From dressler@mso.anu.edu.au Tue Jul 25 15:58 EST 1995 Date: Tue, 25 Jul 1995 15:57:59 +1000 From: dressler@mso.anu.edu.au (Alan Dressler) To: heron@mso Subject: Dressler text Peculiar Velocities within R = 6000 km/s Alan Dressler Carnegie Observatories In introducing this meeting, Peter Quinn emphasized the importance of measurements of peculiar motion -- departures of galaxy velocities from smooth Hubble flow -- as a tool to investigate the degree to which light traces mass in the universe. I want to reiterate the point he made that much of what we think we know about large scale structure comes only from distributions of galaxies, so it is critical that we learn to what extent our conclusions are influenced, perhaps biased and misled, by the assumption that light traces mass in some relatively straightforward way. Peculiar motion studies are our most direct way to study this issue in the universe today, though gravitational lensing, and small scale anistropies of the cosmic microwave background (CMB), are valuable ways to study the distribution of mass at earlier epochs. We might ask simply, "Do galaxies trace the mass?," and try to choose from these conclusions: (1) faithfully; (2) in a (simply?) biased way; (3) in a gross statistical sense; (4) not at all. At present, I think we can only deduce, from our comparisons of the field of peculiar velocities within R = 6000 km/s to the distribution of galaxies in this volume, that light traces mass in a gross statistical sense -- we have at least eliminated (4). It will be the task of future work on peculiar measurements to improve accuracy and sampling to the point that we can test (1) and (2), perhaps even leading to an empirical derivation of what is the nature of biasing, if it exists at all. I. A Brief Review of the Data and Interpretation This field can trace its roots to 1976, when Vera Rubin and Kent Ford reported a large scale anistropy in the Hubble flow based on the hope that Sc I spiral galaxies make good standard candles. We can probably conclude from what they did that, unfortunately, they are not good enough standard candles for this task. The fact that their result remains unconfirmed reflects the large uncertainties in systematic biases that accompany large statistical errors: It is crucially important to use distant estimators with the smallest possible error, because these systematic errors can prevent reliable results that cannot be compensated by large sample size. In the same year the first unquestioned measurement of peculiar motion was made: the observation of the CMB dipole is interpreted, without doubt in my mind, as the Doppler motion of our own Local Group with respect to the cosmic reference frame. This 600 km/s velocity clearly does not arise locally (within R < 1000 km/s); our distance estimators are secure enough to show with certainty that no reflex motion is seen in nearby galaxies that would be seen if it were local. Rather, the fact that the peculiar velocity of the local group is not mirrored in the motions of other galaxies out to 2000 km/s -- that the frame defined by our Local Group is a good "rest frame" for other galaxies in this sphere -- should have alerted us early on that the CMB dipole arises on a much larger scale. The year 1976 is also a banner year in this field because of Brent Tully -- he was completing his work on the Tully-Fisher relation that became the workhorse distance estimator in the 1980's and 1990's. The high accuracy of this method, which gave distances good to 15-20% for late-type spiral galaxies, was the breakthrough that allowed the first serious study of galaxy motions on a large scale, the Aaronson et al. (1982) collaboration that mapped the velocity field around Virgo and the Local Supercluster. Here at Heron Island we are fortunate to have all the members of that breakthrough work, all save Mark Aaronson himself, in attendence. **T1** In a simple world, the Hubble flow would be perfectly smooth, reflecting the uniform density of matter (although if such were the case, there would likely be no one around to remark on it). In an equally ideal case, a spherical overdensity of matter would produce a smooth symmetrical "S-wave" in the Hubble diagram. **T2** Distance estimators like Tully-Fisher, and Dn-Sigma, allow us to make Hubble diagrams to look for such signatures of overdense and underdense regions. **T2a** Of course, the distribution of matter is likely to be very uneven and asymmetric, so it should be no surpise if the patterns we see are less clearly outlined. Even though the Virgo infall pattern found by Aaronson et al. is not as clear as the idealized version shown here, it does bear some resemblance, and it led to the expectation, expressed by Sandage and Tammann among