Candidates for a Southern Extension of the Karachentsev Catalogue of Isolated Pairs of Galaxies

Luca REDUZZI (1) and Roberto RAMPAZZO (2)

1) Dipartimento di Fisica, Università degli Studi di Milano, Milano, Italy
2) Osservatorio Astronomico di Brera, Milano, Italy

(Astrophysical Letters & Communications, 1995, vol.30, n.1-6)

GO TO THE ATLAS



ABSTRACT


The paper presents a sample of double galaxies selected from The Surface Photometry Catalog of the ESO-Uppsala Galaxies using the Karachentsev (1972: K72) criteria. Considering the large and growing number of observing facilities in that hemisphere, the sample aims to supply investigators with targets as homogeneous as possible to those in the Northern Catalog of Isolated Pairs of Galaxies (K72) which have been studied in a large frequency range. The paper discusses the sample degree of completeness, separation and velocity distributions. First inferences are sketched from the morphological association of pair members and from the study of the Holmberg effect. Being the primary purpose of the paper to provide a tool to investigators, an Atlas of images for the 301 best candidates pairs obtained using the Digitized Sky Survey (*) is given. The Atlas reports for each pair member the relevant data as obtained from the current literature.

* The Digitized Sky Survey was produced at the Space Telescope Science Institute (ST ScI) under U.S. Government grant NAG W-2166.


INTRODUCTION


Besides the historical objective of the galaxy mass determination, binary galaxies offer a `simple' tool through which to study the evolution of the gravitational interaction and induced phenomena. So, the study of the dynamical friction, the angular momentum and energy transfer, merging as well as the secular and/or co-evolution of galaxies stellar populations, the development of stimulation effects (starburst, LINER or even AGN activity) found in pairs a fruitful soil (see Barnes and Hernquist 1992 and reference therein).
Most of these studies need a statistical approach. This requires that the sample of pairs under study has to be a priori well determined and homogeneous. The selection criteria have to identify the largest fraction as possible of true gravitationally interacting objects, avoiding or controlling the number of interlopers like false pairs (i.e doubles embedded in high density fields) and optical pairs (i.e chance optical alignments). At the same time the criteria have to be very clearly defined in such a way that main biases can be mastered in the subsequent analysis.
After the pioneer works done by Holmberg (1937, 1954) and Page (1960), Karachentsev built The Northern Catalog of Isolated Pairs of Galaxies (K72: CPG hereafter). CPG contains 603 candidate pairs and claims a completeness up to 15.5 magnitude. The source data base was the Catalog of Galaxies and Cluster of Galaxies by Zwicky et al. (1968: CGCG). CPG cover all the northern hemisphere and extends to delta=-3 degrees. The selection was based on semi-empirical simple criteria of isolation and hierarchy. Different criteria have been adopted in subsequent pair selections in the northern hemisphere by Turner (1976), Peterson (1979) and more recently by Soares (1992). Besides the fact that even starting from the same source (like CGCG) each compilation have a low fraction of pairs in common, due to the different adopted selection criteria, it has been noticed that some compilation failed the aim they set. For instance, the criteria adopted by Turner (1976) produce an high percentage of optical pairs larger than supposed (Rood 1982). Further, Turner's pair are mainly found to belong to groups or even clusters (White et al. 1982). Since Turner's criteria tend to select pairs with small projected separation, Peterson's (1979) compilation relaxed the criteria succeeding to include large pairs, but enhancing the fraction of optical alignment.
After having obtained the velocity difference for the members of the entire CPG, it results that optical pairs represent ~11% of the sample (Karachentsev 1989). On the other side, CPG shows a lack of strongly hierarchical pairs, while those with small separation are predominant as in the previous compilations. The mean separation, after correction for the selection criterion, is <x>=82 kpc (Ho=75 km/s/Mpc) vs. <x>=142 kpc obtained by Soares (1992) using van Albada criteria.
For the southern hemisphere a compilation similar to the CGCG which is complete up to the 15.5 mag does not exist, if we exclude the surveys done with the COSMOS machine (MacGillivray and Stobie 1984) which can be as deep as bj=20. These latters cover different sky areas and the subdivision between galactic vs. extragalactic objects is done in automatic way, then introducing a possible source of interloopers if used as starting list (see Heydon-Dumbleton et al. 1988).
Different approach in selecting doubles has been attempted using velocity difference (obtained from Southern Sky Redshift Survey: da Costa et al. 1991) and projected separation cuts by Charlton and Salpeter (1990). No isolation criterion has been adopted, but galaxy density constraints, so the sample include binary galaxies with separations up to 1.5 Mpc. More traditional compilations of pairs have been produced using different kind of criteria in limited southern sky areas by Arp and Madore (1985), Zhenlong et al. (1989), Sulentic (1991: unpublished). In particular, Soares et al. (1994) have obtained from The Surface Photometry Catalog of the ESO-Uppsala Galaxies (Lauberts and Valentijn 1989: ESO-LV) a sample of binaries using the criteria of likelihood of physical association. They estimated that the frequency of optical pairs (velocity separation > 1000 km/s) is ~20%.
We will adopt the following, already in use, terminology through all the paper. `Binary' refers to the underlying physical objects comprising the population we wish to identify, `double' to an observationally selected sample of objects. `Pairs' are those double systems which are chosen to represent the underlying population. Then, considering the higher and growing concentration of observing facilities in the southern hemisphere, the paper aims to provide a compilation of double systems and propose a sample of southern candidate pairs as partial complement to the CPG. This latter will be useful to study the properties of binary galaxies in different key problems. We started from the ESO-LV catalogue, exploiting the photometric data there available both for identifying double galaxies (following the K72 criteria) and for deriving the first properties of the selected sample. In \S2 will be discussed the candidates selection criteria and their application to the ESO-LV. The sample properties will be presented in \S3. Since this paper is intended as a tool for pairs investigation, in an Atlas are collected the image of each candidate together with a pair description and the basic data for each pair member.


