of the FIR and HI emissions could be related to the diffuse interstellar medium.
(e) It is very difficult to understand the absence of correlation between $S_{\rm HI}$ and $S_{\rm H_{2}}$ for "Nearby'' and "Cluster'' samples. In fact, if we believe that both components of the gas are important for star formation, we expect to observe a relationship between them. Perhaps for some samples this relationship is too weak to be observed. But the absence of correlation definitely does not depend on the environment of a galaxy (field or cluster objects).

(f) Table 4 shows that Markarian galaxies do not differ in star formation properties from other galaxies. The resemblance of "Nearby'' and "Cluster'' samples is noticeable.