others, that the next nearest supercluster -- Hydra-Centaurus -- would show an infall pattern that would account for the rest of the CMB dipole, a larger share of perhaps 400-500 km/s. **T3** It didn't turn out that way. The distance measurements for some 400 elliptical galaxies made by the Seven Samurai -- Burstein, Davies, Dressler, Faber, Lynden-Bell, Terlevich, and Wegner -- found no clear infall patter in the Centaurus direction, and, in fact, a large motion by galaxies in the Centaurus and Pavo-Indus regions, comparable with our own Local Group's motion. Indeed, many of the peculiar velocities in Centaurus far exceeded that of our own CMB-derived motion. In this early version of the Seven Samurai results, in which the peculiar velocities relative to the Local Group of the elliptical galaxies in the general vicinity of the Supergalactic plane are shown as arrows, it is clear that the reflex motion of the CMB dipole (the red vectors at lower right and upper left) are not evident in the data, which covers the region R < 4000 km/s quite well. In other words, the Local Group's motion is not generated principally within this sphere. **T4** The work of the Samurai can be summarized in three conclusions: (1) A large "bulk flow" for galaxies in the entire sphere R < 6000 km/s. Averaged over the whole volume, there is a typical velocity of order 350 km/s in the general direction of the CMB dipole. Note, however, that the flow is not by any means "laminar" -- there are clear signs of many subregions of substantial flows within. (2) In the direction of the CMB dipole large velocities continue out to at least R = 4000 km/s. There is no clear sign of a reflex motion. (3) For certain specific regions, most notably the Centaurus clusters, some very large peculiar velocities, more than 1000 km/s, were found. These conclusions, particuarly (2) and (3), led to a model for the flows that was called the "Great Attractor," which postulates the existence of an extensive mass overdensity beyond the Centaurus clusters whose center is relatively close to the Galactic plane, but whose influence extends far above and below it. This map of the Galaxy distribution by Ofer Lahav, which shows a hemisphere of the sky centered on the apex of large scale flow found by the Samurai, shows that there is a huge glow from hundreds of galaxies in this direction. This concentration is much more vast than the Centaurus or Hydra clusters (marked by H and C) with which it is often identified, and it is easy to see from comparison with the Virgo region (marked with a V) that velocities considerably in excess of the Virgo amplitude of 250 km/s might be expected. I want to emphasize, however, that the Great Attractor model was a model for the mass distribution in this region, based on peculiar velocities, and that correspondence with this map of the luminosity to see if mass is trace by light, has so far been done only in a rather gross way. This is because of many technical problems in applying corrections for bias in the peculiar velocities, and in transforming from maps in redshift space to real space, that accompany such data. **T5** These three conclusions of the Samurai have been confirmed by subsequent studies, using other techniques like Tully-Fisher. (I will, at the end of this presentation, discuss new efforts to use the technique of surface brightness fluctuations to derive more accurate values for peculiar velocities of elliptical galaxies.) In these Hubble diagrams from work by Faber and myself, the "front side infall" into the putative Great Attractor is shown in both spiral galaxies (star symbols) and ellipticals (open circles) that were done subsequent to the original Samurai sample. The diagram at right serves as a calibration in the sense that it shows that such large departures from smooth Hubble flow are not generally seen over the sky. As significant as the large frontside infall, however, which is typically 500-600 km/s, is the apparent end of this infall at a predicted distance of 40 h-1 Mpc. That is, the data scatter around the Hubble line at approximately this distance. Although our data also exhibit a backside infall beyond, data from other workers do not. Nevertheless, the reality of this return to the Hubble line, a return to the CMB rest frame, is critically inportant for identification of the peculiar velocities as due to a mass overdensity. **T6** These same features are seen in this cluster data from Han and Mould. Note the frontside infall and some very large velocities for the Centaurus clusters, and a return to the Hubble line for all the data beyond R = 50 Mpc. **T7** This next diagram is a compilation of the peculiar velocity data up to about 1990, color coded to show some of the features of