SAMPLE SELECTION


Although big efforts have been done in the recent years in order to conduct redshifts surveys most of them are devoted to the mapping of the large scale structure and then they do not reach an homogeneous and complete covering of the sky. Therefore, a realistic expression for recognition of double galaxies must be still formulated operating on more easily obtainable observables like angular separation, apparent magnitudes, and angular diameters of the galaxies as most of the previous investigation on double galaxies. From general considerations a physical pair of galaxies should consist of galaxies sufficiently close each other that the energy of interaction between them and any neighbouring galaxies should be smaller than the potential energy of the interaction between the two galaxy in the pair. To express this using the previous observable properties necessarily leds to some assumptions which basically tend to limit the inclusion of chance optical systems.
The criteria we adopted are similar to the ones adopted in K72 for the selection of the doubles in the CPG. They are based on simple observable properties of galaxies: their angular projected separation and apparent diameters.


Selection criteria applied to ESO-LV


Two galaxies of angular diameter A1 and A2 whose separation is X(1,2) will satisfy the Karachentsev isolation criterion if

X(1,i) / X(1,2) >= K (Ai / A1),
X(2,i) / X(1,2) >= K (Ai / A2),

where i indicates each of the neighbour galaxies whose diameter Ai is included in the intervals set by

C A1 <= Ai <= L A1,
C A2 <= Ai <= L A2.

Since we are looking for isolated pairs, last conditions imply that are not considered possible perturbers those galaxies whose diameters are significantly different from that of the galaxies under investigation. Possible values of X/A measured by Turner (1976) and Peterson (1979) are in the range X/A<5, while in the loose groups the nearby members have separation typically of the order 10 diameters or larger. The selection criteria have then been calibrated following the previous indication. We adopted here as adimensional coefficients in the two previous formulae values similar to those in K72:

K = 5,
C = 1/2,
L = 4.

The ESO-LV catalogue, obtained from the ESO-SERC plates, contains all the galaxies larger than 1 arcmin south of delta=-17 degree for a total of 15467 objects. This catalogue is diameter limited and complete up to B=14.5. If some galaxies were found to show sign of interaction and a possible nearby companion smaller than 1 arcmin is visible it was also included in the catalogue. The authors explicitly say that ``a strong aspect of the ESO-LV is that, in a rather homogeneous way, photometric and structural parameters have been determined for a large sample of galaxies irrespective of their environment. Therefore, the catalogue can be used for studies of galaxy parameters in relation to their environment''. Since ESO-LV gives galaxies positions with an average precision of 3 arcsec and different photometric diameters (e.g. a(25) the diameter at mu=25 magnitudes/arcsec^2, all the parameters needed to apply the criteria are available. As representative diameter of the galaxy we selected a(25), avoiding any `statistical' correction. In fact, a(25) is 1) a reasonably well determined measure (typical uncertainties within 0.2 magnitudes/arcsec^2, see for example Rampazzo et al. 1994) while at fainter surface brightness they rapidly fall to 0.5 magnitudes/arcsec^2 or larger frustrating any sort of correction. 2) a(25) is considered in many astrophysical works, like in galaxy dynamics, the typical optical radius within which to compare galaxy properties. 3) Any correction for inclination for the S members should be due to galaxy internal extinction which is largely unknown and matter of debate (Disney et al. 1989). The latter authors estimate that `if the external galaxies have star:gas:dust ratio not dissimilar to our Galaxy's, their surface brightness and their known column densities in HI and CO suggest that indeed they ought to be optically thick out to isophotes between 23 and 25 mu(B)'. This means that the surface brightness may be considered independent of inclination (while apparent luminosity have to be corrected).
For each galaxy in the catalogue, a circular area of a radius 20 times the diameter of the galaxy under investigation has been inspected. The criteria have then been applied recursively to all the objects found inside. We extracted in total 1302 candidate doubles.

FIGURE 1. Integral distribution of galaxies (panel a) and pairs (panel b) as function of the total B-band apparent magnitude. The solid line indicates the slope of the homogeneous distribution N ~10^(0.6B). The different symbols indicate the entire sample (1302 candidates: open squares) and the final selection (409 pairs: crosses)

In Fig.1 is shown the integral distribution of the member's apparent total magnitude compared with the line slope expected for a uniform distribution in a Euclidean space. Together with perturbations at bright magnitudes due to nearby structure (see \S3), it is possible to infer that the distribution starts to lose galaxies approximately at 14.5-15 magnitudes. A similar information can be deduced from Fig.2. In this latter is shown the result of the application of the classical <V/V(max)> test (see Thuan and Seitzer 1979 for a full explanation).

