6 Discussion

6.1 Detection rates and implications

The present survey has demonstrated that late-type spiral galaxies selected on the basis of their highly flattened disks and large disk axial ratios are readily detectable in H I21-cm emission within the local universe (see also Giovanelli et al. 1997). Figure 5 illustrates the distribution of radial velocities for the galaxies detected in the present survey. In total, we detected roughly 50% of our targets (232 galaxies) within the search range $V_{\rm h}~\rlap{$<$ }{\lower 1.0ex\hbox{$\sim$ }}\ 10\,000$ km s^-1. Seven of the detections included multiple sources detected toward a single target (see Table 1). Seventy-eight per cent of our detected galaxies had no previously reported detection in H I. We emphasize that our detection rate should be regarded only as a lower limit for the H I detectability of late-type FGC sources, since due to scheduling constraints, the velocity coverage and limiting signal-to-noise were not uniform for all of our targets (cf. Tables 1-3). With full velocity coverage observations of sufficient sensitivity, it is expected that virtually all of the FGC galaxies should be detectable in H I within $V_{\rm h}~\rlap{$<$ }{\lower 1.0ex\hbox{$\sim$ }}\ 20\,000$ km s^-1 (cf. Giovanelli et al. 1997).

Figure 5 demonstrates that the present survey has added a number of gas-rich galaxies to samples both within the Local Supercluster ( $V_{\rm h}~\rlap{$<$ }{\lower 1.0ex\hbox{$\sim$ }}\ 3000$ km s^-1) and beyond it. It is no surprise that late-type, pure disk galaxies are generally gas-rich systems, readily detectable in H I. However, the high detection rate of our survey and of the Giovanelli et al. (1997) FGC survey underscore the important fact that late-type, pure disk galaxies are abundant in the nearby universe, and hence represent one of the most common products of galaxy disk formation (see also van der Kruit 1987). Furthermore, since our current survey has explored only nearly edge-on systems, numerous additional examples of analogous galaxies are certain to exist among samples of recently-catalogued extreme late-type and low surface brightness galaxies seen at lower inclinations (see also Dalcanton & Schectman 1996; Matthews & Gallagher 1997). Even if their overall contribution to the mass and luminosity density of the universe is small, late-type, pure disk galaxies are not negligible in terms of number counts. Any model of galaxy disk formation and evolution must therefore account for the abundance and properties of these small disks, as well as their survival into the present epoch.

Table 4 summarizes the range of Hubble types covered in the present survey, and the detection rates for each type. In contrast to the FGC survey of Giovanelli et al. (1997), our survey concentrated on FGC and related galaxies with Hubble types later than Scd, hence we observed relatively few Sc and earlier spirals. Nonetheless, even given the small numbers of statistics, we note a sharp falloff in the detection rate of Sc targets compared with Scd and later systems. This may be partly due to the fact that most Sc targets in our survey were at low declinations (where sensitivity is reduced due to higher system temperatures), and partly due to the fact that previously undetected Sc spirals are rarer within our search range than new examples of smaller and fainter disk systems.

Figure 5: Histogram showing the number of galaxies detected in the present survey as a function of heliocentric radial velocity, in km s-1

6.2 The nature of the detected objects

A detailed analysis of the properties of the objects detected in our survey is beyond the scope of the present paper. Nonetheless, we briefly remark here on a few trends.

6.2.1 Spectral morphologies

An examination of Fig. 2 reveals that the spectra of the galaxies we have detected most often exhibit the classic double-peaked rotational profiles expected for late-type, rotationally-supported disk galaxies seen near edge-on. Very few of the galaxies in the present survey were resolved significantly by the telescope beam; therefore subject to signal-to-noise limitations, our spectra should be accurate representations of the globally averaged rotation profiles of the H I disks of these galaxies.

