Radio-excess infrared galaxies: a new population of active galaxies?

A. L. Roy1,2,3 and R. P. Norris2

1 School of Physics, University of Sydney 2006, Australia

2 Australia Telescope National Facility, CSIRO, PO Box 76, Epping, NSW 2121, Australia

3 Raman Research Institute, Bangalore 560 080, India

ABSTRACT

We present a sample of active galaxy candidates, found by comparing the IRAS catalogue to the Parkes-MIT-NRAO (PMN) southern 6.3-cm radio survey. We have selected galaxies that depart from the radio versus far-infrared (FIR) relation displayed by normal galaxies, to produce a sample of 12 candidate radio-excess, gas-rich, active galaxies. Such objects have been discussed recently with the aim of understanding how they relate to more familiar types of active galactic nuclei (AGNs). Do they represent a different evolutionary stage of radio-quiet AGNs, or are they a familiar type of AGN but in an unusual environment, perhaps dust-enshrouded quasars or their progenitors?

The pilot sample presented here is intended to find a usefully large number of radio-excess AGNs suitable for further study, and to test the efficiency of the selection criteria. We give radio and FIR data for the sample, along with optical identifications, spectroscopic classifications, and redshifts from the literature. These data reveal some unusual and interesting objects, despite the relatively small area covered by this inital survey.

Key words: galaxies: active - galaxies: Seyfert - galaxies: starburst - radio continuum: galaxies - infrared: galaxies - BL Lacertae objects: general

1 INTRODUCTION

The correlation between non-thermal radio and thermal FIR emission from spiral galaxies is so tight and is displayed by a such a wide range of galaxy types (e.g., Wunderlich, Klein & Wielebinski 1987) that it is interesting when a population of galaxies shows a significant deviation from the normal correlation.

Many Seyfert galaxies, like NGC 1068, show large departures from the normal radio-FIR correlation (e.g. Sanders & Mirabel 1985; Wilson 1988; Baum et al. 1993), and such Seyferts tend to be more radio-loud than normal spirals with the same FIR luminosity. We call these "radio-excess", hereafter, to distinguish them from the traditional radio-loud galaxies. The excess radio emission was shown by Sanders & Mirabel, and others, to originate from within the kpc-scale nucleus of the galaxies, since the galaxies returned to the normal radio-FIR correlation after the nuclear radio emission was subtracted off. A. L. Roy et al. (in preparation) show that the normal radio-FIR correlation is degraded in Seyfert galaxies by excess radio emission on scales larger than 20 to 200 pc but less than about one kpc, which provides an interesting constraint on the geometry of the radio-emitting region. It still remains to be understood why a few AGNs produce prodigious radio luminosities whilst most are radio-quiet.

One of the most extreme examples of a radio-excess Seyfert galaxy is the extremely luminous far-infrared galaxy IRAS 00182-7112. It is a type-2 Seyfert (Norris & Allen, unpublished data; and Armus, Heckman & Miley 1989), which produces LFIR = 1012.5 L and L4.8 GHz = 1025.4 W Hz-1 (Roy et al., unpublished data), placing it among the most luminous FIR galaxies. It is significantly more radio-luminous than expected from its FIR emission. Despite its similarity in radio luminosity to powerful radio galaxies, the object does not produce large-scale radio lobes, and is unresolved on arcsec (5 kpc) scales. However, it shows structure on VLBI scales (R. P. Norris et al., in preparation), within the kpc nuclear region. This object may shed light on the cause of the dichotomy between radio-loud and radio-quiet AGNs.

