5.3. Emission Properties of Double Galaxies: Pairs With Active Nuclei
The emission spectra of galaxies exhibit a large variety of processes. The intensity and richness of emission lines indicate the presence of gas, the rate of star formation, and the presence of nuclear activity. The strength of emission lines and their number on a particular spectrogram may be due to a number of factors: spectral range, resolution, spectrograph scale perpendicular to the dispersion, efficiency and characteristics of the light dispersing apparatus, and even accuracy of positioning the slit with respect to the nucleus of a galaxy. Because of such instrumental and subjective effects, comparison of spectral studies conducted by various authors is fraught with difficulty. Nevertheless, we have attempted to classify the galaxy spectra obtained for the present catalogue. Following Tifft (1982) we used a four-step scale. Type A includes the objects with a pure absorption spectrum, and in type S are the galaxies with strong emission. Besides Seyfert galaxies and the majority of the Markarian objects with blue excesses, this category also includes galaxies with large numbers of HII regions, which can contribute to strong Balmer emission. Types M (three to five lower intensity emission lines) and W (one or two weak lines) indicate intermediate categories. The reproducibility of these spectral types can be examined because of the large sub-sample in common with Tifft (1982). The agreement is substantial: in 81% of the cases the spectral types agree, in 17% they differ by one class, and in only 2% of the cases is there a difference by two classes.
This classification may be put on a quantitative basis by using measurements of the equivalent widths of the emission lines in the central regions of these galaxies. For 81 galaxies in the catalogue, such data have been collected in the work of Keel et al. (1985). Because the greatest number of spectral types have been determined from red spectra we compared these with the sum of the equivalent widths W of the emission lines H, [NII], [SII] and [OI] in the same spectral region. The distribution of 81 pair components by spectral type according to log W (in angstroms) is presented in figure 37. In galaxies of class A only very small values of equivalent width were found by Keel et al. (1985). There is an observed correlation between log W and the spectral type with scatter log W = 0.6 in each spectral type, and standard deviation 0.3. These data allow a calibration of the visual estimates of spectral type to the geometric mean equivalent widths of the red emission lines: A-2.5Å, W-10Å, M-40Å, and S-160Å.
Spectral types of galaxies should exhibit a dependence on structural type, since on going from early Hubble classes to late there is a change in the gaseous mass fraction of a galaxy and in the amount of hot blue stars contributing to the emission lines. Such a dependence is clearly visible in the data of table 20. The first line presents the mean relative mass in neutral hydrogen for each type, according to Efstathiou et al. (1982), and the second gives the relative number of double galaxies with emission spectra, qem, according to the data in column 12 of Appendix 1. Among late Hubble types practically all galaxies exhibit emission spectra.
The distribution of various spectral classes of double galaxies of each morphological type is presented in table 21. It is apparent that in going from elliptical galaxies, with a very small observed fraction of emission line objects, to later Hubble types, the number of objects is redistributed toward increasingly stronger emission. If, among E and S0 galaxies, a strong emission (S) occurs in only 6 to 8% of the cases, then for irregular galaxies such spectra predominate. Note the excess S-spectra in Sa galaxies in comparison with neighbouring Hubble types. This may be due to the fact that the concentration of a sufficiently large quantity of gas in the disk, and the depth of the potential well, form an especially fertile ground for the nuclei of Sa galaxies to display Seyfert characteristics.
Examining the distribution of double galaxies according to morphological type (section 5.1), we noticed a significant excess of pairs with the same Hubble type. An analogous tendency is observed as well for the spectral classes of double galaxies. Table 22 shows the distribution of 487 catalogue pairs according to component spectral type. The parentheses indicate the expected number of pairs for a random distribution of spectral properties. The excess of observed pairs with the same spectral class is unmistakable. The greatest excess is characteristic of those double systems in which both galaxies exhibit strong emission. Following expression (5.1) we find for the correlation function of pairs according to spectral type difference the following values: (0) = 0.82 ± 0.12, (1) = -0.03 ± 0.07, (2) = -0.47 ± 0.06, (3) = -0.66 ± 0.08. A differential analysis shows that for interacting systems the amplitude of the correlation function (0) is twice as large as it is for pairs with no signs of interaction.
The same processes which bring pairs closer together in spectral characteristics also appear to produce strong emission in galaxies. All pairs with signs of interaction represent 59% of the catalogue, but among objects with S-spectra more than 80% are interacting. This trend of emission properties with interaction has recently been examined by Kennicutt and Keel (1984), Tifft (1985) and Keel et al. (1985). The distribution of spectral types with respect to interaction is presented in table 23. These results show that the differences between the populations of various emission properties from one interaction type to another are quite large. Most of this difference arises from the varying morphological types of galaxies in each type of interaction. However, even taking this into account the differences do not disappear. Knowing the probability of encountering various Hubble types in one or another interaction class (table 18) and the distribution of spectral classes among galaxies of each morphological type (table 21) we may calculate the expected relative numbers as a correlation matrix, which is shown in parentheses in table 23. A 2 comparison indicates that the distributions differ at the p = 10-4 level. Most of the difference occurs in the observed excess of strong emission in galaxies with tails, ridges and disturbed structure in both components of the pair.