the flow more graphically. The rest frame is now the CMB frame, so the vector of our own Local Group is seen at the origin (center of the diagram). The color coding is such that (remembering that there is noise in these measurements!) galaxies with peculiar motions from right to left are shown in red, while those moving from left to right are shown in blue. The coding within the cone of the dashed lines is different: galaxies moving toward the Local Group are shown in green, those away in magenta. The bulk flow shows up as the perponderance of red vectors over blue in the main zone (actually just a reflection of our own CMB motion and the lack of any detected reflex motion). One can also easily see the Virgo infall region, just above the intersection of the dashed lines, as the opposing patches of green and magenta. But, even more impressive to me, is the obvious quadrupole signature within the narrow cone: the dominance of green vectors shows a compression of the entire region that seems to fit well with the idea that much of the dipole signature is gravitational in origin. **T8** By 1990, then, the picture seemed fairly coherent: a general flow pattern within R=6000 km/s that could be ascribed to some large mass concentration in the general direction of the CMB dipole, probably just within the volume. However, around this time additional observations came in which began to challenge the simplicity of this view. One was Jeff Willick's observation, aided by data from Giovanelli and Haynes, of a bulk flow of the Pegasus-Pisces region in the direction of the Local Group. Although Willick's data, shown here as additional red and blue vectors, showed signs of an infall pattern in this region -- something one would expect -- the more surprising feature was the gross motion of about 400 km/s that seemed to be in the same direction as the previously discovered flow. This region is so far from any putative Great Attractor that it seems unlikely that it could be responsible. These results were later confirmed with the thesis observations by Stephan Courteau. **T9** The greater challenge, however, has come from Don Mathewson and his collaborators. The huge sample of 1355 southern spirals by Mathewson, Ford, and Buckhorn overwhelmed them small amount of data collected by Faber and myself. Although the new sample confirmed the general features of backside infall and a zero crossing at 40 h-1 Mpc that I have discussed earlier (however, the interpretation of this latter feature is disputed by Mathewson), no sign of backside infall was reported by this group. Instead, the flow seemed to continue at the 600 km/s out to the limit of this new sample. Combined with the Willick result, this suggested that perhaps there was no sign of gravitational attraction within the sample, but that the flow was truly a bulk flow over the entire span of R < 6000 km/s. **T10 In this next diagram I've compared the Mathewson et al. data to those of Dressler and Faber. Notice that both inside and outside the GA region (about a steradian of sky centered on the apex of motion found by the Seven Samurai), the data are in good agreement, with the exception of the many galaxies above the Hubble line beyond 60 h-1 Mpc in the Mathewson et al. sample. I will have little more to say about this today, but I believe that the Malmquist and selection biases that operate in this regime are very hard to deal with, and that we must be cautious in interpreting these features as real flows. We need better distance estimators to be sure what is happening at such great distances. **T11** I do, however, want to address the explanation for the "crossover region" that has been made, for example, by Josh Roth, exemplified in this diagram from the comprehensive review article written by Michael Strauss and Jeff Willick. Of course, it has been known for a long time that a very high overdensity, combined with 20% distance errors, will mimick a region of compression or infall. But, notice that such models require huge overdensities, like the factor of 25 used here, to achieve such an effect. This is unphysical, and certainly not supported by map of the sky in this region. Also, notice that such a model does not generate any peculiar motion below 2000 km/s inferred distance, exactly where the signal is largest in the real data. Even if one were to spread the mass overdensity across the region, for example, an extremely dense plane of galaxies at R = 5000 km/s, it would not reproduce the large peculiar motions observed below R = 2000 km/s. Besides, we would all be surprised, I think, if such a "wall" of galaxies were not to generate a strong infall pattern of its own. So, in the end, the explanation is, it seems to me, circular and nonsensical. **T12** 2. Have