Similar radio-excess IRAS galaxies are the subject of a programme of study by Dey & van Breugel (1994). To select such galaxies they compared the IRAS Faint Source Survey (Moshir et al. 1992) to the Texas 365­MHz radio catalogue (Douglas et al. 1980) and selected radio-excess objects. On the basis of radio and optical imaging and spectroscopy and from the FIR colours, they propose that these objects may form an evolutionary link between ultraluminous infrared galaxies, like Arp 220, and quasars. An evolutionary connection between ultraluminous infrared galaxies and quasars was suggested by Sanders et al. (1988). According to this scheme, the ultraluminous infrared-emitting phase is triggered by collisions between two gas-rich galaxies and produces a dusty nuclear starburst and sometimes also an obscured quasar core. The starburst is thought to blow clear the obscuring gas and dust and, in time, will reveal the quasar core to our optical view. At that stage, powerful radio emission might also be produced by a small fraction of such objects. Dey & van Breugel found thattheir radio-excess IRAS galaxies were intermediate, between ultraluminous starbursts and powerful quasars, in radio luminosities, far-infrared colour, and optical spectral class. The galaxies often also produce FIR thermal excesses and strong Balmer absorption which suggest the presence of large quantities of gas. They suggest, therefore, that such galaxies are at intermediate stages along an evolutionary track between ultraluminous infrared galaxies and quasars.

Here, we present a systematically selected sample suitable for studying radio-excess AGNs with gas-rich hosts. We have selected the sample using the radio and FIR properties rather than optical spectroscopy, given that we know relatively little about the optical properties of these AGNs. Large FIR luminosities indicate the presence of large dust masses and hence tend to select for gas-rich systems (e.g. Soifer et al. 1987b). We used the 4.9-GHz PMN southern radio survey (Griffith & Wright 1993) and the IRAS Point Source Catalog (1988), and have selected those galaxies that depart from the normal radio-FIR correlation. We used only the southern part of the PMN survey, as it was the first part to become available and it was large enough to provide a useful pilot sample. The approach adopted independently by Dey & van Breugel (1994) was similar, but used the much lower frequency of 365 MHz. Their sample therefore favours objects with steeper radio spectra than the present sample. From our sample of 20 objects, we concentrate on a sub-sample of 12 for which more accurate radio positions are available. These improved positions allow us to find reliable associations between radio and FIR sources. We present optical identifications from the Sky Survey plates for the whole sample, along with identifications from the literature for some of the galaxies. These have proven the method to be an efficient way to find active galaxies.

The selection criteria and the resulting sample are detailed in Section 2. Section 3 estimates the reliability and completeness of the survey, and Section 4 discusses the nature of the radio-excess objects found in this survey, and identifies future follow-up work.

We adopt H0 = 75 km s-1 Mpc-1 and q0 = 0.5 throughout this paper.

2 SAMPLE SELECTION

The recent release of the PMN radio survey (Wright et al. 1994) makes available a uniform, sensitive, large-area radio survey which complements the IRAS all-sky survey. We have used these two samples together to find active galaxies, by using the tightness of the radio-FIR correlation displayed by normal galaxies to distinguish the active from the normal galaxies.

We used the southern region of the PMN survey, which covers 20 per cent of the sky at a frequency of 4.9 GHz to a flux-density limit of ~ 35 mJy. FIR data were drawn from the IRAS Point Source Catalog Version 2, and we selected objects which satisfied the following criteria. They were:

i) detected at 60 m,

ii) not in a stellar catalogue, i.e. IRAS "idtype" 2,

iii) not identified in the IRAS catalogue with objects from catalogues numbered 1, 2, 4, 5, 7, 11, 13, 15, 16, 17, 18, 19, 22, 23, or 24, in order to reject stars, globules, planetary
nebulae, H II regions, dark nebulae, reflection nebulae, and bright diffuse Galactic
nebulae,

iv) S60 m / S25 m 1.5 when detected at 25 m, in order to reject stars,

v) not in the Magellanic Clouds, i.e., not in the (B1950) regions:

0h 36m < RA < 1h 29m, -74 < dec < -72 (SMC),

4h 40m < RA < 5h 55m, -72 < dec < -65 (LMC),

vi) at high galactic latitude, |b| 7,

vii) detected by the PMN survey as stronger than 70 mJy at 6.3 cm (i.e. 7 to 11 ), within a
2­arcmin square (i.e., 2 ) centred on the IRAS position,

viii) (S6.3 cm / Jy) / (S60 m / Jy) > 1/55, or (S6.3 cm / Jy) / (S60 m / Jy) < 1/875, to retain those objects that are more than 3 above or below the radio-FIR correlation defined by normal galaxies (de Jong et al. 1985). (No objects lay more than 3 below the
correlation.)

