QUASARS, REDSHIFTS AND CONTROVERSIES QUASARS, REDSHIFTS, AND CONTROVERSIES Halton Arp Interstellar Media 2153 Russell Street Berkeley, CA 94705 Interstellar Media 2153 Russell Street Berkeley, CA 94705 Copyright © 1987 by Halton Arp Reproduction or translation of any part of this work beyond the limits of Sections 108 or 108 of the 1976 United States Copyright Act without the permission of the copyright owner is unlawful. Requests for permission or further information should be addressed to: Permissions Department, Interstellar Media, 2153 Russell Street, Berkeley, CA 94705 Library of Congress Catalog Card Number 87-080290 ISBN 0-941325-00-8 Printed in the United States of America 10 9 8 7 6 5 4 3 2 1 TABLE OF CONTENTS Introduction 1 1 Distances of Quasars 7 2 The Battle Over Statistics 17 3 Galaxies Visibly Connected to Quasars 31 4 Certain Galaxies with Many Quasars 47 5 Distribution of Quasars in Space 63 6 Galaxies with Excess Redshift 81 7 Small Excess Redshifts, the Local Group of Galaxies, and Quantization of Redshifts 107 8 Correcting Intrinsic Redshifts and Identifying Hydrogen Clouds Within Nearby Groups of Galaxies 115 9 Ejection from Galaxies 133 10 The Sociology of the Controversy 165 11 Interpretations 173 Glossary 187 Index 193 PREFACE The purpose of this book is to present important information about the nature of the universe in which we live. Knowledge of the laws of nature offers humankind the only chance of survival in a changing environment. It endows us with the power to achieve whatever we consider our most desirable evolutionary goals. Perhaps most of all, the search for knowledge gives expression to a basic curiosity which appears to be the salient defining characteristic of human beings. The information about the physical universe that this book tries to convey is highly controversial. Since I believe that the facts are true and important, and since I have firsthand knowledge of the observations, I have undertaken to present the subject in the following book. Actually, this offers the only possibility of discussing this subject in a meaningful way at this time. The reasons for this are the following: First, the antecedent observations have been published over a span of twenty years in various technical, astronomical journals. In order to construct a coherent picture, these reports need now to be drawn together and related to each other. In the past, it has al- ways been possible to criticize or ignore individual discoveries and avoid the weight of accumulated evidence which a minority of astronomers have felt requires a drastic change in current assumptions about the universe. This book presents an integrated picture of this evidence which it is hoped will be compelling enough to establish the necessity for a new and large step forward in astronomical concepts. Secondly, there have recently been attempts by a few people in the field to suppress new results that disagree with their particular viewpoint. Telescope time needed to follow up discoveries in these directions has been denied. Research reports to journals have been rejected or modified by referees committed to the status quo. It is clear that when scientific results are prevented from appearing or being discussed in standard journals, the only alternative is to publish a book. Then no one who is interested is denied the opportunity of reading the evidence. Since many generalist readers are also interested in this subject, I have tried to write in a way that is comprehensible to nontechnical readers. Then no one, specialist or generalist, will be denied access to this new information if they wish to make use of it. I believe in order to gain the most fundamental knowledge of which we are capable it is necessary to continually and sincerely question our assumptions and test our theories. In a sense, the way we do science is more important than the exact results at any given moment. I have stated the results as correctly as I can in this book but one must always face the possibility that one's current understanding is more or less completely wrong. So, even if my thesis were mistaken—which I consider unlikely in view of the evidence—it may still have been valuable to have discussed how the process of astronomical discovery is actually conducted. The most important thing for us to recall may be, that the crucial quality of science is to encourage, not discourage, the testing of assumptions. That is the only ethic that will eventually start us on our way to a new and much deeper level of understanding. INTRODUCTION Redshifts and the Hubble Law In 1924, Edwin Hubble demonstrated that the small, hazy patches of light we see in the sky on a dark night—the galaxies—are really enormous islands of billions of stars, like our own Milky Way galaxy seen at a great distance. Study with large telescopes revealed that the fainter and smaller a galaxy appeared, the higher, in general, was its redshift. Redshift describes the fact that the characteristic lines in its spectrum due to hydrogen, calcium, and other elements appear at longer (redder) wavelengths than in a terrestrial laboratory. This effect was most simply attributed to a recession velocity of the emitting source—like the falling pitch of a receding train whistle. It was therefore concluded that the fainter and smaller the galaxy, the more distant it was, and the faster it was flying away from us. This is the velocity interpretation of the redshift-apparent brightness relation, the standard interpretation of the so-called Hubble law. About this time, Einstein was writing equations that attempted to describe the behavior of the entire universe, the totality of what exists. His equations pointed to its probable instability. Gravitation was either strong enough to be in the process of contracting the universe or too weak to prevent its expansion. In view of the extant conclusions about galaxy recession velocities, it was natural to interpret them as due to expansion of the universe. Extrapolating these velocities back to an origin in time gave rise to the concept of the universe being born in a primeval explosion, the so-called "big bang" cosmology. If this simple theory could explain all the observations, as it appeared to do for many years, then it would be what people strive for—an elegant solution. But it is my thesis in this book that from 1966 onward, observations began to accumulate that could not be explained by this conventional picture. Some extragalactic objects had to have redshifts which were not caused by velocity of recession. At the very least, it seemed that some modification had to be made to the current theory. The reaction against these discordant observations among some influential specialists was very strong: It was said that they "violated the known laws of physics" and therefore must be incorrect. Alas, it seems that in the intervening years the useful hy- Introduction pothesis had become enshrined dogma. Translated plainly, this dogma simply states: "At this golden moment in human history, we know all the important aspects of nature that we will ever know. In spite of a long record of fundamental revolutions in human thought, there are now no surprises, there is now an end to this history." This seems patently absurd. In fact, like most people, I would instinctively have a kind of temporal Copernican view that no arbitrary epoch is special. I believe that startling forward leaps in knowledge will continue to take place if the human race survives long enough. Nevertheless, I do not believe we can use extrapolation of past experience as a proof that this is taking place at any given moment. We must instead have concrete, specific evidence if a radical change is now needed. Therefore, my major goal in this book is to gather together all the proof existing at this moment, to show that there is massive, incontrovertible evidence for important phenomena and processes, perhaps even new forces or laws, which we cannot currently understand or explain. In order to force a change in the current paradigm, this book presents observations in detail. Unlike the current belief in the field that observations can be accepted or discarded according to whether they fit a theory, I submit that the observations in fact are the known laws of physics. It only is required for us to connect them together in some satisfactory way. In this process, however, the one thing we must guard against is the misinterpretation of chance occurrences. (Throughout history this has been a popular activity called superstition.) Therefore, in each of the many examples I discuss, I first try to establish that if the observation is not a chance occurrence, then a fundamental "law" has been broken. For example, if a high-redshift object occurs close by in space to a low-redshift object, then the redshift-distance relation has been violated. This reduces the proof to a yes or no decision. Either the closeness on the sky is an accidental projection of background and foreground objects, or there is a real physical proximity. We can then concentrate on determining how small the chance is that this observation is an accident. Even more important, we can determine whether there are other independent observations that support and confirm the reality of this result. Ultimately, each reader has to make his own decision whether the conventional law has been contradicted. But