2.3. Radial Velocities
With the proliferation of optical and rado telescopes, the number of galaxies for which radial velocities V have been measured has risen rapidly. At the same time, the accuracy of such measurements has improved. Along with the general progress in this field, the number of double galaxies with measures of V has increased correspondingly.
At the time when our catalogue was published (1972), radial velocities had been measured for only about 10% of the double systems. Typical errors in such measurements were hundreds of kilometers per second which was not sufficiently accurate for estimates of galaxy masses. As an example we may consider the interacting pair, NGC 3786 / 3788. According to the Reference Catalogue of Bright Galaxies (de Vaucouleurs et al., 1976) the difference between the radial velocities of these two components is 410 ± 148 km/s. According to our observations with the 6-meter telescope, the relative velocity is 60 ± 2 km/s, which corresponds to a difference in the orbital masses of a factor of 46.
Completing the catalogue with new velocities proceeded episodically during the investigation of galaxies for various programs. Data from 1972 through 1976 appeared in the second version of the Reference catalogue. Radial velocities of selected double galaxies may be seen in the following important references: de Vaucouleurs and de Vaucouleurs (1976), Rubin et al. (1976), Peterson (1978), Arkhipova et al. (1976), Debye et al. (1976), Chincarini and Rood (1976), Arakelyan et al. (1976), Gudehus et al. (1979), Brown and Wilson (1976), Heckman et al. (1983), Barbon et al. (1979), Schild and Davis (1979), Sulentic and Arp (1983; Shectman et al. (1983), Giovanelli and Haynes (1985), and Hoessel et al. (1985). Particularly notable is the massive spectral survey of galaxies brighter than magnitude 14.5 performed by Huchra et al. (1983) and the compilation of 21-cm radio observations presented by Hoffmeier et al. (1983). This last publication includes very accurate radial velocities for 75 galaxies from the catalogue of pairs.
Comparison of radial velocities measured for one or another galaxy by various authors shows poorly understood divergences. From the data of Sandage (1978) the typical situation is that the rms difference in velocities (lower limit) exceeds by two or more times the internal errors of measurement. There are a number of factors that can lead to systematic errors in velocity measurements, of which the following are some of the more basic.
1. Blending of absorption lines in the spectra of galaxies with emission lines in the night sky. This is especially noticable in the spectra of galaxies of low surface brightness observed in blue spectral regions (Simkin 1977).
2. Disagreement between the mean velocities of galaxies measured via emission and absorption lines in the case in which the gaseous component of an object exhibits large, non-circular motions.
3. The structure of close interacting pairs, in which it can be difficult to distinguish the nucleus as a dynamical center from other bright condensations in the galaxy.
4. Inaccurate positioning of the spectrograph slit on objects with complicated structure.
5. Blending of the components of close pairs when their apparent separation is smaller than the angular resolution of a radio telescope.
6. Errors in the scale measurement on image tubes or other electro-optical systems frequently used for the registration of galaxy spectra.
7. Improper or missing corrections to the galaxy's velocity due to the orbital motion of the Earth about the Sun as well as the motion of the Sun within the galaxy.
If both components of a pair are observed simultaneously from the same spectrum the last two factors will not affect measurements of the radial velocity difference from which orbital masses will be derived. Therefore, measurements of the radial velocities of double galaxies done by the same author apparently have smaller errors than the rms errors of the individual measurements.
The first systematic measurements of radial velocities for double galaxies in the catalogue were taken by the author, together with Pronik and Chuvaev, with the 2.6-meter telescope at the Crimean Observatory (Karachentsev et al. 1974, 1975, 1976). The observations were performed with the three-stage image tube at dispersions of 360 Å/mm and 110 Å/mm. The accuracy of the resulting velocities did not appear to be sufficiently high and the authors observed many of these pairs later at the 6-meter telescope. Further radial velocities were produced with the 2-meter telescope of the Tautenberg Observatory (Afanasiev et al. 1975). These observations used two types of image tube with fibre optics giving dispersions of 190 to 250 Å/mm. Later, new measurements for 44 pairs were produced by the author, together with Sargent and Zimmerman (Karachentsev et al. 1979a), at the Cassegrain focus of the 5-meter telescope at the Palomar Observatory. These observations were performed using the 512-channel television detector in the spectral range 3700-5300 Å with a spectral resolution of 4 Å per channel.
The bulk of the radial velocities, about two-thirds of the catalogue, were obtained at the prime focus of the 6-meter telescope of the Special Astrophysical Observatory of the USSR Academy of Sciences (Karachentsev 1980a, 1981a, 1981b, 1983). Double galaxies were observed with the UAGS spectrograph with a UM-92 image tube with dispersions of approximately 90 Å/mm over the spectral range 5500-7500 Å. The scale of the spectrograph perpendicular to the dispersion is 17"/mm and a typical slit width was 1". Objects not showing emission lines were later observed in the spectral range 3500-5500 Å. For close pairs with angular separations less than the 90" slit length of the spectrograph, the slit was generally oriented between the centers of both galaxies. In the case of wider pairs the slit was often oriented along the major axis of each galaxy. These spectra were also used to measure the rotation curves in these systems.
