An Atlas of DRAGNs

Classification Schemes for DRAGNs

To come: figures of FR division vs Mag; spectrum of 3C84. Other type spectra? Naked jet image?

Fanaroff-Riley Class

In a classic paper, Fanaroff & Riley (1974) divided extended radio sources into the following two structural classes:

The separation between the points of peak intensity in the two lobes is smaller than half the largest size of the source.
The separation between the points of peak intensity in the two lobes is greater than half the largest size of the source.

The remarkable thing about this simple criterion is that it correlates very closely with radio power: there is a break power of about 1025 Watt Hertz-1sr-1 below which nearly all DRAGNs are FR Is, and above which nearly all are FR IIs. Owen & White (1991) found that the break power actually increases with the optical luminosity of the host galaxy; taking this into account, the FR transition is extremely sharp. --Figure here--

By coincidence, when applied to starburst galaxies like M 82, which have very low radio power, the FR criterion tends to give the 'right' answer: FR I. While this at first seemed an advantage, now that DRAGNs and starbursts are recognised as fundamentally different phenomena we prefer to think of the FR classes as subtypes of DRAGN, and from this perspective it is meaningless to discuss the FR class of other types of radio source.

The FR classification of a DRAGN is usually obvious, but there are some ambiguous cases. Occasionally, a peak with relatively low flux density, which initially appears insignificant, turns out at high resolution to be the brightest point because its flux comes from a very small region, and this can change the classification (c.f. 3C 433). Fanaroff & Riley were working with relatively low-resolution maps, and their analysis works best when applied to such images; the FR classes quoted here are therefore based on our C20 maps.

We have ignored steep-spectrum cores when assigning FR classes, which therefore refer just to the outer structure.

In recent years, high resolution images have revealed various kinds of structure which are typical of one or other of the FR classes. There has been a tendency to describe objects as FR I or FR II based on the presence of such typical structures (for instance, hotspots in FR IIs) even when the original definition would have given a different result (c.f. 3C 171). Sometimes structures typical of different FR classes appear in the two lobes of the same DRAGN, suggesting an 'FR 1.5' classification. We have stuck to the original definition in this Atlas, because otherwise it is all too easy to find a reason to reclassify DRAGNs which have the 'wrong' luminosity and so 'improve' the correlation with radio power!

A more detailed scheme

The FR classes, especially the FR Is, cover a broad range of structural types. In fact, the main achievement of Fanaroff & Riley was to find a simple structural property common to the various types of complex low-power DRAGNs that were known at the time (cf. Miley & van der Laan 1973).

The various structural types depend on which of the basic "ingredients" listed on the anatomy page they contain. We propose a two-part classification in which each radio lobe is characterised by the prominent compact structure (if any), and by whether the diffuse lobe is a bridge or a plume:

None Weak-flavour jet Hotspot Strong-flavour jet
Here R stands for relaxed. It is not possible to tell if a relaxed lobe is a bridge or a plume, as the distinction involves the position of the end of the jet or hotspot.

In most DRAGNs the two lobes have the same classification on this scheme. This suggests seven basic types of DRAGN; but in fact prominent strong-flavour jets very rarely occur in both lobes, so there are really only five simple types. These correspond to the types in common to the schemes proposed by Leahy (1993) and Laing (1993):

Lobe typeUsual nameFR classExampleComments
R+RRelaxed Double (RD)I or II 3C 28Sometimes called Fat Doubles
HaloesI 3C 84 See the discussion of lobes.
Peculiar Relaxed Double (RD pec) I3C 338,
3C 442A
These two objects show prominent filamentary structures connecting the lobes which would have been classified as jets, except that they do not come from the active nucleus. The lobe structures are also more complex than typical bridges.
WP+WPTailed Twin Jet (TTJ)I 3C 31This example has plumes, i.e. it shows little overall bending.
Narrow-angle Tail (NAT)I 3C 83.1 BThe jets are bent back to become approximately parallel and the lobes form a tail trailing away from the host galaxy. The structure is caused by the motion of the host galaxy through the intra-cluster medium.
Head-Tails (HT)I 4C 35.40Similar to NAT, but there is apparently only one jet. This may be because twin jets are not clearly resolved, or because one jet is quickly destroyed (perhaps because it is pointing directly into the oncoming wind of passage), or because there only ever was one jet.
WB+WBBridged Twin Jet (BTJ)I 3C 296
HP+HPPlumed double (PD)mostly I3C 171 Both Leahy and Laing call these WATs, but not all fit the conventional WAT stereotype, so we have renamed the group to avoid confusion.

May contain strong-flavour jets.

Overlaps with the "Peculiar doubles" of Law-Green et al. (1995a), which also included mixed types such as HP+HB.

Wide-angle Tail (WAT)mostly I 3C 465 See separate discussion.
HB+HBClassical Double (CD)II 3C 173.1May contain strong-flavour jets.

Our new scheme allows unambiguous classification of the uncommon objects with different types of lobe on the two sides; for instance 3C 319 is HB+R, NGC 6251 is HP+WB, 3C 215 is SP+R.

Other Types

For comparison with other work, we mention the following DRAGN classifications, which do not form part of the system outlined above.

Naked Jets

Contain a core and weak-flavour twin jets, with no sign of a lobe. (FR I). They tend to be the least luminous DRAGNs. There is a tendency for jets to become relatively more prominent in weaker FR Is, and deep observations (e.g. de Ruiter et al 1990) have found weak lobes around some apparently naked jets, so this class probably just consists of weak twin-jets, observed with insufficient sensitivity to detect the lobes.