if we can gather enough different kinds of examples, as this book attempts to do, then perhaps we can achieve agreement that the phenomena are real. Then perhaps we can begin to discern repetitive patterns or similarities in the evidence that can be used to predict further examples and induce new hypotheses which could explain the phenomena. We will be underway into a new era. The first shock to conventional theory came with the advent of radio astronomy and the discovery of quasars. Therefore, we should first define the terms relating to these subjects. Radio Galaxies and Quasars The story of this new branch of astronomy begins for me on a bluff overlooking the ocean on the east coast of Australia, near the end of the Second World War. The operator of a radar station noticed a source of radio noise in the sky that rose 3 minutes and 56 seconds earlier each day. An astronomer immediately recognizes that this is just the amount the stars gain on the sun each day (sidereal time runs "faster" than solar time). This motion marked the source of radio noise as belonging to the realm of the stars and galaxies. That original observer, John Bolton, went on to help found the science of radio astronomy and direct some of the early radio observatories. The original radio source was in the southern constellation of Centaurus Introduction and became known as Cen A. Eventually, thousands upon thousands of cosmic radio sources were discovered. Many of them were eventually identified with disturbed or active galaxies like the galaxy responsible for Cen A. The radio emission is caused by charged particles moving in a magnetic field. Both the charging of the particles (their ionization) and their motion is a result of high temperatures and energetic events. These discoveries have required a fundamental change of concept about galaxies, a change that I believe has not yet been fully appreciated. It is no longer possible to view galaxies simply as relatively quiescent aggregates of stars, gas, and dust, all swirling in some majestic, ordered rotation about their center. Some are ripped asunder by huge explosions. Many have nuclei that vary strongly in brightness and intermittently eject quantities of matter outwards into space. In some ways, galaxies individually are reminiscent of the model of the universe as an exploding, or unfolding "cosmic egg." In my opinion, these new facts about the active nature of galaxies have not yet been integrated into a coherent picture of galaxy creation and evolution. But among radio sources that were identified with visible objects, an even more mysterious class than radio galaxies was found. These were the quasars. Optically they looked like point sources of light—like stars— hence their name "quasi-stellar" radio source, a term soon shortened to quasar. The first of these objects was identified by Allan Sandage and Thomas Matthews in 1963 in a collaboration between an optical and a radio astronomer. Then Maarten Schmidt, an astronomer at Caltech, found the key to the spectrum by showing the initially puzzling lines were those of familiar elements but shifted very far to the red. This was the shock. Why, when the highest redshifted galaxies known had maximum redshifts of 20 to 40 percent the velocity of light, did these stellar-looking objects suddenly appear with redshifts of 80 or 90 per- cent the velocity of light? It was briefly considered whether some other mechanism than velocity of recession could be responsible for quasar redshifts. For example, redshifting (which is equivalently a loss of energy of ~\ photon) might by caused by a very strong gravitational field. Such explanations were quickly discarded, however, and it was decided that quasars were the most luminous objects in the universe, seen at such great distances that the expansion of the universe was giving them the largest possible recession velocities. Difficulties were encountered almost immediately. In the first place, how could an object be so luminous? There was a problem of creating so much energy from known kinds of galaxies. Then, the calculated density of charged particles was so high in some quasars, that there was a problem of actually getting the photons, by which we see the objects, out from the interior. Then, very accurate positional measures by radio telescopes (very long base-line interferometry) revealed the astounding fact that some quasars appeared to be expanding with up to ten times the velocity of light. This was a flat-out violation of the known law of Einsteinian physics that the speed of light is a physical constant that cannot be exceeded in nature. Rather than move the quasars to lesser distances, which would give quite modest expansion velocities, the conventional theorists set up a small industry for rationalizations. They explained, by extremely complicated models, the faster-thanlight expansions as an illusion caused by very special, assumed conditions such as ejection toward the observer at nearly the speed of light. They, of course, ignored the direct evidence that the quasars were associated with galaxies which were much closer to us in space. When enough quasars were identified over the sky it became clear that anomalies also existed in the increase of their numbers with apparent faintness. For constant space Introduction density their numbers should have increased in proportion to the increased volumes enclosed at successively fainter apparent magnitudes. What was observed was completely different. This gave rise to another "gee whiz, isn't the universe wonderful" explanation. In this case, it was concluded that as we look out in space, and therefore back in time, we encounter a higher and higher density of quasars until suddenly—at a certain point—the quasars ceased to exist! In the present book, however, I do not debate whether this peculiar evolution of quasars is a priori improbable. I try to concentrate on the hard evidence of what they actually are and where they are located in space. In that process, we come again and again to observational evidence that redshift is not a good indicator of distances for quasars. Now the debate takes a curious turn. The conventional wisdom says quasars are just abnormal galaxies (superluminous, etc.) and that galaxies can only have redshifts caused by velocity. I say yes. I, myself, pointed out originally that quasars are physically continuous with galaxies. But a large body of evidence now exists showing that galaxies also can violate the redshift-distance relation. In fact, it is just the most peculiar galaxies, those most like the quasars, for which the most compelling evidence for nonvelocity redshifts exists. This has two consequences: First, it enormously strengthens the case that the redshiftdistance law can be broken. After all, it only requires one well-proven, discordant case of quasar or galaxy to establish that an additional cause of redshift—other than velocity—must be in operation. Because of the connection of quasars with galaxies we now have many, interlocking proofs of the phenomenon. But secondly it means that the mechanism for causing this nonvelocity shift must be capable of operating on an entire extended assemblage of stars, gas, and dust. This is much more difficult than finding a mechanism to operate on the more compact, more mysterious quasars. The advance in understanding required to explain these observations has been thereby considerably escalated and now represents a spectacularly exciting challenge. The stakes in the theory game have been sharply raised. The following book reflects the progression of evidence just discussed. The first five chapters deal with the anomalous evidence on quasars. The next three chapters deal with evidence for nonvelocity redshifts in galaxies. The final chapter combines this evidence and tries to explore the possible types of explanations which might account for the discordant data. Throughout the book, however, the reader will also be aware of many comments and anecdotes which bear on how the participants in the controversy have conducted the debate. The next-to-closing chapter deals with how I feel the acrimony arose in the debate and what it means for science. One of the reasons for this commentary is that the reader needs to be aware that many professional astronomers do not believe that there is any need to change the conventional theory. Some accomplished and noted professionals do believe important, fundamental changes are necessary. A number are waiting to make up their mind—or see what others decide. But a number of astronomers vehemently reject the observations or the conclusions from them which are presented in this book. There are only two possibilities: One is that the conventional wisdom is right and that the observations are meaningless accidents. The other possibility is that the present thrust of the observations is correct and that some radical changes will have to be made in current theory. This raises two questions for the reader: The first is, "What are the reasons which some astronomers give for disbelieving the evidence and conclusions of this book?" As Introduction to this question, I cannot possibly represent fairly the other side. Even if I could present it in an unbiased way, it would take an impossibly long time. The counterarguments to the evidence presented in this book, when they have been made in a few cases, are