FIGURE 2. The <V/V(max)> classical test (see Thuan and Seitzer 1979 and reference therein). The distribution of our candidate pairs differs from the 0.5 canonical value of a complete distribution in a Euclidean space at very bright magnitudes and starting from 14.5 - 15 magnitudes. This means that we lose nearby pairs, a known bias due to the selection criteria, and fainter pairs due to the source catalogue incompleteness.

The <V/V(max)> test has the important characteristic that can be applied without the knowledge of the redshifts of the galaxies, as in our case. The value of <V/V(max)> should be 0.5 for objects uniformly distributed in a Euclidean space if the sample is complete. We see that our compilation is reasonably complete up to 14.5 magnitudes (which is the limit of completeness of the ESO-LV) and for diameters greater than 1 arcmin. This is the reason because we cut our sample excluding those pairs in which both members have a total B magnitude lower than 15. Without losing completeness, the final sample is composed by 409 doubles.
Since the ESO-LV catalogue is diameter limited and we automatically applied the selection criteria, it is possible both 1) that a companion is lost if its diameter is lower than 1 arcmin or 2) that a double could be classified as isolated although there are possible faint companions nearby. The first cause of double loss is balanced by the fact that the K72 criteria select nearly similar size members but a full strategy cannot be adopted. Concerning the second point all the extracted doubles have been checked on the Digitized Sky Survey (DSS hereafter) and on the ESO-LV catalogue itself. In some critical cases we looked in detail if all the galaxies, having similar size and nearby the investigated double were in the catalogue so that the K72 criteria have been applied to them. Each double has been then classified on the ground of the inspection of the DSS. We devised different fiducial classes for the doubles in order to take into account the degree of isolation which is, at certain extent, given by the parameter N(tot) in the ESO-LV (see later discussion in \S3). The meaning of the different classes is reported in Table 1. The symbols C+ and C-, used in the Atlas and Table 2 indicate the strength of the K72 criteria satisfied by each pair (see the introduction to the Atlas); in some sense they can be used as another indicator of the degree of isolation of the object.

TABLE 1: Fiducial classes for the selected doubles
Class Meaning
Pthe double verifies criteria and visual inspection on DSS
PTPossible Triplets, the third member is not in ESO-LV
PGPossible Group, some members are not in ESO-LV
NOthe double is in a very high density region and most of the galaxies are not in the ESO-LV
OPOptical Pair when DeltaV is > 1000 km/s
Stwo entries in ESO-LV but indistinguishable in the DSS


The global properties of doubles with fiducial class P are reported in the Atlas together with their images. These constitute a sample of 301 objects which we consider the best candidate pairs. We have also classified them in DIS, LIN and ATM type following the same classification adopted in K72. The specific meaning of such classes is reported in the Atlas legenda. The properties of the doubles in the remaining fiducial classes are resumed in Table 2.
Some of the pairs in the Atlas have been already observed in order to study their structural properties. CCD images have been obtained at the 92cm ESO-Dutch telescope located at La Silla (Chile), during different observing runs. Typically each pair has been observed in the B, V and R Bessel bands. The seeing was in the average of 1.2 - 1.5 arcsec. Typical exposure times were 45 minutes in B, 20 in V and 15 in R. The average image resolution is 0.4 arcsec/px. The detailed study of these objects together with their isophotal maps will be presented in a forthcoming paper (Reduzzi and Rampazzo 1994: in preparation).
The bulk of recession velocities has been primarily obtained from those reported in ESO-LV, but a substantial fraction of new measurements come from the recent surveys made by Fairall and Jones (1991: code F in the Atlas and Table 2), from da Costa (1994: private communication, code dC) and from Maurogordato et al. (1991: code MPB). A small sample (41 objects) has been also observed in CO at the SEST telescope at La Silla, by Combes et al. (1994); the fluxes derived by this work are also reported. In the same paper the R and H(alpha) images of such candidate pairs are displayed. These latter are coded as CPRS94 in the Atlas. CCD images of two objects are also showed in Rampazzo and Sulentic (1992) and are coded as RS92.


Optical pairs, false pairs and comparisons with similar compilations


Karachentsev and Shcherbanovsky (1978) have numerically simulated both the apparent distribution and dynamics (Hubble flow plus peculiar velocities) of a typical POSS 6X6 degrees field examined by K72. Knowing the distribution of doubles in the model they have evaluated what fraction of them were selected applying the criteria adopted. They simulated 127 fields with a distribution of galaxies complete up to 15.7 magnitudes. Applying the criteria a fraction of 11% of doubles demonstrates to be optical, when a velocity difference of 500 km/s is selected as the cut off limit. Further 32% of the pair detected are member of groups of different richness and then have to be considered false pairs. It is in fact obvious that the members of false pairs are not dynamically isolated from the other members of the group, although they may predominantly gravitationally interact with each other. Since peculiar velocities inside groups are of the order of 200 - 500 km/s it is not straightforward to separate false pairs from isolated ones. Adopting more restrictive parameters in the criteria, the number of false pairs would diminish at a loss of the physical ones. In fact, the previous authors computed that in order to reduce of a factor of three the optical pairs and of a factor of two the false ones there is 50% of probability of losing real isolated pairs. Even if estimates done by Karachentsev and Shcherbanovsky (1978) are model dependent, it is realistic to expect similar percentage of optical and false pairs in our compilation. This is among the reasons because we identified in the DSS those candidates pairs with a higher density of nearby objects and devised for them different fiducial classes (classes PT and PG in Table 1 and 2)

FIGURE 3. Distributions for apparent B magnitudes (panel a), diameters (panel b), difference of the member's diameters (panel c), projected separation (panel d), absolute M(B) magnitude (panel e), linear diameter (panel f), difference of member's apparent B magnitude (panel g). Relation between linear diameter and M(B) (panel h). The absolute quantities have been computed for the objects for which the recession velocity is available using Ho=75 km/s/Mpc.