We found 20 objects that satisfied these criteria, out of 6094 IRAS sources and 6650 PMN sources that lay in the above regions and satisfied the FIR and the radio criteria separately. The 20 objects that satisfied both the radio and FIR criteria are listed in Table 1. None of these is flagged in the PMN catalogue as showing signs of extension in the 4.2-arcmin beam.

Due to the relatively large position uncertainty, these associations are not very reliable (Section 3.1). To provide more reliable associations, we used more accurate radio positions that were measured as part of a programme of high-resolution observations of the brighter PMN sources (N. J. Tasker, D. McConnell & A. E. Wright, private communication), using the Australia Telescope Compact Array (ATCA). This survey measured positions with
~ 5-arcsec rms uncertainty for ~ 9000 sources stronger than 70 mJy that lay in the southern region of the PMN survey. Offsets between these ATCA positions and the IRAS positions are shown in Table 1 and are generally less than about 20 arcsec. ATCA positions were not available for some PMN sources due to source structure or confusion. Twelve associations in Table 1 have IRAS-radio position offsets 20 arcsec. These associations are likely to be real as we expect only one to be spurious (see Section 3.1). These 12 reliable associations make up the sample on which our discussion concentrates.

We show the radio-FIR scatter diagram in Fig. 1 for the sample after stage vi, above, before removing any sources on the basis of the radio flux density (criterion vii), or on the radio / FIR flux­density ratio (criterion viii). The AGN candidates lie scattered in the lower left of the diagram above the radio-FIR correlation of normal galaxies. The other objects, which populate the lower central region of the diagram, display a correlation between their radio and FIR flux densities, and they lie between the two lines that mark the ±3- extent of the radio-FIR correlation displayed by normal galaxies (de Jong et al. 1985). Many of the objects that display the radio-FIR correlation are classified morphologically as spirals (NED), and a subset also host Seyfert activity, consistent with the finding by Sopp & Alexander (1991). The sources that lie above the upper line are our AGN candidates, and are listed in Table 1.

3 IDENTIFICATION PROGRAM

3.1 Reliability

The reliability of this sample (i.e. the probability that an association listed in Table 1 is real) depends on the reliability of the IRAS and PMN surveys and on the chance of spurious associations between them. The reliability of the individual surveys is high when the signal-to-noise ratio is good, and begins to degrade appreciably only when approaching the flux-density limits of the surveys. The 60-m channel of the IRAS catalogue is estimated to be at least 99.9 per cent reliable for objects with at least two sightings that were confirmed within hours (i.e. for most objects), and the PMN survey is at least 90 per cent reliable at the survey limit.

To estimate the chance of spurious associations we re-ran the sample selection code in the same manner as when generating the sub-sample of 12 objects in Table 1, but with various small offsets in declination between the catalogues (increments of 15 arcmin were used to destroy any real associations). After 10 trials, this turned up a median of one association within 20 arcsec that satisfied our selection criteria, and so we adopt this as the most likely number of spurious sources. (The ten trials produced a range of zero to one such associations.) Thus, we expect ~ one of the 12 objects in the main radio-excess sample to be a spurious association.

A similar estimate for the whole sample of 20 objects yielded a median of six spurious associations, with a range of 4 to 11. This uses an acceptance area for an association of
2-arcmin square, appropriate for the less accurate PMN positions. Due to this high rate of spurious associations, we concentrate on the sub-sample of 12 objects for which ATCA positions are available and yield close associations.

The differences between the ATCA and IRAS positions for sources in the radio-excess sample are shown in Fig. 2. The differences are approximately Gaussian distributed with a FWHM of 18 arcsec. The position differences are distributed as expected from the combined formal uncertainty of the two surveys. Positions from the IRAS catalogue have quoted uncertainties of up to 14 arcsec (1 ), and positions from the ATCA survey have 1- uncertainties of ~ 5 arcsec for a 70-mJy source. Thus, on the basis of the position differences, we believe that the radio and FIR sources are generally associated.