we Measured Something Sensible With some of these worrisome trends in the data, and the challenges to the techniques, like those raised by Joe Silk, we might well ask whether we can believe these basic results of peculiar velocity fields. I believe that the answer is a solid "yes." I've tried to list some of the arguments on these next two pages, with some supporting figures. **T13, T14** In this diagram I've tried to make a cartoon of what we have seen versus what I think some people prefer to believe -- that the large peculiar velocities in the GA region are due to some systematic effect in the data or the galaxies themselves. Were this true, we would be in the embarrassing (I think) position of believing that in this whole volume R < 6000 km/s our Galaxy and its neighbors would have the largest peculiar motions, with peculiar velocities falling off to zero outside -- a basically Ptolemaic view of the universe. I prefer the Copernican picture, that our Local Group has a more typical peculiar motion, and that we expect some bigger, and some smaller, within this volume. **T15** In this comparison of the POTENT map of the reconstructed density field, heavily smoothed, with the IRAS map of galaxies, smoothed to the same amount, we see good overall correspondence of the three main features of the local universe: two large overdensities and one void. The fact that these three general features are seen in both light and the mass distribution inferred from peculiar velocities -- two completely independent data sets -- is strong evidence that the peculiar velocities are reliable. **T16** Here again is the POTENT reconstruction, but done separately for spirals and ellipticals, using the Tully-Fisher and Dn-Sigma relations independently. The correspondence of these two maps again argues for the reliability of the peculiar velocity data: the results are consistent for these similar but significantly different ways of measuring peculiar velocities (for example, Tully-Fisher seems to depend on the dark matter halos of the spirals, but Dn-Sigma relates more to baryonic matter only). **T17** These diagrams are tests of possible environmental or global zero point shifts in the Seven Samurai elliptical galaxy sample. The diagram on the left is sensitive for errors in Galactic extinction, the one on the right, to variations in stellar population. No such trends are evident. **T18** This last such diagram shows a test that peculiar velocities are the result an environmental difference, like the density of the systems in which they are found. Raphael Guzman has found what he thinks is evidence for such an effect in a study of ellipticals in the Coma cluster core and periphery. It is clear here that, in the Seven Samurai sample at least, there is no correlation of the zero point of Dn-Sigma that might lead to large peculiar velocities as a function of density of the environment. **T19** Finally, I leave this subject with a more whimsical appraisal of whether the large-scale flows are real. **T20** 3. Distance Measurements to Ellipticals from Surface Brightness Fluctuations It is clear that in order to make a breakthrough in this field, we need not only more but more accurate measurements of galaxy distances for determining peculiar velocities. The surface brightness fluctuation (SBF) technique developed by John Tonry promises to be such an advance. John Tonry has already spoken about the method and our program, so I will just reemphasize that, unlike the Dn-Sigma or Tully-Fisher technique of estimating galaxian distances, the SBF techique has an explicit correction for stellar populations. A linear correction term in V-I colors appears, in theory and in practice, to be a sensitive and necessary correction for variations in both age and metallicity in the stellar population that produces most of the light. I believe that this is why SBF is such an improvement over Dn-sigma and Tully- Fisher, and why the accuracy for estimating distances with these two techniques is unlikely, by the addition of other as yet undiscovered parameters, to be reduced to the SBF level of 5-10%. That is, these other relationships, because they measure the luminosity -- the first moment of the light distribution, are subject to variations caused by age or metallicity in the stellar population. SBF, on the other hand, by being a second-moment measure of the typical stellar magnitude that contributes most of the light, is not insensitive to these matters. For example, if an elliptical galaxy contains a signficantly younger population than another, the brighter expected magnitude of the typical star producing the light will be well correlated with its bluer V-I color. In fact, it is remarkable that Dn-Sigma and Tully-Fisher work as well