In Fig. 3 are plotted the main characteristics of the selected 409 doubles whose distribution on the celestial sphere is shown in Fig. 4. The Fig. 3 (panel a) shows, once more, that the distribution starts to miss objects starting from 14.5 magnitudes. The distribution of apparent separation (panel d) is peaked toward small values. Similarly there is a small spread between member galaxies apparent magnitudes and diameters, as expected by the application of the K72 criteria (panels c and g).
Some biases of CPG are emphasized by Sulentic (1990). We wish to remember, in particular, that since small projected separations are privileged this means to select special orbital phases, i.e. near the perigalacticon. At the same time merger candidates, near coalescence, can be lost if they are classified as a single galaxy, irrespective to the double nuclei appearance. Merger candidates may rightly be considered the final phase of the dynamical evolution of pairs. Bergvall (1981 a,b) in his survey, for instance, found a large number of isolated objects that show signs of disturbance. In the ESO-LV there are many objects with two entries in the catalogue that at an inspection on the DSS reveal as single galaxies. Many of such objects have disturbed appearance and this is the reason because we did not discard them. A further analysis is needed in order to investigate the nature of this class that we indicate as S in Table 1 and 2.

FIGURE 4. Distribution of the sample in celestial coordinates.

None of the compilations are similar to another or free from biases, not only for the different adopted criteria but also because of the different methods used in measuring the applied employed parameters (in particular isophotal diameters). With this in mind we will discuss in the next paragraphs the comparison of our sample with two similar compilations which have been obtained from the direct inspection of the plate material. This will give us a direct check about how reliable is our automatic search procedure.
A first comparison may be obtained with a compilation of 29 pairs published by Sadler and Sharp (1984). 13 objects out of 29 (45%) of the Sadler and Sharp list of objects appear also as candidate pairs in our catalogue. The percentage is justified by the fact that the authors used less restrictive Karachentsev criteria. The effect of the different values adopted on the sample selection can be illustrated with some examples in nearby pairs. NGC 1549/1553 is present in the Sadler and Sharp compilation. The double is in the Dorado Cloud (53 -1 +1; Tully 1988) together with 11 galaxies having an average recession velocity of 1006 km/s. These objects have a recession velocity respectively of 995 and 1024 km/s (Rampazzo 1988). NGC 1549/1553 is also known for the system of shells and ripples present in both galaxies (Malin and Carter 1983) which origin is ascribed to the accretion of a small companion (Barnes and Hernquist 1992), so they are probably in a rich environment. The double NGC 1531/37 (our Pair #85) is also in the Dorado cloud and forms a subclump (53 -13 +13), probably a triplet (Tully 1988), with NGC 1532. On the other side, there are different evidences that NGC 1531/32, excluded by our criteria being too hierarchical, is a real pair. For instance, the spectra of the spiral member NGC 1532 shows an inversion of the [N II]/H(alpha) (see Sperandio et al. 1994), a phenomenon which is probably connected to the interaction induced bar formation (see Rampazzo et al. 1994). In any case, it is not possible to exclude that all these objects in the Dorado cloud may fall into the general category of false pairs described above.
A second useful comparison is the unpublished compilation by Sulentic (1991: private communication) used as basis in the study done by Rampazzo and Sulentic (1992). Besides the fact that it has been extracted directly from plates, the compilation includes objects fainter than those listed in ESO-LV. Then it may give us an indication about the completeness of our compilation. The Sulentic compilation contains 119 candidate pairs in the range 15h - 3h. The comparison gives the following results. 47 pairs are not found because members do not belong to ESO-LV, but for only 10 among them one member has a magnitude lower than 15 (8% of the total sample) than have to be in our sample but the lack of the companion in ESO-LV prevents us to find it. Finally 12 are not found because they do not satisfy in a rigorous way the criteria we applied. These latter are, in all cases, candidate pairs in rich environment, where the strict application of the rules is critical. The direct work on the plates does not facilitate the measurement of the diameters and, in some sense, is a more subjective and less repeatable measurement than one obtained from accurate surface photometry. In any case, even if we consider this 12 as discordant cases, the agreement with the Sulentic compilation reaches the 80%.