exceedingly complex and obscure and become hopelessly lost in technical detail. As Fred Hoyle has remarked, the establishment defends itself by "complicating everything to the point of incomprehensibility." I try to deal with valid, alternate possibilities as they arise, but the reader will either have to search out any original arguments from the references I give in the appendices to each chapter, or wait for a comprehensible rebuttal to this book to be published. The second question raised is: "Assuming for the moment that the evidence in this book is correct, why have many professional astronomers disbelieved it?" That is an exceedingly important question because it bears on how human beings discover and gain knowledge and avoid harmful, entrenched mythologies. That is the reason I have included personal anecdote and commentary in this book. In case the thesis of this book is correct, we want to know what the factors are that led to this long, implacable rejection of new knowledge, the wasted effort, and the retardation of progress. Inevitably these factors involve emotional, personal, and ethical questions. These are explosive subjects down through the history of mankind. I am sure emotions will be stirred as a result of my comments in this book. That I am willing to endure because I feel that the way in which research is conducted is one of the most crucial of mankind's activities. If the research is imaginative and accurate, and the human relations promote and protect this process, then the results will inevitably be worthwhile. If the process is biased, the practitioners too hostile or competitive for personal gain, or lacking in the crucial element of sportsmanship, then the results will inevitably be delayed and distorted. The reader will have to make up his own mind on both questions, the correctness of the thesis of this book and also the validity of the comments on why some people tried to reject and suppress the results. Each reader will have to make up his own mind on these two questions, as in most important matters in life, on the basis of the evidence he has seen or heard. Introduction DISTANCES l OF QUASARS The opposite page shows a photograph of knowledge toward the end of this book, but three quasars closely grouped around a first, let us just follow for a while the thread of large galaxy. The chance of these three qua- one particular story, the history of the claimed sars accidentally falling so close to a galaxy is association of quasars with galaxies. between 10"5 and 10~7, that is about one In 1966 while checking galaxies in my chance in a million. This is an enormously in- newly completed Atlas 0/ Peculiar Galaxies, I teresting observation because the quasars, noticed that radio sources, including some with high redshifts, are conventionally sup- quasars, fell close to, and aligned across, some posed to be far behind, and unrelated to the of the particularly disturbed galaxies. Quasars galaxy which has a much lower redshift. Nev- had been just discovered in 1963 and already ertheless, there was an attempt to suppress were being hailed as the most distant objects the discovery and observation of these qua- visible in the universe. If they were associated sars. When finally submitted to the Astrophys- with relatively nearby galaxies, such as in the iced Journal, publication was held up nearly Atlas, however, they would themselves have IV2 years. An anonymous referee stated, to be relatively nearby. Some explanation for "The probability arguments are completely their high redshifts would have to be found spurious." other than expansion of the universe at large What is the truth about this matter? Are distances. In March 1966, I gave the evi- the quasars related to the galaxy or not? And dence for these closer quasar distances in a why the emotion, intrigues, and deadly pro- colloquium at Caltech. I was told that one of fessional combat which the subject has in- the audience, later to become a vociferous op- spired for the last 20 years? To answer these ponent of the local hypothesis, had remarked questions, I believe, gives insight into the in a characteristically loud and over- state of knowledge in astronomy today and confident voice before the start of the confer- also illuminates the passions, prejudices, and ence, "Well, this will be the shot heard power relations in a modern science. We can around the room." In contrast, after the collo- explore the consequences of this for human quium, Fred Hoyle came up to the lectern Distances of Quasars 7 and said, "Chip, we did not know about your results, but Geoffrey Burbidge and I have just submitted a paper which comes to the same conclusion, namely that quasars need not be the most distant objects, but could originate from nearby galaxies." But, in retrospect, it seems that the remark made that day, even before the observations were presented, signalled the implacable opposition of those who had accepted the original assumptions about quasars. The first independent test of the newly found association between radio sources, quasars, and certain types of galaxies was published in Nature magazine in 1966 (see Figure 1-1. Three quasars close to the galaxy NGC1842. The chances are about one in a million o{ finding this association accidentally. Appendix for reference). It concluded that the chance of obtaining the observed associations between peculiar galaxies and radio sources arranged randomly on the sky was about 1 in 100. But curiously, the authors took the standpoint that since the chance of being accidental was only 1 in 100, that there was no need to accept the significance of the association and test further the current assumptions. This attitude is distilled into the aphorism, "In order to make extraordinary changes in accepted scientific assumptions, one must have extraordinary observational evidence." Unfortunately, we will see that experience suggests that what this expression has come to mean in 8 Distances of Quasars practice is: "In order to make basic changes in conventional assumptions, there is no evidence which is extraordinary enough". Naturally, this last attitude cuts the very foundation from under science. It will be of great importance for us to see, in the following pages, whether the evidence really is strong enough, and if it is, whether extragalactic astronomy has a hope of once again becoming a science. Quasars Associated with Galaxies In the years since 1966 some few astronomers produced many investigations that purported to demonstrate the association of quasars with low-redshift galaxies at high levels of significance. A few papers appeared attacking these conclusions and much private opinion was circulated that the associations were meaningless accidents. The reaction of the rest of the field seems to have been not to ask which of this published evidence was correct, but only to fall back on mutually spoken reassurances that the association was not accepted. Among the parade of papers demonstrating association, however, one particularly startling result emerged in 1979, namely that there were three quasars projected near the edge of the spiral galaxy NGC 1073. This was the first example of multiple quasars very close to galaxies, which, of course, would be very much less likely to occur by chance than single quasars close to galaxies. Figure 1-2 shows this very beautiful barred spiral with the quasars measured by myself and Jack Sulentic marked by arrows. Ironically, the galaxy was originally photographed by Hubble in 1950, and is featured in the Hubble Adas of Galaxies. This situation enabled me to make the pointed joke to my friend Allan Sandage, the author of that Atlas, and the co-discoverer of the first quasar, that his catalog of nearby galaxies seemed to contain many images of quasars well before they were discovered. The probability that three quasars would be observed by chance so close to NGC 1073 is about 2 x 10~5 or about 1 chance in 50,000. But NGC 1073 appears so large and bright that it is one of only 176 galaxies in the Hubble Atlas and one of only 1246 galaxies in the Shapky-Ames Catalog of Bright Galaxies. Of course, the vast majority of these have never been searched for the presence of nearby quasars. It was only by curiosity after a small radio source was discovered near the galaxy that I looked for other quasars nearby and found a total of three close to NGC 1073. Therefore, the previous prediction that quasars fell close to nearby galaxies had been confirmed with an extremely high probability of significance by this system. Even if no more cases like this were found among bright galaxies, it would still be a significant confirmation of the hypothesis that some low redshift galaxies have associated quasars. Usually in science one would expect an observation such as this to lead to the acceptance of the predicting hypothesis. In this case, the reality of the quasars was checked by some Caltech astronomers at Mt. Palomar Observatory, who refused to publish their confirmation and ignored the result. Confirmation was eventually published by the University of California astronomers E. Margaret Burbidge, V.T. Junkkarinen, and A.T. Koski. What the observation, and others like it, did lead to was a written warning from the allocation committee for Palomar telescope time threatening to cut my telescope time unless I refrained from such observations. In 