3.2 Data from the literature

We have searched the literature for data on the 20 candidate radio-excess objects in Table 1, and found that all 13 objects with morphological or spectral classifications were galaxies or AGNs. The remaining seven are likely to have similar classifications. We searched for catalogue names, redshifts, and spectral classifications in NED, de Grijp et al. (1992), Allen et al. (1991), Spinoglio & Malkan (1989), Véron-Cetty & Véron (1991), Edelson (1987), Jones & McAdam (1992), Neugebauer et al. (1986), Hutchings & Neff (1991), and Jaffe & Perola (1976). We required that the positions agree with the IRAS coordinates within one arcmin before we associated the data with the object. In most cases the positions agreed to within 10 arcsec, and the remainder agreed within 46 arcsec. In the one case where the position offset approached 46 arcsec it is distinctly possible that the object was misidentified. The position offsets can be seen in Table 1 from the column giving the distance between IRAS and optical positions.

Only one of the candidate radio-excess AGNs (IRAS 05373-4406) had more than one catalogued object within one arcmin of the IRAS coordinates. In addition to the identification that we adopted in Table 1, there was also an 18th-magnitude E2 galaxy, catalogued as PKS 0537­441G1, which lay 20 arcsec distant from the IRAS position. We have adopted the galaxy PKS 0537­441 as the identification, since it has a smaller separation from the IRAS position, and is brighter optically than PKS 0537-441G1.

3.3 Morphological information from the Sky Survey

We examined the SRC J sky survey plates for optical counterparts to the candidate AGNs in Table 1. Morphologies were estimated from polaroid copies of the sky-survey plates, using the following criteria.

i) Galaxies that were visibly extended, being surrounded by a diffuse structureless halo with an axial ratio 2, were classified as elliptical (E).

ii) Galaxies that satisfied the definition of an elliptical, except that the axial ratio was greater than 2, were classified as S0.

iii) A galaxy with considerable structure in the halo was classified as a spiral (Sp).

iv) An image that was diffuse but too faint to assign to the above classes was called "G".

v) An image with a sharp circular boundary was classified as stellar (St).

The resulting morphological classifications are given in Table 1.

4 DISCUSSION

4.1 Morphology

The character of the radio-excess galaxy sample was revealed, in part, by the optical data from the Sky Survey and from the literature. We have morphological information for the closer radio-excess objects. Of the 12 objects in our main sample, we found that two are S0, two are elliptical galaxies, one is a strongly interacting pair, and one appears stellar and is classified spectroscopically as a BL Lac object. Among the remainder of the 20 galaxies there are five spirals, one S0, and two elliptical galaxies. The remaining six objects are too distant for their morphologies to be clear. The presence of spirals is expected for an FIR-selected sample, and the S0s and ellipticals are probably favoured by the radio-excess selection criteria. Four of the 12 main objects show signs of morphological peculiarities, like distorted arms or rings, or tidal tails. Such peculiarities occur with a frequency that is indistinguishable from the that of the control sample of field galaxies of Lawrence et al. (1989). However, one must bear in mind that the visibility of such features depends on the depth of the images, and that this control sample was observed using CCD images that were deeper than the Sky Survey plates.

The radio-excess objects range in mV from ~ 11 to 21, and have optical major diameters from 5 arcmin to several arcsec. We found redshifts for six of the 12 main radio-excess objects and they lay between 0.005 and 0.896 with a median redshift of 0.105. This range is typical for FIR-selected galaxy samples (e.g. Soifer, Houck & Neugebauer 1987a).

4.2 Spectroscopy

All six of the main radio-excess objects with spectral classifications from the literature were classed as AGNs of various types. There were four type-2 Seyferts, one type-1, and one BL Lac object. Therefore, the success rate for finding AGNs using the radio-excess sample was at least 50 per cent, and so the selection procedure is an efficient method for identifying AGN candidates. We will now comment on each class of object found in the survey.