There are no naked jets in the Atlas sample.

Wide-angle tails

Contain strong-flavour jets which disrupt at quasi-hotspots, and plumes or tails. (FR I). Despite the name, WATs do not have to be bent at all; see the comments on the prototype, 3C 465.

Traditionally, WATs have been defined as DRAGNs closely resembling the prototype; they have generally only been sought in clusters of galaxies, where they tend to be associated with cluster-dominating galaxies. WATs are clearly members of our plumed double class. Compared to other PDs they tend to have weaker hotspots, but it is not clear how distinct WATs really are from other PDs.

Jetted Doubles (JD)

Defined by Law-Green et al. (1995a) as cases where one lobe is dominated by a strong-flavour jet. This includes SB+HB (3C 200, 3C 346 and 3C 401), SB+HP (3C 433), as well as extreme cases like 3C 48 where the DRAGN is so jet-dominated that it is difficult to classify the lobes.

To avoid confusion between these types, we do not use the JD classification in the Atlas.

Classification based on size

A popular alternative approach to classifying DRAGNs is via their size. A number of acronyms exist for objects in various size bins:

0 to 1 kpc
Compact Symmetric Object (CSO), Compact Double.
1 to 10 or 15 kpc
Medium Symmetric Object (MSO), or Compact Steep-Spectrum (CSS) source.
>666 kpc
Giant radio source.

The choices of the various size bins is generally for historical reasons, based on observational constraints. Different authors may use different names for nearly equivalent classes of objects, and the precise definitions are not well established; for some authors the angular size is the relevant factor. There is not a close correspondence between size classes and the morphological classes we have defined above. Since we quote the linear and angular sizes for all objects, the size class gives no extra information and we tend to avoid using this terminology.

Classification based on spectrum

Another alternative to classification by structure is classification in terms of the radio spectrum. This is widely used because it is much easier to determine spectra (at least roughly) for large numbers of objects than to make radio images. To some extent, spectral and structural classifications are complimentary, although one can learn something about the structure from the spectrum and vice versa.

Steep-Spectrum Sources

The typical objects found in low-frequency surveys, with a nearly power-law spectrum with spectral index greater than about 0.5, characteristic of optically thin synchrotron radiation. They normally have angular sizes of tens of arcsec or more, and are mostly identified with galaxies.

Note that these spectra are "steep" in contrast to flat or inverted spectra; the term does not imply "steeper than average spectrum" (c.f. USS sources).

For reasons discussed under flat-spectrum sources, these objects are now generally called "lobe-dominated" rather than "steep-spectrum" sources.

Almost all the Atlas DRAGNs are in this spectral class.

Ultra Steep-Spectrum (USS) sources

Lobe-dominated sources can have spectral indices ranging from about 0.4 to 1.5 or even more, with most bunched between 0.6 and 0.9. For some purposes it is useful to concentrate on those with unusually steep spectra, the so-called USS sources. The definition is arbitrary, but objects with spectral index greater than 1 are often taken to be USS. Samples of USS objects contain more or less the same range of structural types as any sample of steep-spectrum DRAGNs, but tend to include a larger number of extremely high redshift DRAGNs, and also relaxed doubles and related types in clusters of galaxies.

High-redshift objects are excluded by definition from the Atlas. Examples of cluster-type USS objects are 3C 310 and 3C 338.

Gigahertz Peaked Spectrum (GPS) sources

These have spectra with a single maximum at about 1 GHz. The turnover from "normal" steep spectrum at high frequency to inverted spectrum at low frequency is probably due to synchrotron self-absorption, and should occur in all steep-spectrum synchrotron radio sources. The turnover happens when the radio intensity reaches a critical strength, and so occurs at higher frequencies when powerful emission is generated in a very small volume. In most sources the turnover frequency is too low for us to observe, but for bright and extremely small sources (typically with total size of a few hundred parsecs, and with hotspots only a few parsecs across), we get a GPS spectrum. Thus GPS sources are the smallest (and presumably youngest) DRAGNs, the CSO's and compact doubles. They typically have classical double structure.

There are no GPS sources in the Atlas sample.

Flat-Spectrum Compact (FSC) sources

These have roughly flat or inverted radio spectra, and at first seemed to have very small angular size and were normally identified with quasars; in effect they seemed to be pure compact core.

The distinction between extended steep-spectrum sources and compact flat-spectrum sources once seemed absolute, but is now recognised to be rather misleading. It turns out that most steep-spectrum sources contain faint compact flat-spectrum cores, while deep images of FSC's usually show that they are surrounded by faint extended steep-spectrum lobes. We therefore prefer to talk about "core-dominated" and "lobe-dominated" DRAGNs. According to unified schemes, even this distinction may largely be an illusion, with the same object appearing core-dominated or lobe-dominated depending on the angle at which we view it.

Because the core and lobes have such different spectra, if they are comparable in flux density then the core will dominate at high frequency and the lobes at low frequency. Thus, all the DRAGNs in the Atlas sample are lobe-dominated at the selection frequency of 178 MHz, but 3C 84 is core-dominated above about 1 GHz. ---Insert GIF of spectrum of 3C84---. We quantify how "core-dominated" a particular DRAGN is by quoting the ratio of core to extended flux density at an emitted frequency of 5 GHz, the so-called R parameter.

<<<< Atlas Index >>>>
Alphanumeric List Icon List Jodrell Bank home page

Last modified: 1997 June 26
J. P. Leahy