In recent years a number of facts have accumulated which indicate that the distinction of interacting galaxies from single objects extends to the strength of their nuclear activity. The effects of interaction may lead to a redistribution of gas from one component of a pair to the other, through their common Roche lobes. Another mechanism may be an increase in the tempo of star formation in members of tight double systems due to the redistribution and clumping of gas clouds in the disk of each galaxy. The observational data tend to favour the second mechanism, for which we may cite two lines of evidence: a) the excess galaxies with strong emission in table 23 are located in just those types of interaction which predominantly include spiral galaxies, which contain sufficient reserves of gas to support an increased level of star formation; the gas-poor EE pairs do not exhibit strong emission even when they show strong signs of interaction (see type A(am) in table 23); b) if in tight double systems there is an active exchange of gas from the spiral component to the nucleus of the elliptical, then E galaxies in mixed ES pairs should have a much higher percentage of emission spectra than those in EE pairs. However, among interacting doubles containing an elliptical the fraction of emission objects is only 13/74 0.18, almost exactly the same emission fraction (10/54) as observed in elliptical galaxies seen to be interacting with spiral components (7) .
As to the secondary role of gas exchange in double systems, Keel et al. (1985) studying the spectral properties of a large number of interacting pairs concluded that such a role was quite limited. As the underlying cause of strengthening activity in pairs of galaxies, these authors considered the orbital changes of the gas clouds in the disks of interacting systems. Such changes could substantially increase the probability of collisions of interstellar clouds, which could then lose angular momentum and result in active transport of gas to the nucleus. Unmistakable evidence of the comparatively high rates of star formation in double systems is furnished by studies of pairs containing objects with ultraviolet excess. Markarian (1967, 1969a, 1969b), Markarian and Lipovetskii (1971, 1972, 1973, 1974, 1976a, 1976b) and Markarian et al. (1977a, b, c, 1979a, b) performed an objective-prism survey of the northern sky with the 1-meter Byurakan telescope and discovered 1500 blue galaxies, which are known as Markarian objects. Numerous authors have performed structural studies of these galaxies, measuring their colours and the properties of their apparent distribution. It has been shown that Markarian objects are strongly concentrated in small systems of galaxies, in large part in pairs (Huchra, 1977), and Heidmann and Kalloglyan (1975) used this circumstance to estimate the orbital masses of Markarian galaxies.
A sample of isolated pairs with Markarian components, including double systems with Seyfert galaxies, was examined by Karachentsev (1981b). Among 65 such pairs only 3 were excluded from further analysis as optical pairs with f > 100. In the resulting sub-sample there were 50 pairs with one, and 12 pairs with two, Markarian objects. The morphological type distributions of the 74 Markarian and 50 normal components of these pairs, along with their spectral types, are shown in table 24. It follows that galaxies with UV excess exhibit a significant shift towards strong emission lines. NOTE: This is probably guaranteed by the selection criteria. The distribution by structural type is not significantly different from the normal distribution of such types, except for an excess at types Sa and Sb. It is known that the Markarian galaxies do not comprise any single category of objects. In addition to Seyfert galaxies, the nuclei of which show strong non-thermal luminosity, the Markarian survey selected blue dwarf galaxies rich in super-associations and intergalactic HII regions, and also a different population of galaxies of an intermediate nature, `liners', with rather weak emission or low levels of excitation. The excess number of blue galaxies of types Sa and Sb arises primarily from objects with Seyfert nuclei.
We note some additional characteristics by which pairs with one or indeed two Markarian objects are distinguished from the general run of double galaxies.
1. Upon accounting for various selection effects, the Markarian components of pairs have brighter absolute magnitudes by (0.45m ± 0.13m) than do normal galaxies. This excess luminosity may be interpreted as the amplitude of increased star formation which itself accounts for the blue colours of the galaxies.
2. Pairs containing Markarian galaxies are distinguished by small values of the radial velocity difference and linear separation of the components. Thus, systems with two blue galaxies have <y> = 50 ± 12 km/s and <X> = 17 ± 7 kpc. This property is apparently due to stronger interaction producing active star formation.
3. Markarian objects exhibit a decrease in mass-to-luminosity ratio. For 50 pairs with a single Markarian component, the mean orbital mass-to-luminosity ratio is (6.1 ± 1.6) f instead of the (11.1 ± 0.8)f for the entire sample, and for pairs with two Markarian galaxies it is only (1.8 ± 1.7)f. According to Thuan and Martin (1981) such low values of f are characteristic of gas-rich galaxies in which bursts of star formation can occur (and, much the same, an increase in integrated galaxy luminosity).