4.2.1 Seyfert galaxies

The five Seyfert galaxies had radio luminosities between 1022.1 and 1025.4 W Hz-1 at 4.9 GHz, which is two orders of magnitude more luminous, on average, than the Seyferts studied by Ulvestad & Wilson (1989). The larger luminosity is a natural result of our selection criteria, since galaxies less powerful than ~ 1023.7 W Hz-1 at 4.9 GHz at the median redshift of 0.0615 would not be detected above our flux-density limit of 70 mJy. Two of the Seyferts (IRAS 00182-7112 and PKS J1557-7913) were also radio loud in the traditional sense (i.e. they have a large radio / optical ratio or, more specifically, they have log10(S6 cm / SB-band) > 1.3 to 1.8, following Visnovsky (1992) and Sramek & Weedman 1980). This is rare among Seyfert galaxies. There were three other radio-loud objects, but they were of types that frequently display powerful radio emission. The FIR luminosities of the 12 main radio-excess objects imply gas masses of 108-9 M, following Hildebrand (1983) and assuming the FIR emission is thermal reradiation from dust with a gas to dust mass ratio of 200. Thus our selection strategy is efficient at finding interesting gas-rich galaxies with large radio luminosity.

Most of the Seyfert galaxies were type 2s (4/5). It is well known that FIR-selected samples contain more Sy2s than Sy1s (e.g. de Grijp et al. 1992), and our sample is consistent with their result, within the limits imposed by small-number statistics.

4.2.2 The BL Lac object

The presence of the BL Lac object, PKS 0537-441, in a sample as small as six classified objects, is remarkable given the extreme scarcity of BL Lacs, and it shows the power of the selection strategy. Even though the BL Lac object met the FIR selection criteria, it probably resides in an elliptical or S0 host like all other BL Lac objects (e.g. the review by Browne & Marchã 1993). Elliptical and S0 galaxies are rarely detected by IRAS (Soifer et al. 1987a), and so it is most likely that the FIR emission is generated by the AGN. This object is a strong GRO -ray source and has a strong VLBI source, which suggest that Doppler boosting is important. The emission is therefore probably synchrotron rather than thermal dust emission.

Dey & van Breugel (1994) found that BL Lac objects comprised ~ 18 per cent of their sample of ~ 100 radio-excess objects. They selected their sample in a similar way to that used here, except that the radio survey was made at a much lower frequency (365 MHz). The proportion of BL Lac objects is therefore approximately the same whether selected using the 365­MHz or the 4.9-GHz radio surveys, subject to small-number statistics. BL Lac objects are mostly flat-spectrum (e.g. Kühr & Schmidt 1990), unlike many other radio-excess objects which are very much weaker at 4.9 GHz than at 365 MHz, and so we should expect a larger fraction of BL Lacs than that found by Dey & van Breugel.

4.2.3 The radio galaxy

The radio-excess object PKS 0410-756 is identified as a powerful radio galaxy by Alvarez et al. (1993). However, they convincingly identify the radio and FIR sources with different objects, and so it seems that this object is one of the spurious associations that we expect may exist in this sample.

4.3 Unifying radio-excess Seyferts

These radio-excess Seyferts present a challenge to the existing unification schemes, which are reviewed, for example, by Lawrence (1987). The popular orientation unification scheme for Seyfert galaxies explains the spectroscopic differences between the two Seyfert types, but it does not attempt to explain the radio properties. In particular, it does not address why some Seyfert nuclei are much more radio-loud than others.