FIRST ANALYSIS OF THE SAMPLE GENERAL PROPERTIES


For 169 pairs out of 409 (41%) is known the velocity difference of the members while the recession velocity for at least one member is known for 491 objects (60% of the total). In many cases there exists more than one measure for each galaxy member. The estimate given by the different authors for the error on the recession velocity is, in average, 50 km/s. It corresponds to the typical formal error of the mentioned surveys. In the following analysis we subdivided the pairs into three classes: E+E pairs are composed by members which are both classified as early-type galaxies in ESO-LV (Type < 0.0), in S+S both are late-type (Type > 0.0), while E+S are mixed morphology pairs. 150 out of 169 pairs (89%) have a velocity difference lower than 1000 km/s. We selected this limit differently from Karachentsev (1989) because pairs with velocity difference larger than his limits (500 km/s) are actually known to show strong sign of interaction (see Combes et al. 1994 and reference therein), although they may represent unbound encounters. We then estimate that 11% of our doubles could be optical alignment.


Distribution of projected separations and relative velocities of members


In Fig. 5 we plot the distribution of the projected separation for the candidates in our sample and for each of the three morphological categories of pairs above described. As noticed before, it appears evident that the pairs with small projected separation are privileged. All categories have a median projected separation ~35 kpc similarly to the average value of 32 kpc obtained by Karachentsev (1990) before any bias correction. E+E pairs show a median separation lower than the other two categories (see also Demin 1984).

FIGURE 5. Distribution of linear projected separation for the sub-sample of doubles with recession velocity difference known (panel a). The same is shown for the three different classes of candidate pairs (panels b,c,d).

In Fig. 6 is shown the distribution of the recession velocity for all the galaxies for which a measurement is available (simple shading) and for the doubles for which the velocity of both members is known (double shading). Since the trend of the two distribution is very similar, the analysis we performed on the latter sample (41% of the total) may give useful information on the general properties of all candidate pairs.

FIGURE 6. Distribution of the recession velocity for the galaxies in the sample for which a measure is available (simple shading) and for pairs with velocity difference known (double shading). Since the trend of the two distribution is very similar, the analysis we perform on the second sub-sample (41% of the total) may give useful information on the general properties of the full sample of candidate pairs.

The spatial distribution of pairs obeys a common law of hierarchical clustering as in the northern hemisphere (Karachentsev 1990), i.e. they are found, in general, at the edge of superclusters as can be shown with appropriate cuts in velocity (see also Fig. 12 at the beginning of the Atlas). In Fig. 7 we show the distribution, in cz and R.A., of the objects and indicate the main structure visible in the Southern Hemisphere as deduced from Fairall and Jones (1991). As noticed before, the bulk of the known redshifts come from large scale structure dedicated surveys as emerges from the figure. In fact, in the distribution of Fig. 6 it is possible to identify a peak in the area of Fornax ~1500 km/s, while to the peak around 3500 km/s contribute pairs in Centaurus, Hydra I region and in the Sculptor Wall. Pairs in Centaurus-Pavo and the extension of the Sculptor Wall contribute at the peak around 9000 km/s. On the other side, in Fig. 1 is already visible that the distribution of pairs is not uniform at brighter magnitudes where the influences of the near Large Structures can emerge with more evidence. In the ESO-LV catalogue is given for each galaxy a projected surface density, N(tot), i.e. the number of galaxies per square degree, inside a radius of 1 degree. The authors demonstrated that N(tot) can be an indicative measure of the true local volume galaxy density for N(tot) > 10 which means that the galaxy is located in rich clusters like Centaurus, Hydra, Fornax. When N(tot) < 10 the value is linearly correlated with the `fraction of galaxies in cluster' (Fig. 11 in ESO-LV).

FIGURE 7. Distribution of pairs and main nearby Large Scale Structures in the southern hemisphere. Starting from the center the dashed circles marks 5000 and 10000 km/s respectively. The information on the large scale structure is derived from Fairall and Jones (1991).

In Fig. 8 (right panel) is shown the distribution of the N(tot) density parameter for our best candidates sample. If different N(tot) values are reported in ESO-LV for each pair members we plot and report in the Atlas and Table 2 the larger one.

FIGURE 8. (left panel) Distribution of the velocity difference for the candidate pairs for which this measure is available. The distribution can be represented by an exponential function similarly to the one obtained in the CPG. The median of the velocity difference is 129 km/s which compares quite well with 120 km/sof CPG. (right panel) The distribution of the N(tot) parameter. For values larger then 10 the objects are in cluster. The arrow indicates that two pairs have values larger then 20.

Our best candidate pairs have a median value of <N(tot)> = 1.59 +/- 0.80. Following ESO-LV, the distribution indicates that less than 20% of the pairs could be members of rich clusters. Fig. 8 (left panel) also shows the velocity difference distribution. It can be approximated by an exponential law and the global median value is 129 km/s.


Pair members morphology


Dressler (1980) has verified that galaxy morphology depends on the environment where the galaxy resides: this is known as the morphology-density relation.
Since the N(tot) median value, previously discussed, is quite low the pair environment can be safely considered as intermediate between field and group. We can then compare the percentage of pair types with that fraction of morphological types derived from the study of previous environments.
Fig. 9 shows that, according to the classification given in ESO-LV, there is not a strong tendency in pairs to show a morphological concordance between members. This effect is also found by Soares et al. (1994) in their sample selected with criteria completely independent from ours. On the other side this contrasts with both Karachentsev (1990) and Yamagata (1990) studies which indicate that there is a predominance of pairs whose members have similar morphological type.