The causal relation between blue colour excess in galaxies and the distance of their nearest neighbour was first illustrated statistically from the distribution of Markarian objects. The catalogues of isolated galaxies (Karachentseva, 1973), doubles and triples (Karachentsev et al., 1979) cover one and the same region of the northern sky to identical limits, but the probabilities of encountering Markarian galaxies among them are remarkably different. For pairs, the relative number of blue objects is 7.6% while among single and triple objects it is 1.4% and 2.8%, respectively. Further evidence for the non-random association of Markarian galaxies in pairs comes from the relative numbers of double systems with one and two blue objects. The observed number of pairs (n = 12) containing two Markarian galaxies exceeds the expected number for a random distribution by more than four-fold. This substantial excess of Markarian galaxies in double systems compared to both singles and triples shows that the excess star formation is controlled or triggered by the presence of the nearest neighbour. This also accounts for the decreased number of blue objects among triple systems, where both the separation between galaxies and the relative velocity differences are greater than in pairs. The exact same reasons explain the decreased number of Markarian galaxies in rich clusters and groups.
The idea that active conditions in galactic nuclei are enhanced by interactions with close neighbours has received much support in recent years. Dahari (1984) surveyed the neighbours of Seyfert galaxies and found that 15% of the Seyfert systems have close satellites, rather than the 3% for a control sample of galaxies with normal nuclei. Hutchings and Campbell (1983) performed an analogous survey of 45 quasars with redshifts z < 0.62. From deep images of the surroundings of these quasars the authors came to the conclusion that more than 30% of them exhibit signs of interaction with neighbouring galaxies. Measuring the redshifts of such galaxies near quasars, Heckman et al. (1983) found that among 19 investigated quasar-galaxy pairs the 18 with X < 50 kpc exhibited radial velocity differences less than 1000 km/s. The small mean radial velocity difference, 190 km/s, confirms the physical association of the galaxies and quasars and further allows orbital mass estimates of the quasars, which are typical of galaxies of the highest luminosity. Those authors noted that the observed interactions apparently serve as a triggering mechanism, igniting the nucleus of the galaxy and turning it into a bright quasar. Bothun et al. (1982) advanced the radical proposition that almost all quasars may result from close interactions of galaxies which are rich in gas, and that the observed strong rise in the number of quasars from z = 0 to z = 3 reflects the larger number of close galaxies at early epochs when the separation of galaxies was much smaller. In such a picture the evolutionary role of double systems in accelerating the process of star formation could be very large.
The gravitational disturbance present in many double galaxies may cause not only ultraviolet excesses, but also a brightening in the infrared regime. According to the observations by Rieke et al. (1980) strong infrared emission is present in pairs 161 and 466 in the current catalogue. According to Soifer et al. (1984), the interacting pair 470 (Arp 220) exhibits a record infrared excess. Its luminosity in the infrared spectral range exceeds that in the B-band by two orders of magnitude. Based on observations in the 1-10µ range of two dozen interacting systems, Lonsdale et al. (1984) concluded that the rate of star formation in interacting and merging double galaxies exceeds by a factor of three that in normal galaxies. According to their estimates the extra star formation in interacting systems increases the mean infrared luminosity by 30%; around 10% of massive stars in galaxies of all types may be triggered by such disturbances. This excess of infrared emission in interacting pairs by comparison to control samples was soon confirmed by Cutri and McAlary (1985).
Smirnov and Tsvetkov (1981) compared the relative numbers of supernovae in isolated and double galaxies, and showed that the supernova frequency normalised to unit luminosity is increased in double galaxies. They attributed this result to the much higher rates of massive star formation in interacting pairs.
Evidence of increased activity in double galaxies was also found in the radio regime. Stocke (1978) and Stocke et al. (1978) conducted a large scale program of observations of all pairs in our catalogue at wavelengths of 6 cm and 11 cm. They established that for tight pairs the radio luminosity exceeds by a factor of two that for wide pairs. This holds true for spirals as well as for ellipticals. The excess number of radio sources in pairs was confirmed by Condon and Dressel (1978) and Condon (1980). For many double systems detailed radio maps with high angular resolution were obtained by Condon (1983) and Condon et al. (1982). Comparison of these results with optical data showed that the regions of radio emission are usually rich in massive stars. Therefore, the authors concluded that the increased radio emission is caused by active star formation, which is in turn enhanced by interactions. Excess radio emission in pair members by comparison with isolated galaxies was demonstrated by the surveys of Altschuler and Pantoja (1984) and Dressel (1981). Properties of the radio emission in double galaxies were given in the works of Wright (1974), Sramek and Tovmassian (1975), Sulentic (1976), Hummel (1981), Biermann et al. (1981), Wirth et al. (1982), Bieging and Biermann (1983), and Petrosian (1984). NOTE: Add also the recent Hummel et al. pair of survey papers.
7 Our earlier conclusion (Karachentsev, 1981e) of an excess of emission in elliptical galaxies in mixed pairs is not valid, after revising the morphological types. Back.