1984, my observing time at Palomar was terminated. But the quasars sit there, apparently in the outer arms of NGC 1073. From where did they originate? What can we learn about the nature of these mysteriously redshifted sources of energy if they indeed are intermingled in the filaments of gas and young stars in this spiral galaxy? The most direct way to answer these questions, as well as to confirm the reality of the physical association, was to find further examples of such associations. Distances of Quasars It was helpful, therefore, when about the same time that the three quasars in the edge of NGC 1073 were being discovered, a pair of quasars turned up very close to the spiral galaxy NGC 622. The chance of finding two quasars this close to a galaxy is less than 4 x 10"1 or less than 2 x 10~5 if one takes into account that the second quasar is quite bright). I discovered this system during the inspection of plates that registered ultraviolet objects over about 100 square degrees of sky. There should be from 10 to 50 galaxies as bright as NGC 622 in the region searched. Therefore, this was another very significant confirma- Figure 1-2. Three quasars near spiral arms of the galaxy N G C 1073. Quasars discovered by Arp and Sulentic. tion. But the especially significant aspect of the NGC 622 configuration was a filament of material that came out of the galaxy and reached to the quasar, B1. Quasars Ejected from Galaxies This luminous connection is shown in Figure 1-3. It appears similar to a spiral arm of the galaxy except that it does not curve around as the edge of the galaxy does. Instead it comes straight out to end on what looks like an H II (gaseous emission) region. Right next to this knot is the quasar. The point is 10 Distances of Quasars aw*-.- .: _. • Figure 1-3. Two quasars improbably close to the galaxy NCG 622. The fainter, higher-redshi/t quasar Bl lies near the end of a straight /ila• ment emerging from the galaxy. that the chances of this exceedingly unusual spiral arm ending almost exactly at the position of the quasar by accident is vanishingly small unless it is physically related. It also suggests ejection from the galaxy as the explanation for origin of the quasar. From the initial discovery of alignment of quasars across disturbed galaxies, the similarity to radio sources whose alignment is caused by ejection from the nucleus suggested an ejection origin for the quasars as well. The implication has always been that the quasars are ejected from the nucleus of the associated galaxy. We will see in coming chapters continuing and inter- locking evidence for ejection of material from galaxies. Second only to the question of origin of nonvelocity redshift, the ejection of material seems to be the most significant puzzle related to galaxies. Ejection from the nuclei of active galaxies raises the question of whether conditions in that innermost center involve the normal terrestrial physics that we know about. Could the strangeness of quasars be related to the very different nature of their material if they originated in active galactic nuclei? The quasar/galaxy associations shown in Figures 1-2 and 1-3 of the preceding pages Distances of Quasars 11 Figure 1-4. Two quasars are seen projected very dose together near edge of N G C 470. Photograph try Allan Sandage. Note material in edge of galaxy apparently associated with quasars. were announced at the Texas Symposium on Relativistic Astrophysics in Munich in December 1978. This review paper presented evidence from systematic searches around a few dozen galaxies in selected regions of the sky. It was shown that a number of companion galaxies to spirals (apparently the most active kinds of galaxies) had single quasars close by to them with a probability of chance association, for the whole sample of only 10"* to 10~14. This was, of course, in addition to the highly significant confirmation of multiple quasars associated with galaxies which was already discussed here in relation to Figures 1-2 and 1-3. Most of the participants in the Texas Symposium were impressed by the results. But riding back to the hotel in the bus, I recall that in the deep seats immediately in front of me were Walter Sullivan, science writer for the New York Times and the most prominent quasar researcher of the day. The latter was explaining to Sullivan, with great patience and kindness, how all the apparent associations were accidental and how the quasars could not be local. To the credit of Sullivan, the stories on the associations did appear in the Times and similar evidence was also later reported in the Times over the years. But, the quietly spoken opinion of the astronomical authority was to prevail over the published evidence within his inner circle of science. The sharp and lethal battle which came in 1983 about the statistics of quasars near galaxies will be the subject of the next chapter. But first let us bring the story of multiple quasar associations up to date by reporting a more recent discovery. This is shown in Figure 1-4. The bright, spiral galaxy from the 12 Distances of Quasars Shipley-Ames Catalog of Bright Galaxies, N G C 470, is shown to have two newly discovered quasars in the edge of its disk. The probability of finding two quasars like the fainter one this close to a given point in the sky is about 2 x 10"4. Since one quasar is considerably brighter, and therefore less common, the probability of finding the actual quasars is even smaller. This discovery comes from an area in which about five galaxies this bright were present. Therefore, the total probability of finding this configuration by accident in that investigation was less than one in a thousand. It is interesting to note how my collaborators in this discovery computed the probability. Putting forward the hypothesis that this is what drew attention to the galaxy, they began by pretending the first quasar did not exist. Then they asked, what is the probability of finding only the second quasar this close to a galaxy? Since this probability is still quite small, they said, well, let's move it further away from the galaxy where it would still attract our attention but where it would be more probable to encounter by accident. They still obtained a significant improbability for their modified configuration but, of course, much less than the improbability of the actual association. I am quite pleased that they were willing to publish their computation in our paper because I think it furnishes a surprisingly vivid record of how, when astronomers of conventional belief have a choice between two possibilities, the more bias toward the current assumption they make, the more proper they feel the calculation is. Of course, the latest discovery is the strikingly close arrangement of three quasars around the galaxy shown in Figure 1-1. This object will be discussed further in later chapters. Table 1-1 summarizes the properties of those four systems in which we, so far, have Name GALAXY Redshift TABLE 1-1 Galaxies With Multiple Quasars QUASAR Name Dist. Mag. Red (arcsec) Shift Probability NGC 622 0.018 UB1 71 BSO1 73 18.5 0.91 0.001 20.2 1.46 0.02 NGC 470 0.009 68 95 19.9 1.88 0.015 68D 95 18.2 1.53 0.002 NGC 1073 0.004 NGC 3842 0.020 BSO1 104 BSO2 117 RSO 84 QSO1 73 QSO2 59 QSO3 73 19.8 1.94 0.01 18.9 0.60 0.006 20.0 1.40 0.02 19.0 0.34 0.003 19.0 0.95 0.002 21.0 2.20 0.01 Distances of Quasars 13 found galaxies with multiple, apparently associated quasars. An estimate of the improbability of chance occurrence is given for each quasar. It is noticeable that the quasars have a tendency to be fainter for larger redshift. This will be discussed in later chapters where evidence is advanced that the high-redshift quasars (z=2) have the lowest intrinsic luminosity. It is also noticeable in Table 1-1 that certain preferred redshift values appear more frequently than one would expect by chance. Preferred Values of Redshift This observed property of quasars having certain preferred redshifts has an extraordinary history and followed a typical course in the field. Geoffrey Burbidge noticed early in the measurements of quasar redshifts that too many redshifts occurred too close to the value z=1.95. He argued vigorously for the reality of the effect but others put emphasis on the quasars of other redshifts which were observed and ridiculed the effect. There was a sort of heroic underground of analysis of quasar periodicities starting with Burbidge and Burbidge in 1967 and continued by a number of astronomers, particularly Karlsson in 1971, 1973, and 1977, Barnothy and Barnothy in 1976, and Depaquit, Pecker, and Vigier in 1984. They all more or less agreed if you looked at all quasars known, that preferred values of redshift were apparent. In the latest and most complete analysis those preferred values of redshift were: TABLE 1-2 z (ALL QUASARS) z (NGC 1073) 0.30 0.60 0.60 0.96 1.41 1.40 1.96 1.94 As the small table above shows, the redshifts of the quasars belonging to NGC 1073 (Figure 1-2) average only 0.01 from three of the magic values. This is like a key fitting into a lock. But, of course, what is behind that locked door is terrifying to conventional astronomy. The three quasars around NGC 3842 (Figure 1-1) fit a pattern of slightly different periodicity, a period characteristic of a different group of quasars. These slight differences in period interval between