3 HI detection rate

The investigation of the HI line detection rate is of great interest with regard to the HI-completeness of Tully-Fisher catalogues. In particular, this HI-completeness is required to insure against biases by using the inverse relation ($\log V_{\rm m} = aM+b$; see Ekholm & Teerikorpi 1997), or the TF-21 relation ($M_{\rm HI} = a\log V_{\rm m}+b$, where $M_{\rm HI}$ is the HI mass); note that the latter relies exclusively (except for the inclination of the objects) upon parameters derived from the HI profile of a galaxy (Dole & Theureau 1997).

Concerning the detection rate, it is pertinent to examine separately the observations for which the radial velocity is known from the literature, and those which first require a radio velocity search. 1821 galaxies of our programme (hereafter Sample I) had already a published radial velocity, and 1572 of these have been successfully observed at Nançay. This proportion corresponds to a mean detection rate of 86.3% (In fact, almost 90% if confused profile due to close companions are taken into account). In the same way, for the velocity search mode which concerned 890 objects (hereafter Sample II), we got 540 new velocities and HI profiles, corresponding to a mean detection rate of 61%.

However, it is important to make a more detailed study: we could suspect, for example, that very inclined objects (edge-on spirals), which have generally a rather wide profile for a low signal to noise ratio, are less easily detectable than face-on objects. Early type spirals, which contain less neutral hydrogen gas than late type spirals, are also expected to be less detectable. Finally, for galaxies having a known radial velocity, it is interesting to study the detection rate with increasing redshift. Figures 1a (resp. 2a), 1b (resp. 2b), and 1c show the detection rate (broken line) respectively as a function of inclination, morphological type, and radial velocity for Sample I (resp. Sample II). The histogram on the bottom of each panel represents the ratio of the number of objects relative to the total sample size (samples I or II). For this ratio $n/n_{\rm tot}$, in Figs. 1a, 1c, and 2a, the scale of the actual vertical axis must be divided by 10.

  

Figure 1: detection rate as a function of: a) inclination, b) Type, c) radial velocity, for galaxies of known velocity

  

Figure 2: detection rate as a function of: a) inclination, b) Type, for galaxies in velocity search mode

Note the following conclusions: (i) there is no special dependence against inclination, (ii) there is as expected a slight significant increase of the detection rate from early to late types ($\simeq$ 15%, well seen in Fig. 1b), and (iii) no clear decrease can be observed against the distance (at most 5% between v=0 and v=6000 km s^-1). These results are promising for Tully-Fisher applications.