FIGURE 9. Comparison of morphological types for the members of P doubles having DeltaV < 1000 km/s. No morphological concordance between pair members is evident from the distribution. A large number of pairs is of mixed type (35%).

Sulentic (1990) computed the expected frequency of E+E, E+S, S+S random association from available compilations of isolated galaxies (Noerdlinger 1979, Gisler 1980). The expected rate from the field is 4% for E+E, 32% for E+S and 64% for S+S. Jerjen et al. (1992) in order to represent the global luminosity function (LF) of `extreme field' and `group' galaxies used a percentage of early-type galaxies respectively of 0% and 9%. So the expected fraction of E+E pairs has to be in this range.
Using the pairs of our sample which have DeltaV < 1000 km/s we obtain the following distribution: 14% of E+E, 35% of E+S and 51% of S+S. If then compared with the values expected there is an overaboundance of E+E with respect to both random association and even with the fraction in groups. Following Jerjen et al. (1992) it is comparable to that used to fit the luminosity function of Virgo cluster. The same effect is also noticed in the CPG. One of the possible explanation suggested is that E+E pairs represent an evolved phase of loose groups (Sulentic 1990). On the other side also the large fraction of E+S poses a challenge to current theories of galaxy formation which foresee for early and late-type galaxies quite different formation mechanisms. Rampazzo and Sulentic (1992) in some sense lightened the problem showing that a significant fraction of E+S are actually S0+S or S0+S0 with tails, erroneously classified as spiral arms.
An accurate morphological classification obtained from CCD high resolution images need, in any case, to be done in order to clarify the fraction of morphological concordant pairs and twin galaxies. This phenomenon is deeply connected with the effect we will describe in next paragraph.


The Holmberg effect


Holmberg (1954) suggested that there is a correlation between the integrated colors of the pair members. There are few systematic studies in the literature about the so called `Holmberg effect' and most of them obtained on a relatively small number of objects (Sharp and Jones 1980: 36 pairs; Demin et al. 1982, 1984: 40 S+S, 13 E+E). We studied the Holmberg effect on our pair candidates showing results in Fig. 10 and 11. We computed the color of each galaxy starting from the magnitudes in ESO-LV. Since in RC3 the galactic absorption was not available for all the galaxies, we adopted the simple model where the Galaxy absorption is described by a uniformly absorbing layer similar to that proposed in RC2 (de Vaucouleurs et al. 1976). The internal absorption has been evaluated using RC3 recipe (de Vaucouleurs et al. 1991) for the B magnitude while we followed Cardelli et al. (1989) in building the correction for the R filter. For the analysis we applied the Spearman rank correlation coefficient test giving in the Tables 3 and 4 the following columns: N the number of data on which the test has been applied, rs the Spearman rank correlation coefficient and t the Student's-t distributed significance parameter.

FIGURE 10. Study of the Holmberg effect. The (B-R)o colors of the members a and b are plotted one vs. the other for the whole sample for which colors are available (panel a). Pairs are then distincted into three classes according to their separation: contact pairs (panel b), intermediate (panel c) and large pairs (panel d).



TABLE 3: Spearman analysis applied on separation classes
SampleN rst
Total 3730.5111.32
sep <= (Aa+Ab) 2200.5910.80
(Aa + Ab) < sep <= 3*(Aa + Ab)105 0.394.35
sep > 3*(Aa + Ab)480.29 2.04


The total sample (Fig. 10, panel a) shows a positive correlation which do not ameliorate significantly if we consider contact pairs (X(1,2) <= A1+A2 ) but get worse when larger pairs are considered (see Table 3). In fact, while for the contact pairs the level of significance is well besides 3 sigma, for the wider ones it is not larger than 2 sigma. This can be explained in a simple framework in which the stronger interaction took place after the passage at the perigalacticon. A large noise is expected because on the ground of different effect: 1) some pairs may have already undertaken interaction, 2) others are viewed just before the interaction take place, 3) in some pairs the effect is long lasting depending on the population of stars which has been triggered.

FIGURE 11. The Holmberg effect using the P sample (panel a) is displayed also separately for the three different classes of pairs: S+S (panel b), E+S (panel c) and E+E (panel d). Largely significant is the correlation for E+E and S+S pairs.



TABLE 4: Spearman analysis applied on pair classes
SampleNrst
Total2750.52 10.04
S+S1480.54 7.75
E+S910.131.22
E+E360.69 5.62


We further studied the Holmberg effect using the P sample divided in the 3 subclasses of pairs E+E, E+S and S+S (see Fig. 11). The Spearman test applied to the three categories offers levels of significance well besides 3 sigma for E+E and S+S (see Table 4). Not significant appears the correlation for mixed (E+S) pairs, where the null hypothesis can be rejected with a 20% level of confidence.