as they do -- this alone indicates that the range in stellar populations is not as great as it could be. But it also seems to suggest that this is the difference between SBF and these two techniques, and why the latter can not be expected to be as accurate as SBF. For the last five years I have been carrying on an extensive program of SBF measurements using the 2.5-m du Pont telescope at Las Campanas Observatory. These observations are part of a collaborative program with Tonry, John Blakesee, Ed Ajhar, and Gerry Lupino. For my part, I have collected data for upwards of 100 galaxies. I have paid special attention to the more challenging observations in the Great Attractor region -- galaxies in the Centaurus clusters and those partipating in the more general frontside infall. I have used the good seeing at Las Campanas to collect dozens of observations of galaxies in the 20-40 Mpc range. These have typical exposure times of 2-3 hours, with seeing that ranges from 0.6 - 1.0 arcseconds. A special effort has been made to make multiple observations in a variety of seeing conditions, and to observe more than one member of a group, to investigate the accuracy of the data. I report here on 43 observations of 31 galaxies. The predicted distances given here are preliminary -- there is significant uncertainty in zero points and colors that will be addressed in the near future, so that the values here could be uncertain to systematic effects at the level of up to 10%. Those tempted to use these data as if they were final numbers are hereby warned against it! Here I show the comparison of predicted distances (in kilometers per second, assuming a Hubble constant of Ho = 85) of galaxies with more than one measurement in the sample at hand. The small numbers by the solid dots record the seeing of the observation. The SBF measurements are extremely senstivie to the seeing when the limit has been pressed to, or beyond, its limit. This is not surprising, since not only is the signal declining rapidly as the smoothing by seeing increases, but the ability to properly correct for point sources -- Galactic stars, globular clusters, and field galaxies -- is greater hampered by poorer seeing. It is significant, then, that these measurements show excellent repeatability even when the seeing is substantially different. Comparison of these multiple measurements suggest an accuracy per measurement of 8% or better, perhaps as good as the best SBF measurements, about 5%. In this diagram we see a particular example, a pair of observations of NGC 3087. These target-like zones show that the SBF measurement of mIbar, the characteristic I-band magnitude of the stellar population, can be done independently for many different regions of the galaxy. For example, the excellent seeing observation on the top produced 7 measurements of mIbar that are in excellent agreement, this despite the fact that the galaxy intensity falls over this range by more than an order of magnitude and the sky contribution goes from unimportant to dominant, as shown in the following transparency. When one further considers how the contribution of point sources changes over these zones, it is a remarkable demonstration of the robustness of the method that the derived distance moduli will all agree to an accuracy of 5%. This kind of multiple measurement of a single galaxy is something Dn-Sigma or Tully-Fisher cannot do. Note that, under poorer seeing conditions (the observation shown on the bottom), the inner three zones are in excellent agreement with each other and with the better observation, but a systematic error has crept in for the outer zones, though the measurement has apparently not "fallen apart." In the data I am presenting, only measurements from the three inner zones have been used. **T22, T23** I've used these data to compare the peculiar velocity predictions made by the Seven Samurai using Dn-Sigma with the predictions of SBF measurements. The agreement is not perfect, and, because of the preliminary nature of the SBF calibrations it is too soon to evaluate how good it is. However, there is a clear correlation: the SBF distances and Dn-Sigma distances are in reasonable accord considering the errors, as indicated at lower right. This diagram addresses the claim made by Djorgovski that different methods of measuring peculiar motion get different results, specifically a comparison of Dn-Sigma and Tully-Fisher measurements to groups and clusters. Considering the correspondence of SBF and Dn-Sigma measurements shown here, the Djorgovski comparison may be telling us more about the spatial segretation of spirals and ellipticals that are allegedly in the same groups and clusters than it does about the reliability of measurements if peculiar