Figure 6: An R-band CCD image of FGC 175 obtained with the WIYN telescope. The exposure time was 750 s and seeing was $\sim$ $1\hbox{$.\!\!^{\prime\prime}$ }0$. The image is roughly $1\hbox {$.\mkern -4mu^\prime $ }1$ across; north is on top, west on the left

Figure 7: An R-band CCD image of UGC 825 obtained with the WIYN telescope. The exposure time was 750 s, and seeing was $\sim$ $0\hbox{$.\!\!^{\prime\prime}$ }54$. The image is roughly $1\hbox {$.\mkern -4mu^\prime $ }7$ across; north is on top, west on the left

Figure 8: An R-band CCD image of FGC 2366 obtained with the WIYN telescope. The exposure time was 750 s, and seeing was $\sim$ $1\hbox{$.\!\!^{\prime\prime}$ }3$. The image is roughly $1\hbox {$.\mkern -4mu^\prime $ }7$ across; north is on top, west on the left. Based on the blue-band axial ratios measured by Karachentsev et al. (1993), FGC 2366 is the axial ratio "record-setter'' for the present sample, having a/b=21.6

Since in some cases, optical classifications of edge-on, pure disk spirals can be difficult from images on survey plates alone, the H I profile type can serve as an additional check. In most cases, we find that the H I profile type correlates reasonably well with the optical Hubble classification of the object. For example, the profiles of the Sd and earlier galaxies typically exhibit more well-defined rotation horns than the Sdm and later systems, and the more luminous Sd systems with visible dust lanes tend to have broader rotation profiles than the more diffuse Sd objects. To illustrate this point, in Figs. 6, 7 and 8 we show optical R-band CCD images of 3 galaxies from the present sample. These images were obtained with the WIYN telescope at Kitt Peak. FGC 175 (Fig. 6) is an Sdm galaxy; UGC 825 (Fig. 7) is a fairly bright Sd with a prominent dust lane; and FGC 2366 (Fig. 8) is a moderate surface brightness Sd with no obvious dust lane, only a modest central light concentration, and with the highest catalogued axial ratio of all the galaxies in the present survey (a/b=21.6). The optical morphologies of these galaxies may be compared with the corresponding global H I profiles in Fig. 2.

Figure 9: Histogram showing the distribution of linewidths (in km s-1) for the galaxies detected in the present survey. The linewidths shown were measured at 20% peak maximum and corrected for spectral resolution and cosmological effects, as described in Sect. 4.1

Figure 10: The distribution of H I linewidths (in km s-1) from the present survey as a function of apparent axial ratio. The linewidths were measured at 20% peak maximum and corrected for spectral resolution and cosmological effects, as described in Sect. 4.1

Figure 11: The same as in Fig. 10, but with data from Giovanelli et al. (1997) overplotted as diamonds, and with five additional superthin galaxies from the literature overplotted as filled circles (see Text). In order to be commensurate with the range of linewidths covered in the present survey, only datapoints with W20<500 km s-1 are shown from the Giovanelli et al. sample

In spite of some general trends, within the broad categories of H I profile types we nonetheless still see a fair amount of diversity even for a give Hubble type (see Fig. 2). This suggests that the H I properties of late-type, pure disk systems do show variations, hence one needs to be cautious in too widely generalizing their properties before more detailed investigations of a significant number of the individual objects have been undertaken.

6.2.2 Rotational velocities

Figure 9 summarizes the distribution of corrected rotational widths (full width at 20% peak maximum) of the H I profiles of our detected galaxies. These data are taken from Col. 10 of Table 1. Since all the galaxies are presumed to be nearly edge-on, no corrections for inclination have been attempted.

The observed range of linewidths is as expected for the Hubble types covered in the present survey, and we see a strong peak in the range $W_{\rm 20,c}=200-225$ km s^-1. For the Sd component of our sample, the mean corrected W₂₀ is 244 km s^-1, and the median is 234 km s^-1; for the Sdms, the mean is 197 km s^-1 and the median is 202 km s^-1. Using the Nearby Galaxies Catalog (TNGC) of Tully (1988), we compare these averages to other nearby, late-type spirals. For the 75 Sd (T=7) galaxies in the TNGC with $i>40^{\circ}$, we find a mean inclination-corrected W₂₀ value of 235 km s^-1 and a median of 238 km s^-1. For the 79 Sdm (T=9) galaxies in the TNGC with $i>40^{\circ}$, we find a mean W_20,i of 221 km s^-1 and a median of 213 km s^-1. We thus find no significant offsets if the rotational velocities of our sample of highly flattened Sd systems compared with other Sds. There is some hint that the Sdms in our present sample rotate slightly more slowly on average than the TNGC sample; this slight difference may be partly a consequence of our new sample containing larger numbers of less luminous Sdm systems than the TNGC.