The dichotomy between radio-loud and radio-quiet AGNs has been addressed by a number of models but it still is not understood why some AGNs eject powerful jets of relativistic plasma and others do not. Scheuer & Readhead (1979) proposed that all quasars have relativistic jets, and these appear powerful only on the occasions when one points towards us. However, this model does not explain the different host galaxy types of radio-loud and radio-quiet AGNs; radio-louds live in elliptical galaxies and radio-quiets tend to live in spirals (e.g. Véron-Cetty & Woltjer 1990). Also, among the radio-quiet AGNs the most luminous live in S0 galaxies, and so it seems to be the mass of the spheroidal component that matters for the AGN phenomenon (Lawrence 1987). The same process of accretion is likely to drive AGNs in both ellipticals and spirals but, for some reason, spiral nuclei do not generate radio jets. It may be that jets are produced within spirals but are smothered by the dense interstellar medium and do not emerge. Certainly, evolution is important, and AGN radio jets probably switch off and on periodically. In his review, Lawrence (1987) suggests that the duty cycle might somehow be influenced by the mass of the spheroidal component. Recently, Wilson & Colbert (1995) suggested that the angular momentum of the black hole determines the radio luminosity. Galaxy mergers can spin up black holes and also produce elliptical host galaxies. Wilson & Colbert suggest that large angular momentum of the resulting black hole powers the jets and that these systems become the radio-loud AGNs.

Among the radio-excess AGNs that our selection procedure will select may be further examples of the remarkable double-lobed Seyfert galaxy found by Beichman et al. (1985). In this object, 0421+040P06, radio jets have formed and produced a classic, albeit small, radio-loud AGN morphology. To unify radio-loud and radio-quiet AGNs we need to understand what stops jet formation in spirals. Further examples like that of Beichman et al. would provide unique material with which to work, and the approach used here may provide an efficient way to find such objects.

Follow-up imaging will be required to show clearly whether radio-excess IRAS galaxies have spiral host galaxies. Evidence for spiral hosts is indirect so far, coming from the FIR emission, the spectroscopic Seyfert classification, and from images from the Sky Survey plates. Obtaining good-quality imaging is a high priority for follow-up work and should reveal a key part of nature of radio-excess IRAS galaxies.

4.4 Future prospects

Extrapolating from this pilot sample, we estimate that there may be as many as ~ 130 candidate radio-excess AGNs in the whole sky, going down to the flux-density limit of the PMN survey. The number of spurious associations could be as low as ~ 17 for an all­sky survey when 5-arcsec positions become available in the northern sky, matching the positional accuracy of the ATCA follow-up to the PMN survey. The resulting sample of 130 radio-excess AGN candidates may be sufficient to form the basis for powerful statistical studies.

We plan to make follow-up optical and radio observations to compare the radio-excess AGNs to ultraluminous far-infrared galaxies and quasars. Optical spectroscopy and broadband imaging should reveal greater detail of the host-galaxy type and the presence of starbursts and active nuclei. Medium- and high-resolution radio observations can reveal possible jets and large-scale lobes analogous to those in powerful radio galaxies. Such properties might be expected to be intermediate between those of ultraluminous far-infrared galaxies and quasars if the radio-excess AGNs do indeed fill an evolutionary link between such objects.

5 CONCLUSION

Active galactic nuclei that reside in spiral host galaxies are usually radio-quiet. However, a few gas-rich hosts harbour nuclei that are much more luminous in the radio than expected from their FIR emission. They comprise an interesting and important group of AGNs that does not fit easily into existing AGN unification schemes. We seek to understand how they relate to other types of AGNs; are they a short-lived stage in the life of radio-quiet AGNs, or are they perhaps dust-enshrouded quasars?

As the first stage of a programme to study these objects, we have selected from the PMN and IRAS surveys a pilot-sample of 12 AGN candidates, some of which are likely to have gas-rich hosts, and which produce much greater radio luminosity than do normal spirals with the same FIR luminosity. To do this, we have evolved a set of selection criteria, and we demonstrate here that they are efficient at selecting radio-excess AGNs. We have thus established the viability of using these criteria to select a usefully large sample for further study.

ACKNOWLEDGMENTS

We thank Alan Wright and his team of collaborators for the countless hours they have spent, both at the telescope, and during the preceding preparations and the following months of data reduction, to produce the PMN Radio Survey. Without their Herculean effort, this work would not have been possible. Niven Tasker kindly provided us with more accurate positions from their Australia Telescope survey, before publication. This survey was conducted by himself, Dave McConnell and Alan Wright, as a follow-up to the PMN survey.