groups of quasars confirm the main periodicity and at the same time confirm that each individual association of quasars is a physically related group in spite of their being composed of different redshifts. This result will be analyzed further in Chapter 5, where evidence for the reality of different groups of quasars in different regions of space will be presented. References and further comments on the past analyses of periodicities in redshifts of quasars are given at the end of the Appendix to this chapter. Summary To summarize this initial chapter, I would emphasize that with the known densities with which quasars of different apparent brightness are distributed over the sky, one can compute what are the chances of finding by accident a quasar at a certain distance from a galaxy (see Appendix to this chapter). When this probability is low, finding a second or third quasar within this distance is the product of these two or three improbabilities, or very much lower. It is perhaps difficult to appreciate immediately just how unlikely it is to encounter quasars this close by chance, but when galaxies with two or three quasars as close as we have shown here are encountered one needs only a few cases to establish beyond doubt that the associations cannot be accidental. We should also note that the hypothesis that quasars fall closer to galaxies than would be expected by chance was made and demon- 14 Distances of Quasars strated by the available evidence in 1966. Long before the writing of this book a number of investigations had confirmed this association. But the present chapter, which presents the latest evidence on just the multiple quasar associations alone, confirms with overwhelming significance the fact of the associations. The vast majority of galaxies remain to be investigated. It is clear that many of these will yield additional confirmations and give further valuable data on the properties of the quasars which would help us understand the real nature of their redshifts. The last few observational programs of this kind have now been blocked by the committees which have been appointed by most observatory directors to allocate telescope time. It is with a considerable sense of relief, however, that I can say that I think the observations have been suppressed too late. As I hope we can continue to demonstrate in this book—the observational proof of this extremely important phenomenon has already been gathered. Appendix to Chapter 1—Probabilities of Associations The basic quantity needed to compute the probability of a quasar falling within any given distance from a point on the sky is just the average density of that kind of quasar per unit area on the sky. For example, if a quasar of 20th apparent magnitude falls 60 arcsec away from a galaxy, we simply say that within this radius of a galaxy there is a circular area of 0.0009 square degrees. The average density of quasars from the brightest down to 20th apparent magnitude is about 6 to 10 per square degree. Therefore, the most generous probability, on average, for finding one of these quasars in our small circle is about 0.001 x 10 = 0.01, that is, a chance of about one in a hundred. The crucial quantity is the observed average density. The comparison of what various observers have measured for this quantity is given in Arp 1983, page 504 (see following list of references). Overall, the various densities measured agree fairly well, certainly to within 140%. For the kinds of quasars considered in these first few chapters this gives probabilities that cannot be significantly questioned. Of course, on the cosmological assumption, quasars of various redshifts must project on the sky rather uniformly. Therefore adherents of this viewpoint cannot object to taking an average background density, as observed, to compute probabilities of chance occurrences. Although quite smooth enough to compute average probabilities of association as done in the previous chapter, the cosmological assumption of uniform quasar background has led to some consternation for certain other kinds of quasars discussed in later chapters. Typically one astronomer will measure one kind of quasar in one direction and get either a large or small difference from previous measures. He will then argue that his measure is "the" correct answer and other observers were in error. He will seldom consider that the differences are real. This has led to some considerable gymnastics to try to avoid inhomogeneities of certain kinds of quasars in certain regions. A good example of this is in the 1981 reference below, which tries to rationalize a difference of more than a factor of 10 in bright apparent magnitude, high-redshift quasars in one direction in the sky. In another instance, noted below, differences in densities are dismissed as scale errors when in fact they are due to the use of continuum magnitude systems that exclude emission lines, being incorrectly compared to broad-band systems that include them. Some references which will amplify subjects discussed in this chapter are listed below with some comments. 1967, Arp, H., Astrophysical Journal, 148, p. 321. This is the first detailed paper discussing associations between radio sources and peculiar galaxies. 1966, Lynden-Bell, D., Cannon, R. D., Penston, M. V, and Rothman, V. C. A., Nature. 211, p. 838. First tests of the above associations. 1968, van der Laan, H., and Bash, F. N., Astrophysical Journal, 152, p. 621. 1968, Arpi H., Astrophysical Journal, 152, p. 633. • First paper critical of the associations and reply by Arp. 1973, "The Redshift Controversy," ed. G. Field, W. A. Benjamin, Inc., Reading, Mass. Three points not widely known about this, a report of the only actual debate to take place on the subject, are: (1) I tried to challenge the best-known quasar experts in the field at that time but none would accept; (2) after the debate had Distances of Quasars 15 been arranged, the director of my observatory heard that it was going to take place and telephoned to try to stop it; (3) the profits from the sale of the book went to support the work of Section D, the astronomy section of the AAAS (American Association for the Advancement of Science) before which the debate was held. The book summarizes and discusses the main developments before 1972. 1979, Arp. H. and Sulentic, J. W., Astrophys. Journal, 229, p. 496. This is the report of the three quasars closely spaced around the galaxy NGC 1073 (Fig. 1-2 here). 1980, Arp H., Annals of the New York Academy of Sciences, 336, p. 94. This is a review paper given at the Ninth Texas Symposium on Relativistic Astrophysics held in Munich in Dec. 1978. The paper summarizes the associations of quasars and galaxies to that date, reports the associations with NGC 622 and NGC 1073, and introduces the associations of quasars with the famous disturbed or exploding galaxy, M82. 1981, Smith, M.G., "Investigating the Universe," ed. F. D. Kahn, D. Reidel Publishing Co., Dordrecht, Holland p. 151. This article attempts to rationalize order-of-magnitude discrepancies in quasar densities in different directions and backward running Hubble relations as "selection effects." See articles below: 1983, Arp, H., Nature, 302, p. 397. 1984, Arp, H., Astrophys. Journal, 285, p. 555. These two articles discuss the evidence against excusing quasar groupings and redshift-apparent magnitude anomalies as selection effects. 1982, Veron, P. and Veron, M. P., Astron. and Astrophys., 105, p. 405. This paper dismisses discrepancies in quasar densities in different directions as "due to errors in magnitude scales used." The paper below shows it failed to distinguish between continuum and broad-band magnitudes. 1983, Arp. H., Astrophys. Journal, 271, p. 479. On page 504 of this article the density of quasars on the sky is discussed, comparisons are made to values measured by various observers, and different magnitude systems are discussed. 1984, Arp. H. and Gavazzi, G., Astron. and Astrophys., 139, p.240. Discussion of three quasars newly discovered around NGC 3842 as shown in this chapter in Figure 1-1 and Table 1. 1984, Arp, H., Surdej, J., and Swings, J. P., Astron. and Astrophys., 138, p. 179. Discussion of two newly discovered quasars at the edge of NGC 470 as shown in this chapter in Figure 1-4 and Ta- ble 1. Periodicities in the observed quasar redshifts have been analyzed by a number of authors. The latest references, from which the earlier references may be gleaned are: 1984, Depaquit, S., Pecker, J.-C., and Vigier, J.-P., Astronomische Nachrichten, 305. p. 339. 