DISCUSSION AND CONCLUSION


The large and growing concentration of observing facilities in the southern hemisphere suggests the convenience of extending homogeneously well studied samples of pairs, restricted to northern declination, in order to enrich the study of the subject. From this viewpoint a sample of candidate pairs has been selected from ESO-LV following the criteria used by K72 in selecting CPG. Our compilation, 409 candidates, has to be considered a partial complement to CPG both because it shows sign of incompleteness starting at 14.5 magnitudes and because reports candidate pairs south delta =-17 degrees. The above limits are inherited from the starting catalogue. We have assembled an Atlas of the images of our best candidate pairs. Images are derived from the DSS.
The isolated pairs are found at the edge of clusters or super-clusters since the selection criteria tend to eliminate pairs which are in dense environments.
41% of the candidate pairs have the velocity difference directly available from the literature. 11% of such pairs have a velocity difference larger than 1000 km/s and were considered optical alignments. We subdivided the pairs having DeltaV < 1000 km/s into three large categories: E+E (14%), E+S (35%) and S+S (51%). We do not see any overaboundance of morphologically concordant pairs, although a first analysis of CCD images and previous studies (see Rampazzo and Sulentic 1992) suggest that a not negligible fraction of ESO-LV morphological classification has to be revised.
Previous percentage of E+E confirms the presence of an overaboundance of early-type galaxies in pairs with respect to association deduced from the field distribution (Sulentic, 1990) and groups (Jerjen et al. 1992). Sulentic (1990) suggests that these pairs could be the debris of loose groups in which merging has played a dominant role. Also E+S are very well represented in the sample challenging galaxy formation theories unless merging is taken into consideration, since chance encounters which led to physical pairing are negligible.
The Holmberg effect presence has been studied for all the three classes of pairs in a significant number of objects with respect to the previous attempts in literature. While the color of the members correlate significantly for E+E and S+S, the correlation is less significant for E+S pairs, as found by Demin et al. (1984). Bergvall (1981b) shows that the effect is also present significantly in hierarchical systems, but the companion is found to be redder than the primary galaxy. Since there is a nearly monotonically decrease of colours with morphological type (see Roberts and Haynes 1994), different classes of pairs may ask for different explanations being the gas richness the dominant factor. In gas poor pairs (E+E), in fact, it could simple reflects the similar stellar population. The consistency of the hypothesis of merging origin for the Es in pairs has to be tested versus the concordance between the member's colors. In fact, during a merging event both stellar populations mixing and star formation are foreseen (Barnes and Hernquist 1991) leaving long lasting signs in colors, especially if the E type merger remnant comes from gas rich progenitors. Concerning gas rich pairs (S+S and at certain extent E+S) observations in many wavelength regimes show that galaxy interactions can enhance the rate of star formation, both for galactic nuclei and disks. Together with the optical evidence cited before, evidences comes from emission lines (Keel et al. 1985, Bushouse 1987, Kennicutt et al. 1987), radio continuum emission (Hummel et al. 1987) and FIR emission (Xu and Sulentic 1991 and reference therein). Keel (1993) has studied the connection between star formation, the galaxy kinematic and the encounter parameters. Direct percentage of young stellar population induced by interaction has been computed by de Mello et al. (1994) on Pair #76 using population synthesis model. All these facts suggest that the so called Holmberg effect may not be a simple consequence of the morphological concordance of the pair members. On the other side, we have shown that our pairs do not show a clear morphological concordance between members.
The proposed sample have to be refined at least obtaining missing redshifts. Even if not aimed to this purpose, a spectroscopic study of the pairs in the sample is going on using ESO telescopes.


ACKNOWLEDGMENTS


We are deeply indebted with the referee, Prof. Igor D. Karachentsev, for the careful reading of the manuscript and comments and with dr. Luiz Da Costa for having provided us redshifts from his survey before publication. We thank Prof.s A. Franceschini, F. Combes, W. Keel and J.W. Sulentic for useful suggestions and discussion during the preparation of the work. Mrs. De Gregorio have greatly helped us in typing tables. We acknowledge the kind hospitality of the CNR Istituto di Fisica Cosmica in Milan during the printing of the image material. One of us (LR) acknowledges the use of the facilities of the Osservatorio Astronomico di Brera in Milan.


