Results

The lower panel of Fig. 1 shows the observed distribution of HI masses binned per half-decade, with errorbars given by Poison statistics. The HI masses were calculated from either the VLA observations, or from the Arecibo measurements if the fluxes were lower than 1.0 Jy km s.

 
Figure 1: Lower panel: The distribution of HI masses of the detected galaxies. with errorbars given by Poison statistics. Upper panel: The thin line is the sensitivity of our survey. The measured HI mass function per half decade is shown by the points. An analytical HIMF is represented by the fat line, using the parameters given in the upper right corner. The arrows show upper limits to the volume density of HI clouds. The two measurements on the right are from a complementary survey with the Arecibo telescope over the range 19,000 to 28,000 km .

The inverse of the survey volume as a function of HI mass is indicated by the thin line in the upper panel of Fig. 1. The curve indicates the upper limit to the space density of intergalactic HI clouds without stars as a function of HI mass.

The HIMF was determined following Schmidt's (1968) method, which consists of summing the reciprocals of the volumes corresponding to the maximum distances to which the objects could be placed and still remain within the sample. For a survey such as the Arecibo Strip Survey, is a complicated function, dependent on velocity width, total flux, declination offset from the center of the survey strip and feed gain, which is a function of frequency (i.e. redshift).

The solid dots in Fig. 1 show the HIMF. Briggs (1990) derived an analytical expression for by using a Schechter luminosity function and a relation between HI richness and optical luminosity: L, , where . This function is represented by the fat solid line, using , a faint end slope and a normalization , which is a satisfactory fit to the points. The parameters of this fit agree quite well with those of optical luminosity functions. Hence, an HI selected sample of galaxies does not yield a population of gas rich dwarf galaxies (Tyson and Scalo 1988), that might be missed by optical surveys. If a large population of underluminous galaxies does exist, they must be either HI deficient, or have extremely low column densities ().

 
Figure 2: Space density of HI mass contained in objects of different masses per half decade. Thin line indicates again the sensitivity of the survey.

The cosmological mass density of HI at the present epoch, , can be determined from the distribution function of HI mass in galaxies. This function is plotted in Fig. 2. The fat solid line indicates the converted best fit HI mass function, the thin line represents the sensitivity limits. The distribution function clearly illustrates that the integral HI mass density is dominated by high mass galaxies, which are galaxies. From this figure we derive that or , with a statistical error of . This result agrees surprisingly well with earlier estimates by Rao and Briggs (1993), who find the same value by using optically selected galaxies. This implies that there is not much neutral gas hidden in objects like LSB galaxies or intergalactic clouds that are missed by optical surveys. The ratio of HI mass density to the critical mass density of the universe at z=0 is , consistent with a smooth decline of from high z to the present.

 
Figure 3: The volume corrected surface brightness distribution of our HI selected galaxy sample. Hatched area shows the possible range of values for this distribution function. The two lines represent the upper and lower limit to the distribution proposed by McGaugh (1996). The y-axis is arbitrarily scaled.

One final result concerns the distribution of central surface brightnesses. Since this galaxy sample is selected regardless of any optical properties, it is well suited to test the distribution function of optical surface brightnesses. The hatched area in Fig. 3 indicates the possible range of values for the distribution function for the 24 galaxies observed so far. Despite the large variations due to small number statistics, it is clear that this distribution is consistent with the `flat' distribution proposed by McGaugh (1996), of which the boundaries are given by the dashed and dotted line. It is noteworthy that no galaxies observed thus far have central surface brightnesses fainter than -, even though the measurement threshold is -. We therefore appear to be observing a lower limit to the central surface brightness of gas-rich galaxies in the local universe.

© Copyright Astronomical Society of Australia 1997