1984, Box, T. C. and Roeder. R. C , Astronomy and Astrophysics, 134, p. 234. Note on Periodicities in Quasar Redshifts. Periodicities in quasar redshifts have been found in all samples except one where the person who analyzed it truncated the sample in a particular way that removed the periodicities (see Depaquit et al. reference). Some authors have argued selection effects are responsible for the periodicities. This is clearly untrue because major emission lines can be seen with objective prism searches throughout the redshift range. Concentrations of redshifts close to z = 1 for optically selected quasars around companion galaxies and in dense groups of quasars prove that techniques of photographic discovery by ultraviolet excess are not significantly biased. Of course, quasars selected by their radio emission should not be biased in redshift at all. It is shown in Chapter 5 that all quasars tend to have certain rather discrete, permitted redshifts but that different groups have slightly different periods. It is the addition of these slightly shifted peaks from group to group which broadens the overall peaks as observed in the total quasar sample. The bottom line is that quasars have the astonishing property of occurring at certain preferred values of redshift, these values occurring with a definite period whose origin is a mystery at this moment. The crucial discovery of periodicities in quasar redshifts was made by K. G. Karlsson. That discovery was that the redshift peaks fit a formula A log (1 + z) = const. As Table 2 in the preceding text shows, the observed redshift peaks for the average of all quasars fit the formula with const. = 0.089. Individual physical groups of quasars have slightly different constants (see Arp, review paper presented at JAU Symposium 124, Beijing, China, August 1986). Unfortunately, a possibility of which all young astronomers are aware occurred in the Karlsson case. This creative researcher was not employed in astronomy and subsequently went into medical science. 16 Distances of Quasars THE BATTLE 2 OVER STATISTICS As mentioned in the first chapter, after 1966, a number of investigations built up the evidence that quasars were associated with nearby galaxies. One of the first systematic investigations of quasars over the sky was an analysis I published in 1970. I was still a faculty member at Caltech at the time, and I remember well the custom of astronomy luncheons at the Faculty Club every Friday. I would bring in new examples of quasars falling improbably close to galaxies and share these photographs with my colleagues. Finally, the consensus was communicated to me that they believed these to be specially selected cases and that as scientists they could only accept the effect if a full statistical test were performed on a complete sample. I thereupon took about six months away from normal activities, enlisted the aid of Fritz Bartlett, a radio astronomer, to program the large IBM computer which Caltech then relied upon, and proceeded to analyze the position of all the then-known 3CR quasars (Third Cambridge Catalog Revised Survey of Strong Radio Sources) with respect to all the galaxies listed in the ShapkyAmes Catalog of Bright Galaxies. Figure 2-1 shows the striking result of those computations. It shows how the separations on the sky between a set of radio quasars and cataloged galaxies steadily decreases as brighter and brighter galaxies are considered—that is, the association with these quasars is stronger as galaxies closer to us in space are considered. The powerful computer enabled many imaginary sets of random quasars to be generated and compared to the galaxies, and thus showed that it was only the real quasars which had this property of falling closer and closer to brighter and brighter galaxies. I returned with excitement and anticipation to the Friday luncheon and explained what I had found out. There was a unanimous response "Oh, no one believes statistics!" The paper containing these results was published in 1970 in the Astronomical Journal and little notice was taken of it. Eventually, in 1983, I utilized some of the clues developed in that paper to make the most recent and detailed proposals as to the location of quasars in space. These concepts are developed further in Chapter 5. The Battle Over Statistics 17 1 1 1 1 i rr to '2° - - o RANDOM QUASARS LL o io° - - XUON < 8° - - u (1970). But, other investigators were also developing evidence that quasars were associated with galaxies on the sky. In 1971, G. R. Burbidge, E. M. Burbidge, P. M. Solomon, and P. A. Strittmatter showed that among the quasars then known, those that fell very close on the sky to bright galaxies fell much closer than would be expected by chance. In a carefully worked-out statistical analysis, they showed that with even the few cases known from casual investigation, the chance that these closest coinicidences occurred accidentally was less than 5 x 10~3 or of the order of one in two hundred. The result was never criticized in print. As usual, however, it was excoriated in private. One of the major techniques for dismissing such results was introduced about this time. The catch-phrase is known as "a posteriori statistics." Normal people may not find that so catchy. But the idea is rather simple: After any event has happened, the probability of it happening in that precise way can al- ways be computed to be vanishingly small. For example, if two people are photographed in the streets of a city of one million inhabitants, we would say that the chance of A being directly adjacent to B is one in a million. But in any random street scene there are perforce many A's next to unrelated B's. This is all quite evident from common sense. But what is also quite evident from common sense is, that if we continue to get photographs at different times and places of A next to B, we had better conclude some relationship exists between A and B. So far as the charge of "a posteriori statistics" which has been levelled at each new piece of quasar evidence is concerned, the association of quasars with galaxies was demonstrated in 1966. Each succeeding example has therefore been an additional confirmation of an "a priori" prediction. The dismissal of each of these on a case-by-case basis with the excuse of " posteriori statistics" has been, at best, poor science and, at worst, a tactic of evasion. 18 The Battle Over Statistics 30 300 3000 30000 REDSHIFT OF GALAXY (km/sec) Figure 2-2. Relation showing that the greater the distance of the galaxy from the observer, the smaller is the apparent separation of the associated quasar. From Arp (1983). G. R. Burbidge, S. L. O'Dell, and P. A. Strittmatter in a later paper in 1972 showed that quasars associated with more remote galaxies appeared closer to the galaxy of association, as if the whole partnership was viewed from a greater distance. This relation is very important because it is what we must expect if we view associations through a range of distances, some associations close by and some more distant. The relation was later strongly confirmed over a much larger range in distance of the central galaxy and with larger numbers of examples in a paper by Arp in 1983. This is shown here in Figure 2-2. At first sight this result seems to contradict the previous result that quasars fall closer to brighter, more nearby galaxies. The key point to understand, however, is that nearby quasars are statistically close (closer, on the average, than one would expect by chance) to nearby galaxies and that more distant quasars are statistically close to more distant galaxies. But, the nearby associations can subtend a large angle on the sky and have rather large separations compared to the separations involved in more distant associations. The difficulty comes when people assume on the local hypothesis (quasars closer than their redshift distances) that all quasars are at the same distance and then try to analyze this mixture of quasars at different distances with galaxies at different distances. Naturally they get porridge. An example of this follows. In 1980, a test analysis was made by one researcher of some quasars which had been found in sample areas of the sky by objective prism techniques. (The objective prism on a telescope enables the selection of objects The Battle Over Statistics 19 which have strong emission lines in their spectrum—the stellar images among these are mostly quasars.) These objective prism quasars were analyzed to see how close on the sky they fell to NGC galaxies. (Perhaps indicative of the fast pace at which astronomy moves, the New General Catalog of Gahxies (NGC) by Caroline and William Herschel was completed by J. L. E. Dreyer in 1888.) The NGC contains over 7000 objects, most of which are fainter objects at medium to large distances. There are no accurate magnitude limits for this Catalog and in addition an inhomogeneous selection of objects was made. But the Arp paper in 1970 showed bright quasars to be generally associated with the brightest galaxies in the sky. Why did the later paper in 1980 try to associate quasars with more distant galaxies? And, should it have been such a surprise when it reported no significant associations? Perhaps there is some clue to be had in reading carefully the words of the author as he coyly "suggests" that the "seemingly high frequency of quasar/galaxy pairs reported, primarily by Arp" may be due to "uncertainties" in the adopted quasar densities used and "it is this effect that makes many astronomers skeptical about the statistical significance of Arp's configurations." The true state of affairs was illuminated a year later by the only young astronomer to have ever regularly dared to test the claims of the establishment. Jack Sulentic analyzed the quasars just discussed as having been claimed to disprove association with galaxies. He analyzed these same quasars and also additional samples, but this time with respect to the bright nearby galaxies with which they were supposed to be associated. He found consistent and significant quasar/galaxy associations in all the quasar samples! Moreover, he found the quasars associated with fainter galaxies fell closer to them on the sky as