This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, Caltech, under contract with the National Aeronautics and Space Administration.

Table 1. The candidate active galaxies

RA (B1950) dec separations S60 m S6.3 cm LFIR L6.3 cm morphological morphological spectral

IRAS name alias h m s ° ' " " " " / Jy / Jy q / L / W Hz-1 type type ref. class ref. redshift ref.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)

Notes to Table 1:

Col 1: IRAS name,

Col 2: alias (see § 3.2),

Col 3: IRAS position,

Col 4: separation in arcsec between IRAS and PMN coordinates,

Col 5: separation in arcsec between IRAS and AT coordinates,

Col 6: separation in arcsec between IRAS and coordinates of optical identifications
from the NASA Extragalactic Database (NED),

Col 7: 60-m flux density in Jy from the IRAS Point Source Catalog, Version 2

Col 8: 6.3-cm flux density in Jy from PMN point source catalogue,

Col 9: radio / FIR ratio, calculated following Helou et al. (1985), i.e., q = log10 (1.26x10-14 Hz (2.58 S60 m /Jy + S100 m /Jy) / 3.75x1012 Hz / (S6.3 cm / W m-2 Hz-1 )),

Col 10: log10 of FIR luminosity in solar luminosities, with FIR flux calculated
following Helou et al. (1985),

Col 11: log10 of 6.3-cm spectral power density in W Hz-1,

Col 12: morphological type, from Sky Survey plates (see § 3.3),

Col 13: morphological type, from the literature,

Col 14: reference for morphological type,

Col 15: spectral classification, from the literature,

Col 16: reference for spectral classification,

Col 17: redshift, from the literature,

Col 18: reference for redshift.

References: A = Allen et al. (1991),

B = Alvarez et al. (1993),

C = R. P. Norris & D. A. Allen (1993, unpublished data),

D = de Grijp et al. (1992),

E = Corwin, de Vaucouleurs & de Vaucouleurs (1985)

L = Lauberts (1982)

N = NED,

V = Véron-Cetty & Véron (1991).

* The association of IRAS 04099-7514 with PKS 0410-756 is probably spurious (§ 4).

† These associations have a high probability of being real (see § 3.1).

REFERENCES

Allen D. A., Norris R. P., Meadows V. S., Roche P. F., 1991, MNRAS, 248, 528

Alvarez H., Aparici J., May J., Navarrete M., 1993, A&A, 271, 435

Armus, L., Heckman, T. M., Miley, G. K., 1989, ApJ, 347, 727

Baum S. A., O'Dea C. P., Dallacassa D., de Bruyn A. G., Pedlar A., 1993, ApJ, in press

Beichman, C., Lonsdale, C. J., Wynn-Williams, C. G., Heasley, J. N., Becklin, E. E., Persson, S. E., Miley, G. K., Soifer, B. T., Neugebauer, G., Houck, J. R., 1985, ApJ, 293, 148.

Browne I. W. A., Marchã M. J. M., 1993, MNRAS, 261, 795

Corwin, H. G. Jr., de Vaucouleurs, A., de Vaucouleurs, G., 1985, Southern Galaxy Catalogue
(Austin, TX: University of Texas)

de Grijp M. H. K., Keel W. C., Miley G. K., Goudfrooij P., Lub J., 1992, A&AS, 96, 389

de Jong T., Klein U., Wielebinski R., Wunderlich E., 1985, A&A, 147, L6

Dey A., van Breugel W., 1994, in Mass-Transfer Induced Activity in Galaxies. ed. I Shlosman, CUP, Cambridge.