REFERENCES


  • Arp, H.C. and Madore, B. 1985, A Catalog of Southern Peculiar Galaxies and Associations, Cambridge Univ. Press.
  • Barnes, J. and Hernquist, L. 1992, A&A Ann. Rew., 30, 705.
  • Barnes, J. and Hernquist, L. 1991, Ap.J., 370, L65.
  • Bergvall N., 1981a, Uppsala Astronomical Obs. Report, N. 19.
  • Bergvall N., 1981b, Uppsala Astronomical Obs. Report, N. 20.
  • Bushouse H.A., 1987, Ap.J., 320, 49.
  • Cardelli, J.A., Clayton, G.C. and Mathis J.S. 1989, Ap.J., 345, 245.
  • Charlton, J.C. and Salpeter, E.E. 1991, Ap.J., 375, 517.
  • Combes, F., Prugniel, P., Rampazzo, R. and Sulentic, J.W. 1994, A&A, 281, 725: CPRS94.
  • De Mello, D., Keel, W., Sulentic, J.W., Rampazzo, R. and White, R.III 1994, A&A, in press.
  • Demin, V. 1984, Soviet Astron., 28, 622.
  • Demin, V., Dibai, E. and Tomov A. 1982, Soviet Astron., 25, 528.
  • Demin, V., Zasov, A., Dibai, E. and Tomov, A. 1984, Soviet Astron., 28, 367.
  • de Vaucouleurs, G., de Vaucouleurs, A., Corwin, H.G. Jr., 1976, The Second Reference Catalogue of Bright Galaxies, Austin, Texas.
  • de Vaucouleurs, G., de Vaucouleurs, A., Corwin, H.G. Jr., Buta, R.J., Paturel, G. and Fouquè, P., 1991 The Third Reference Catalogue of Bright Galaxies,. Springer-Verlag.
  • Fairall, A.P. and Jones, A. 1991, Southern Redshifts Catalogue and Plots, Publication of the Dept. of Astron., Univ. of Cape Town, N. 11, South Africa: F
  • Gisler, G. 1980, A.J., 85, 623.
  • Heydon-Dumbleton, N.H., Collins, C.A. and MacGillivray, H.T. 1988, in Large Scale Structure in the Universe. Observational and Analytical Methods, ed.s W.C. Seitter, H.W. Duerbeck and M. Tacke, p.71.
  • Hummel, E., van der Hulst, J.M., Keel, W.C. and Kennicutt, R. C.., 1987, A&A Supp. Ser., 70, 517.
  • Karachentsev, I.D. 1972, Comm. Spec. Astrophys. Obs., 7, 1: CPG, K72.
  • Karachentsev, I.D. 1987, Dvoinye Galaktiki, Nauka Moscow.
  • Karachentsev, I.D. 1990, in Paired and Interacting Galaxies, I.A.U. Colloquium N. 124, Edited by J.W. Sulentic, W.C. Keel and C.M. Telesco, NASA, 3.
  • Keel, W.C. 1993, A.J., 106, 1771.
  • Keel, W.C., Kennicutt, R.C., van der Hulst, J.M. and Hummel, E., 1985, A.J., 90, 708.
  • Kennicutt, R.C., Keel, W.C., van der Hulst, J.M., Hummel, E. and Roettiger, K.A., 1987, A.J., 93, 1011.
  • Jerjen, H., Tammann, G.A. and Bingelli, B. 1992, in Morphological and Physical Classification of Galaxies, eds. G. Longo, M. Capaccioli and G. Busarello, Kluwer Ac. Publisher, pag. 17.
  • Lauberts, A. and Valentijn, E.A. 1989, The Surface Photometry Catalogue of the ESO-Uppsala Galaxies, ESO, Garching bei Munchen, Germany: ESO-LV.
  • Holmberg, E. 1937, Annals Lunds Astr. Obs., N.6.
  • Holmberg, E. 1954, Medd. Lunds Astr. Obs., 186.
  • MacGillivray, H.T. and Stobie, R.S. 1984, Vistas in Astronomy N.4, 433.
  • Malin, D.F. and Carter D. 1983, Ap.J., 274, 534.
  • Maurogordato, S., Proust, D., Balkowski, C. 1991, A&A, 246, 39: MPB
  • Noerdlinger, P. 1979, Ap.J., 229, 470.
  • Page, T. 1960, Ap.J., 132, 910.
  • Peterson, S.D. 1979, Ap.J. Supp. Ser., 40, 527.
  • Rampazzo, R. 1988, A&A, 204, 81.
  • Rampazzo, R. and Sulentic, J.W. 1992, A&A, 259, 43.
  • Rampazzo, R., Reduzzi, L., Sulentic, J.W. and Madejsky, R. 1994, A&A Supp. Ser., in press.
  • Roberts, M.S and Haynes, M. P., 1994, Ann. Rev. A&A, in press.
  • Rood, H.J. 1982, Ap.J. Supp. Ser., 49, 111.
  • Sadler, E. and Sharp, N. 1984, A&A, 133, 216.
  • Sharp, N. and Jones, B. 1980, Nature, 283, 275.
  • Soares, D.S.L. 1989, PhD Thesis, Univ. of Groningen.
  • Soares, D.S.L., de Souza, R.E., de Carvalho, R.R. and Couto da Silva, T.C. 1994, A&A, preprint.
  • Sperandio M., Chincarini, G., Rampazzo, R., de Souza, R.E. 1994, A&A Supp. Ser., in press.
  • Sulentic, J.W. 1990, in Morphological and Physical Classification of Galaxies, eds. G. Longo, M. Capaccioli and G. Busarello, Kluwer Ac. Publisher, pag. 293.
  • Thuan, T.X. and Seitzer, P.O. 1979, Ap.J., 231, 680.
  • Tully, B.R. 1988, Nearby Galaxy Catalogue, Cambridge Univ. Press.
  • Turner, E.L. 1976, Ap.J., 208, 20.
  • Yamagata, T. 1990, Paired and Interacting Galaxies, I.A.U. Colloquium N. 124, Edited by J.W. Sulentic, W.C. Keel and C.M. Telesco, NASA, 25.
  • White, S.D.M., Huchra, J., Lataham, D., and Davies, M. 1983: M.N.R.A.S., 203, 701.
  • Xu, C. and Sulentic, J.W. 1991, Ap.J., 374, 407.
  • Zhenlong, Z., Jiangsheng, C., Xiaoying, T. and Yulin, B. 1989, Pub. Beijing Astr. Obs., 12, 8.
  • Zwicky, F., Herzog, E., Wild, P., Karpowicz, M. and Kowal, C.T. 1968, Catalogue of Galaxies and Cluster of Galaxies, Caltech, CA U.S.A.: CGCG.