would be expected if these fainter galaxies were more distant. All this poses an interesting question: Why, when the establishment believes so fiercely in the different distances of galaxies (as indicated by their different redshifts), do they always insist on testing the association of galaxies and quasars by assuming that all galaxies (bright and faint) are at the same distance from us? We will see this obviously incorrect assumption used again and again in attempts to disprove the association of quasars and galaxies. A. Quasars near Companion Galaxies During these investigations of quasars associated with galaxies, it became apparent that a particularly favorable configuration existed when a large spiral had an associated companion galaxy. Strikingly closer to these companions than expected by chance were found quasars. A representative selection of the cases is shown in Figure 2-3. One reason for this could be that the companion galaxies are younger and more active and tend to produce more quasars. But whatever the explanation may be, the associations furnish vivid evidence of quasars associated with much lower redshift galaxies. As this evidence built up, the opposition became more silent. I felt that one final test would enable a resolution of the question. Starting in 1978, I outlined a considerable section of the sky, defined beforehand what kinds of galaxies I was looking for quasars around, and started to make the observations. (The area of the sky sampled was defined by bright spirals with companions that were contained between the right ascensions of NGC 2460 and 3184.) For long nights on the Mt. Palomar 48-inch Schmidt telescope I photographed these areas in the sky in two colors in such a way as to be able to pick out the ultraviolet-excess, stellar-appearing objects which were the quasar candidates. I labori- 20 The Battle Over Statistics ously scanned the plates for the candidates and struggled with the spectrograph on the 200-inch Mt. Palomar reflector to obtain their individual spectra so that I might know with certainty which were bona fide quasars. Finally, after three years, I had finished. From 34 predefined candidate galaxies, I had found 13 cases where the quasars fell so close that the chances were, in each individual case, only about 1 in 100 of being accidentally associated. To find 13 such cases out of a limited number of trials implied fantastic odds against being a chance occurrence. I calculated about 10"17 (one chance in a billion would be 10-9) The result was published in the Astrophysical journal. A month or so passed. Suddenly the storm broke. Two papers arrived, each denouncing the probability calculations. They had been sent to the Astrophysical Journal Letters for quick publication and copies had been sent on to me by the Journal as a customary notification of critical papers. Both papers took essentially the same tack, that because I had used a probability of about one in a hundred of being accidental as a criterion for being associated, all cases where the probability was even slightly greater should be excluded. They wound up in the ludicrous position that associations where the probability was 0.012 or 0.013 should be excluded from the calculation as not improbable. (Basing my original probability calculation on p < 0.015, instead of the rounded-off p — 0.011 used, would have raised my final probability of accidental association from p ~ 10"" to p = 7 x 10""). These papers also used scaled densities of faint quasars to calculate the probabilities of finding bright quasars, despite the fact that it was clear from the literature that these brighter quasars were much less common. They also assumed that areas had been searched for quasars that had not been searched. These papers, after extensive refereeing, were never published. What did the damage was a paper that used the same in- correct arguments which was published with extreme rapidity by a British journal, Monthly Notices of the Royal Astronomical Society. I first became aware of this paper when I received an unprecedented note from the editor of the main Astrophysical Journal. (The main Journal has a separate editor from the Letters section.) He mailed me a preprint of the paper which was to appear so quickly thereafter in the Monthly Notices. The note from the Ap. J. editor read essentially: "We received this from the author and are conveying this simply for your information." Professional ethics required that both the author and the Monthly Notices send this kind of preprint directly to me. But, since from that day forward, my papers had enormous difficulties appearing in the Astrophysical Journal, I eventually understood what was the probable reason for this highly unusual maneuver. There was a further effect of the Monthly Notices paper, however, which presented me with additional problems. Another astronomer from the British establishment, who had been in communication with the first, sent a note to the Astrophysical Journal Letters which was eventually published. It took a different, apparently more valid tack. It pointed out that when I computed the probability of finding a quasar of a certain brightness at a certain distance from a galaxy, that I should also take into account that I could find other quasars at different brightnesses and distances that could also have a low probability of chance occurrence. In other words, there was more than one way of obtaining a chanceoccurrence probability of about 1 in 100. That gave me a number of sleepless nights because I had to acknowledge that this appeared to be a flaw in my original thinking. After pondering the question deeply, I finally came to the conclusion that I had intuitively assumed that galaxies at a certain distance would possess quasars at a certain, The Battle Over Statistics 21 figure 2-3. A representative set of examples of large spirals that have companion galaxies with apparently associated quasars. The arrows mark the quasars. See also N G C 5296/97 in following chapter (Fig. 3-6). None of these examples were used in the complete statistical analysis of the test area discussed in the text and shown in Figs. 2-4 through 2-6. characteristic separation from the galaxy and that other separations would not occur. Actually, this is just the relation shown in Figure 22. With considerable anxiety, I recalculated the probabilities in a completely different manner, taking into account explicitly instead of implicitly the distance of the central galaxy calculated from the conventional redshift criterion of its distance. I got the same improbability of association. I felt vastly relieved that my initial assumption had been correct. I encouraged publication of the paper that criticized this aspect of my previous probability calculation and accompanied it with a recalculation, by the different method, which confirmed my original results. Actually, the critical paper, even after making the most unrealistic and unfavorable assumptions, still derived an improbability of 10"' that the original result could be accidental. This was still enormously strong confirmation of my original result. But the whole point of that paper was suddenly contradicted by a note added in proof (unseen by me or any referee) that referred to the Monthly Notices paper as disproving my original result. So the whole scientific content of the exchange was defeated. Reading now the papers, from some distance in time, it is clear that this was a rather effective double-play between two authors. In order to eliminate what seemed to be the inescapably small probability of chance occurrence originally obtained, one claimed that part was due to a large factor of error due to one cause and the other a large factor of error from another cause—neither had to make a formal calculation from the actual data. An extensive recalculation on these data was later made by two other authors, E. J. Zuiderwijk and H. R. de Ruiter. It was reported in the Monthly Notices 1983 (see Appendix). With essentially the same precepts as the earlier authors, they found instead an association of quasars with the galaxies with an overall chance of only about 1 in 100 of being accidental. The reason that this still fell short of my original improbability is that they also ignored the fact that the galaxies with which the quasars are associated have varying distances from us. The absolutely key point is that my original calculation used a simple way of taking into account the different distances from us of the galaxies around which the quasars were found. To ignore this elementary piece of astronomy in which everyone believes leads to a strong dilution of the effect by looking in areas where quasars are not expected to be found and ignoring areas where they are. The nub of the matter is the following: When testing the null hypothesis, the assumption that the quasars are not associated, the last two authors find evidence for association. The burning question then becomes, why not test the hypothesis that they are associated, taking into account the varying distance of the galaxies? Why not repeat the test which gave the original, overwhelmingly significant association with the galaxies? Of course, when two sides differ so sharply over calculations performed on the same data, one side must be wrong. If my side is wrong, I have to wonder whether I got the wrong answer because I wanted very much a certain result and this desire prejudiced my judgment on how to make the calculation. The other side would have to face the same question. Perhaps this is what is preventing the calculation from being pressed to the decision that it is certainly capable of. Finally, in 1983, a paper by myself was published in the Astrophysical Journal giving the final data on the observed fields and reanalyzing the statistics. Figure 2-4 here is from that paper and shows how the quasars concentrate between about 7 to 20 kpc radius around the particular galaxies that were sampled in that investigation. The average background density of quasars could not conceivably contribute significantly to the 24 The Battle Over Statistics 10 on 8 / _ in QUASARS OBSERVED o on mID 4, /, 2 '/, 2 -//, I—1 I 0 AVERAGE BACKGROUND DENSITY EXPECTED r- i XL r ' 1 1n, . n . 