Extensive work measuring radial velocities for double galaxies in our catalogue was presented by Tifft (1982). Tifft observed a total of 370 pairs with a 2.3-meter telescope at a dispersion of 240 Å/mm. In addition to measures of the radial velocity, Tifft's paper presents descriptions of the spectral characteristics and structure of the double galaxies. To describe the general character of the spectra of galaxies, Tifft used the following classification: A-absorption spectrum, W-the spectrum has weak emission, M-emission lines in the galaxy are moderately strong, S-the spectrum exhibits strong emission (such as found in Seyfert galaxies or most of the blue Markarian systems). We have adopted this simple scheme to classify the spectra of all double galaxies in the catalogue.
We return to the question of errors of measurement of radial velocities. The internal accuracy is limited by several factors, such as the spectral resolution of the observation, the number of spectral lines measured, and their contrast against the spectrum of the night sky. When measurements of the night sky can be used as references we find that, for galaxies observed at the 6-meter telescope, the mean internal accuracy v 40 km/s. To address the absolute (external) errors we consider the data for radial velocities taken with very high accuracy using 21-cm observations. From the compilations of Hoffmeier et al. (1983) and the paper by Sulentic and Arp (1983), we find radial velocities for 96 double galaxies in the catalogue measured with an accuracy no worse than 15 km/s. The results of comparing our measurements with these 21-cm data are summarised in the upper portion of Table 1. The first column gives the spectral type of the galaxy; the second, the number of objects; the third, the mean difference in the estimates of radial velocity; the fourth, rms errors; and the last one, the mean internal error of measurement = (K2 + 212)1/2. From these data the following conclusions may be drawn: a) there is no significant systematic error between the two kinds of measurements: the difference is somewhat larger for galaxies with absorption lines, b) the inferred absolute accuracy of the radial velocities compared to the internal accuracy, i.e., <V 2>1/2 / <2>1/2, decreases with increasing intensity and number of emission lines in the spectra of galaxies: the mean absolute error exceeds the internal error by a factor of 1.8.
Another possibility for comparison arose upon publication by White et al. (1983) of a set of radial velocities for double galaxies in the Turner sample, among which were 75 objects in our catalogue. The excellent accuracy of these measurements was produced by obtaining the spectra on a Reticon at very high signal-to-noise and by using cross-correlation analysis. The comparison of our data with White et al. is presented in the lower half of Table 1. Obviously the mean errors in the radial velocity and the absolute rms errors are smaller than those obtained from earlier methods. Further discussion concerning errors in velocity measurements and their effects on the orbital mass can be found in Karachentsev (1983). The overall distribution of radial velocity differences measured optically and in 21-cm for 171 galaxies is shown in Figure 2, where the vertical lines are restricted to objects of spectral types S and M. The distribution is close to normal and has a mean value close to zero.
Note that the difference between absolute and internal errors is reduced by considering the peculiar velocities of pair components (Karachentsev 1983). This supports the notion of correlated errors in radial velocity between the members of a single pair.
In conclusion we will note that the catalogue contains 302 galaxies whose radial velocities were measured both by Tifft and by the author. We will not present here a detailed comparison of these series of observations, but note only that our measurements (VK) agree with those by Tifft (VT) worse than with the data presented by White et al. (1983). For the entire sample, we obtain <VK - VT> = 28 ± 9 km/s and <V 2>1/2 = 154 km/s.
To illustrate the difficulties that sometimes occur when comparing velocity measures given by various authors, take the following case. The pair of galaxies NGC 5679 was noted as an interacting system in the Arp Atlas (A274) and by Vorontsov-Vel'yaminov (VV458). Peterson (1979a) gives a mean radial velocity 1638 ± 25 km/s. According to Arkhipova and Esipov (1979), the radial velocity difference between the components of this pair is small but the mean velocity was 3100 ± 60 km/s. According to our measurements the west and east components have radial velocities Va = 7447 ± 20 km/s and Vb = 8631 ± 30 km/s. The correctness of these values is supported by Tifft (1982), who obtained Va = 7524 km/s and Vb = 8670 km/s. The significance of such large differences in radial velocity is not clear, but fortunately the number of such curiosities among our systems does not exceed 1% of the sample (2) .
2 We have recently obtained new measurements of radial velocity for a number of double systems, primarily wide pairs, and for several double galaxies radial velocities have been measured by other authors. We limit ourselves here to presentation of radial velocities which appeared in the magnetic tape version of our catalogue (Karachentsev et al. 1985). Note, however, two cases of large differences in radial velocity. For component 3a (Giovanelli and Haynes, 1985) V0 = 8084 instead of the 7131 from our data. For component 291b, Huchra et al. (1983) give 2606 instead of our 6403. By performing further observations we have confirmed the newer values for radial velocity. Back.