Douglas, J. N., Bash, F. N., Torrence, G. W., Wolfe, C., 1980, Univ. Texas Publ. Astron. #17

Edelson R. A., 1987, ApJ, 313, 651

Griffith M. R., Wright A. E., 1993, AJ, 105, 1666

Hildebrand R. H., 1983, QJRAS, 24, 267

Helou G., Soifer B. T., Rowan-Robinson M., 1985, ApJ, 298, L7

Hutchings J. B., Neff S. G., 1991, AJ, 101, 434

IRAS Point Source Catalog, Version 2, 1988. Joint IRAS Science Working Group, US Government Printing Office; Washington, DC.

Jones P. A., McAdam W. B., 1992, ApJS, 80, 137

Jaffe W. J., Perola G. C., 1976, A&A, 46, 275

Kühr H., Schmidt G. D., 1990, AJ, 99, 1

Lauberts, A., 1982 The ESO/Uppsala Survey of the ESO (B) Atlas (European Southern Observatory)

Lawrence A., 1987, PASP, 99, 309

Lawrence, A., Rowan-Robinson, M., Leech, K., Jones, D. H. P., Wall, J. V., 1989, MNRAS, 240, 329

Moshir, M., et al., 1992, Explanatory Supplement to the IRAS Faint Source Survey, Version 2, JPL D-10015 8/92 (Pasadena: JPL)

Neugebauer G., Miley G. K., Soifer B. T., Clegg P. E., 1986, ApJ, 308, 815

Sanders D. B., Mirabel I. F., 1985, ApJ, 298, L31

Sanders D. B., Soifer B. T., Elias J. H., Madore B. F., Matthews K., Neugebauer G., Scoville N. Z., 1988, ApJ, 325, 74

Scheuer, P. A. G., Readhead, A. C. S., 1979, Nature, 277, 182.

Soifer B. T., Houck J. R., Neugebauer G., 1987a, ARA&A, 25, 187

Soifer B. T., Sanders D. B., Madore B. F., Neugebauer G., Danielson G. E., Elias J. H., Lonsdale C. J., Rice W. L., 1987b, ApJ, 320, 238

Sopp H. M., Alexander P., 1991, MNRAS, 251, 14p

Spinoglio L., Malkan M. A., 1989, ApJ, 342, 83

Sramek R. A., Weedman D. W., 1980, ApJ, 238, 435

Staveley-Smith L., Norris R. P., Chapman J. M., Allen D. A., Whiteoak J. B., Roy A. L., 1992, MNRAS, 258, 725

Ulvestad J. S., Wilson A. S., 1989, ApJ, 343, 659

Véron-Cetty M. P., Véron P., 1991, A Catalogue of Quasars and Active Nuclei. 5th ed. European Southern Observatory, Garching bei Munchen, Germany.

Véron-Cetty, M. P., Woltjer, L., 1990, A&A, 236, 69.

Visnovsky, K. L., Impey, C. D., Foltz, C. B., Hewettt, P. C., Weymann, R. J., Morris, S. L., 1992, ApJ, 391, 560.

Wilson A. S., 1988, A&A, 206, 41

Wilson, A. S., Colbert, E. J. M., 1995, ApJ, 438, 62

Wright A. E., Griffith M. R., Burke B. F., Ekers R. D., 1994, ApJS, in press

Wunderlich E., Klein U., Wielebinski R., 1987, A&AS, 69, 487

FIGURES

Figure 1. 6.3-cm radio v. 60-m FIR flux-density scatter diagram showing 84 sources which satisfied the FIR selection criteria for the radio-excess sample. A correlation can be seen which involves normal spirals, lying between the two lines that indicate the ±3- extent of the radio-FIR correlation found by de Jong et al. (1985). A few galaxies in the lower left show significant scatter, tending to be more radio-loud than normal galaxies for the same FIR flux density, and the objects that lie above the +3- line and have 6.3-cm flux densities greater than 70 mJy are included in the radio-excess sample.

Figure 2. Scatter diagram showing the position offsets between the AT and IRAS positions for the radio-excess sample. Plotted is the IRAS - AT positions. The position differences are distributed approximately as a Gaussian, within the 2-arcmin-square search area. The width of the Gaussian is consistent with the formal errors of the IRAS and AT positions.