6.2.3 Disk axial ratios versus rotational velocity

Figure 10 shows the $W_{\rm 20,c}$ values for the survey objects plotted as a function of apparent axial ratio. In Fig. 11 we show the same data, but this time we also include galaxies from the Giovanelli et al. (1997) FGC survey with W₂₀<500 km s^-1. The Giovanelli et al. sample contains a number of galaxies with $a/b\le 7$ due to the fact that these authors included many of the FGCA galaxies in their survey. Galaxies for which we detected 2 emission profiles toward a single target were excluded from Figs. 10 and 11, as were cases where our line profile was detected near the edge of the bandpass. Using filled circles, we have also overplotted in Fig. 11 five well-known superthins from the literature, all with a/b>11: UGC 711, UGC 7170, UGC 7321, UGC 9242, and ESO 146-014 (see Rönnback & Bergvall 1995; Cox et al. 1996; Matthews et al. 1999, 2000a).

Figures 10-11 demonstrate that a wide spread is seen among the $W_{\rm 20,c}$ values for the less highly flattened galaxies (a/b<10) in the samples illustrated. This is expected, since pure disk systems with a/b<10 will include a mixture of both "superthin'' galaxies viewed several degrees away from edge-on, as well as edge-on, intrinsically thicker systems.

Of perhaps greater interest are the "extreme'' superthin galaxies in the sample (i.e., those with $a/b\ge 15$). We note that no such galaxies were found with W₂₀ <$\sim$ 190 km s^-1, suggesting that perhaps there exists some minimum rotational velocity (or equivalently, mass) below which such systems cannot exist (see also Karachentsev 1999). Moreover, there appears to be a steep rise in the maximum permitted axial ratio for disks between $W_{20}\approx$ 100 km s^-1 and $W_{20}\approx$ 200 km s^-1.

In the combined sample of Figs. 11, 7 of the 27 galaxies with $a/b\ge 15$ fall in the narrow interval $190\le W_{20}\le 210$ km s^-1, including the most extreme object in the sample, FGC 2366, with a/b=21.6 (see Fig. 8). Figures 10 and 11 also show a high density of objects with $a/b\le 10$ in the interval and $210\le W_{\rm 20,c}\le 280$ km s^-1, while for a/b>15, this range of rotational velocities is unpopulated. Although we still have only a small number of statistics for galaxies with $a/b\ge15,$ the combination of these trends raises the interesting possibility that there may be certain mass ranges over which the most extreme superthin disks are most likely to form, or to retain their svelte appearances (see also Matthews et al. 2000a).

6.3 Environments of pure disk galaxies

It has been proposed that many flattened, pure disks galaxies, particularly those "superthin'' objects with a/b>10, must necessarily have remained largely unperturbed in order to preserve their thin, bulgeless, dynamically cold stellar disks (e.g., Reshetnikov & Combes 1997; Matthews 1998). The redshifts obtained in the present survey (together with those of Giovanelli et al. 1997), can be used to test this proposition by exploring the environments of these galaxies for the first time in three dimensions. A detailed examination of this problem is deferred to a later paper, but here we remark on a few trends.

In our present survey we find that among the 232 targets detected with a moderately high level of certainty (i.e. those listed in Table 1), incidences of true blended or confused H I profiles are relatively rare, in spite of the rather large beam sizes of the Nançay and Green Bank 140-ft telescopes ( $4'\times\ge 22'$and 21' FWHP, respectively). In only 10 cases were we able to identify a neighbor within 1.5 beam diameters of the target and having a similar redshift ( $\Delta V_{\rm h}\le400$ km s^-1). These cases are described individually in more detail in Appendix A. In six of these instances, the target object is not a superthin, i.e., a/b<10. The four exceptions to this are FGC 1845 (a/b=13.3), UGC 11243 (a/b=11.2), FGC 2264 (a/b=14.3), and ESO 467-063 (a/b=12.4). Although the thinness of the disks of these objects in the presence of a neighbor within a projected distance of $\lesssim 0.28$ Mpc is surprising, we do find that 3 of the 4 objects show signs of optical peculiarities, most likely due to the perturbation of the companion. ESO 467-063 was noted by Karachentsev et al. (1993) to have "curved ends'', and on the DSS, both FGC 2264 and UGC 11243 can also be seen to exhibit curvature. Only FGC 1845 seems to show no obvious optical peculiarities on the DSS or in optical CCD images obtained by Matthews et al. (unpublished). Overall we find the results of our survey to be consistent with the notion that highly flattened pure disk galaxies tend to be relatively isolated objects, and that the presence of a close neighbor tends to thicken their disks and/or alter their optical morphologies.