velocity. This is an interesting scientific matter related to galaxy formation, so it certainly bears a heavy scrutiny in the future. **T24** The Hubble diagram composed of these 31 galaxies bears a strong resemblance to the Han and Mould figure I showed above. There is a sequence defining the smooth Hubble expansion (the sample includes galaxies outside the GA region), and a spattering of points above the Hubble line. Most notable is the large peculiar velocities found for the Centaurus clusters, Cen 30, Cen 45, and the I3370 group. **T25** These data can be plotted in the familiar vector diagram, with the corresponding diagram from the Seven Samurai measurements of the same galaxies. Note the similar results: small peculiar motions for most galaxies, but large velocities for the Centaurus clusters. A remarkable difference in the new data, compared with the Dn-Sigma results, is that all of the Centaurus galaxies, in all three subgroups, coalesce at the same distance, within the errors. This is an example of what we frequently see, and should expect, with the SBF technique, which is accurate enough to make "fingers of God" disappear. When questioning the reality of these measurements, it may be relevant to consider that for these clusters it is the very-accurately-measured radial velocities that scatter widely, while the distances all agree very well. This to me is a confirmation of the reliablity of SBF. **T26** Two years ago, at the IAP meeting on Cosmic Velocity Fields, I reported early results of SBF measurements of these Centaurus clusters. Here I am adding some new data, but the numbers are still preliminary because the observing runs have yet to be properly tied together photometrically. As can be seen in the following table, the agreement of the distances in the groups, with many multiple measurements combined to make these numbers, is remarkably good. In fact, all 12 galaxies are consistent with being at the same distance, a predicted velocity of 2302 km/s, with only an 8% scatter! This is a result somewhat different from the Seven Samurai's data, which had a significantly larger distance (though still high peculiar velocity) for Cen 45. However, in some ways the new data may make more sense: these large peculiar velocities are more probable in a situation where Cen 30 and Cen 45 are very close to each other. This is the model originally proposed by Lucy and collaborators. **T27** I've tried to sketch out the only plausible explanation I can come up with in this last transparency. It suggests that the I3370 group samples the average peculiar velocity in this region in space, which is a wopping 1300-1400 km/s. Alongside of this group are the Cen 30 and Cen 45 clusters, which are projected on top of each other. If they are moving with the same bulk velocity of 1300-1400 km/s, but also falling together, perhaps that would explain the extraordinary velocity of Cen 45, which I determine to be almost 2500 km/s. It would require a mutual infall of Cen 30 and Cen 45, with a mass ratio of something like 3:1, to achieve the measured peculiar velocities. Perhaps this is what happens when a rich cluster is being assembled from pieces; after all, such velocities are not excessive for rich clusters. On the other hand, one can't forget that all three systems are also moving at over 1000 km/s, a much greater peculiar velocity than is found on average in this part of the sky, so the cause for this great disturbance in the velocity field remains a mystery. If these results are correct, it may be a mistake to associate the truly extraordinary velocities in Centaurus with the Great Attractor, which was postulated to explain the lower-amplitude, overall flow. To achieve this very large anomaly might require some caustic in the mass distribution that is something quite different, one that may be one of our first clues that, in some regions, galaxy light does not accurately trace the distribution of mass. **T28** 4. Summary Many methods have now confirmed the basic features of the velocity field discovered by the Seven Samurai: signifcant bulk flow over the sphere R < 6000 km/s; a larger amplitude flow in the general direction of the CMB dipole; even larger velocities for the Centaurus clusters. I believe these data and interpretations are robust for the region R < 6000 km/s, though at greater distances systematic errors may be substantially influencing our results. Distances to early-type galaxies dervied using the surface-brightness-fluctuation technique promise to greatly improve our ability to map the dark matter distribution within 3000 km/s and, eventually, with better ground-based observations and the Hubble Space Telescope, to establish a much more reliable picture of flows on the R < 10,000 km/s scale.