Determining the sample

There are many advantages in using HIto determine a sample for the BBD. The most important of these is that it is possible to select a sample that spans the full range of luminosity and surface brightness without the need for large numbers of optical observations. In an ordinary, magnitude-limited, optical sample almost all of the galaxies found will have luminosities near L and surface brightnesses near the peak of the visibility function. This makes the construction of a BBD spanning a large range in both luminosity and surface brightness impossible without making observations of thousands of galaxies. By selecting the sample using HIit is possible to cover the required range of luminosity and surface brightness with a much smaller sample due to the lack of optical selection effects. It is therefore possible to determine the BBD with far fewer optical observations than would be needed otherwise.

As the HImasses of the galaxies are known, it is possible to choose the sample in a way that avoids selecting L galaxies preferentially. It appears likely that HImass works as a tracer of luminosity for spiral galaxies, as larger galaxies will be more luminous and will contain more HIwhile smaller galaxies will be less luminous and will contain less HI. Previous work on optically selected galaxies has found such a relationship, although LSBG's do tend to have more HIthan HSBG's for a given luminosity (de Blok, McGaugh & van der Hulst, 1996) If we assume this to be true, then it is possible to select our sample to have equal numbers of dwarfs and giants by binning the galaxies by HImass and then selecting equal numbers from each bin. This approach means that only a small proportion of the galaxies around , analogous to optical L galaxies, need be observed and should enable us to cover 2 decades in HImass, from 10⁸ to 10¹⁰ solar masses. This range should be approximately equivalent to a magnitude interval from -17 to -22 in B-band.

The selection in surface brightness is also avoided, to a certain degree, by using HI, as there are no explicit surface brightness selection effects. It is known (de Blok & McGaugh, 1996) that LSBG's are generally of lower column density than HSBG's, thus Disney & Banks (1997) relate column density and optical surface brightness using a simple scaling based on the size of the galaxy. Using this scaling it is estimated that we could detect objects in HIPASSdown to a central surface brightness = 25.5 mag arcsec^-2, so we expect to cover a range of around 5 magnitudes in surface brightness.

In order to get sufficient numbers in these bins to expect statistically meaningful results it will be necessary to obtain data for around 400 galaxies - if these were distributed evenly across a 5 x 5 grid there would be 16 in each bin, giving 25% accuracy. From the galaxies found so far in HIPASSit appears that approximately 75% are catalogued. Almost all of these galaxies have luminosities and surface brightnesses in the ESO-LV catalogue (Lauberts & Valentijn, 1988). This leaves around 100 galaxies in our sample which will require optical follow up. This should be possible in a reasonable time-period.

We hope to use an automatic galaxy finder to determine the catalogue our sample will be taken from (see §4). This will enable the sample to be determined objectively and uniformly across different data cubes, which is not possible using the human eye. As full reliability is needed for the optical follow-up involved in this project, spectra selected by the finder will be inspected by eye before the final sample is selected. The primary function of this final screening is to determine which spectra are real galaxies and which are either noise or baseline ripple. While this does introduce some subjectivity into the process, the impact is minimal as the selection of the spectra to be inspected has been made totally objectively and it is normally clear which spectra represent real galaxies.

© Copyright Astronomical Society of Australia 1997