20 40 60 80 RADIAL DISTANCE FROM GALAXY (kpc) figure 2-4. Numbers of quasars found as a function of radius around companion galaxies in Arp's 1983 study. Expected average background counts are shown Iry the dashed line. observed numbers in the small area on the sky in which these quasars were found. In fact, the density of quasars at these radial distances from their associated galaxies (in this investigation the investigated galaxies were all companion galaxies to larger galaxies) exceeded by more than 20 times the measured density of these kinds of quasars away from such galaxies. This result appears in Figure 2-5, where the error bars on the determinations from quasars of various apparent brightness class are also shown. All three magnitude classes of quasars agree that the excess density around these galaxies must be, taking a liberal estimate of the possible background density of quasars, between 10 and 30 times that background density. This is the same result that was the conclusion of the first, much maligned paper. Apparently the chance of accidentally getting an overdensity by the original factor of 20 is about 10"" (or more accurately, 7 x 10"16 )• The final published data and exhaustive analysis, however, made no difference. It was sufficient for many astronomers to have something—anything—in print claiming the associations were spurious. This was brought home with particular vividness after a physics-astronomy conference in Geneva in November 1983.1 was chatting with a British cosmologist about C fields and inflationary theory when the subject of quasars was mentioned. This theorist looked distressed and said apologetically: "Well, I was interested in your quasar investigations, but then I was told your observational evidence had been proved wrong." It was clear that the research had been successfully discredited. But, before that I was even more painfully aware that the event had been used with disastrous effect within my own home observatory. While the critical paper that had been published so quickly in the Monthly Notices was still in preprint form, it had also been sent, special communication to the new director of my observatory, courtesy of the most eminent quasar researcher on the Caltech faculty. This occurred shortly after Caltech had broken the agreement to jointly operate Palomar Observatory with Carnegie Institution of Washington. Soon after this, a friend of mine met the director at the airport in Washington, D.C., and asked about me. The answer came back: "Well, I wish he would get his statistics right!" When I heard this, I went to his office The Battle Over Statistics 25 at, Wolstencro/t, and He. the result that the paper was eventually published after a delay of only one year and three months. Certain significant aspects of the NGC 1097 situation, however, serve to introduce the next result in this chapter: (1) The radio emission in NGC 1097 occurs asymmetrically placed over on the side of the strong northern jets. (2) The X-ray emission is asymmetrical, likewise on the side of the northern jets (both the inner and outer X-ray emission). (3) The quasars are preferentially aligned with the jets, more of them being on the side of the northern jets. These latter points are illustrated in Figures 4-4, 4-5, and 4-6. These are extraordinarily important points because they make clear that the radio and inner X-ray material is associated with the galaxy. Since the quasars are associated with the outer X-ray material, which is con- tinuous with the inner, and since the quasars are also aligned with the jets, this shows that the association of the quasars cannot be coincidental. It must have some physical significance. Specifically, these results imply that the X-ray material, radio material, and quasars are all part of the ejection which is marked by the optical jets. Do other examples exist to support this picture? B. The Disturbed Galaxy NGC 520 The first chapter of this book describes how, in looking around galaxies in the Atlas of Peculiar Galaxies, I found apparently associated radio sources and quasars. In order to amplify this point a little here, I should say that these associations were mostly with numbers 100 to 160 in the Atlas. Those categories Certain Galaxies with Many Quasars 51 JET 2 10'- •30° 30' DET1 X-RAY 50' »m 1.2m NGC 1097 /, \ \ N q \ 26' \ \ f m > 30°30" - JO 0 VAI RADIO 30s © 1 1 1 2hUm00s Figure 4-4. XTOJ and radio maps o / N G C 1097. Note quasars 1 tnrougn 6 (X symbols) coincide uiit/i man> of the patches o/X-ray emission. Placement of jets and then extensions indicated by full and dashed tines. 52 Certain Galaxies with Many Quasars 12 in UJ CD ols_i N JETS • •i •• • • I m •• S JETS n « «» »««|«« 0 (20 240 POSITION ANGLE FROM NGC 1097 (DEG.) Figure 4-5. The alignment of the direction of all the quasars (filled points) with the direction of the jets in N G C 1097. From Arf>, Wolstencro/t, and He. X-RAY EMISSION AROUND NGC 1097 1.5 HARDNESS RATIO 0.5 X '' } */ / 1f / 2 4 6 DISTANCE IN ARC HIN 20 INNER 32 TO 64 ARC SEC 120 360 COUNTS 40 20 INNER 64 TO 90 ARC SEC 120 360 POSITION ANGLE Figure 4-6- These plots demonstrate continuity of high energy to low energy ratio in X-ray sources outward from center of N G C 1097. Also peaks of inner X-rays in direction of strongest jet (pos. angle n'360°). Calculations from Wolsttncro/t. Certain Galaxies with Many Quasars 53 QUASARS QUASARS Figure 4-7. The disturbed galaxy NGC 520. It is number 157 in the Atlas of Peculiar Galaxies. Directions of cone of radio sources discovered in 1967 and lines of quasars discovered in 1970 and 1985 are marhtd. in the Atlas represent the most chaotic and disturbed objects, presumably contorted by inner activity and explosions. Galaxies in categories with much higher and lower numbers represented a selection of interacting doubles, dwarfs, and other peculiar objects which one would not expect to show violent activity. Thus a significant point, which most critics insisted on overlooking, was that the associations of sources turned out to be primarily with these central numbers in the Atlas. One of the most disturbed objects in the entire Atlas is number 157. It is shown here in Figure 4-7. In 1967, I knew of no quasars around it. I did, however, notice an unusual number of radio sources which appeared to define directions of ejection to the northeast and southwest. Three years later, in 1970, I became aware of four radio-loud quasars to the southwest. They defined an almost perfect straight line pointing back toward NGC 520. The properties of the quasars in this line resemble each other in a number of ways such that the overall probability of such a chance configuration is less than one in a million. The line of quasars also lies within the previously defined direction of radio source ejection. The line is shown in Figure 4-8. I remember showing this line of quasars to John Bolton in 1970. As one of the founders of radio astronomy, John has made many identifications of radio sources with quasars over the sky. He said he had not seen any so straight but had seen many approximate lines and apparent chains. It seems amaiing that until very recently no one has systematically investigated these features. I suppose the reason is that the configurations involve quasars 54 Certain Galaxies with Many Quasars 1 - \ \ \ ^ 4°- \ NGC52Cy'v. ®J).77 \ X®067" ID \ ^N Q ^©2.11 - 32m ^©0.72 \ i i6m 1 i room RIGHT ASCENSION i i um Figure 4-8. A plot o/strongest radio quasars in a larger area around N G C 520. Redshifu are written next to symbols. 28m of different redshift. (Recently, tentative alignment results for faint quasars were put forward in the literature with great difficulty by Clube and Trew, Liege Conference, p. 374 and Monthly Notices of the Royal Astronomical Society, submitted.) Again the investigation of NGC 520 languished, this time for about 10 years, until 1980. By that time, I had started finding quasars over homogeneously searched areas of the sky by using the ultraviolet-excess techniques described in Chapter 2. Jean-Pierre Swings and Jean Surdej of the Institut d'Astrophysique in Liege, Belgium, joined me in a collaborative project of searching several areas of about 20 square degrees each on the sky. I had already obtained plates covering the NGC 520 region but here was an opportunity to have the quasar candidates surveyed by astronomers without previous experience with the region, who would therefore be unbiased. Over the next few years I measured their candidates with the Carnegie telescope in Chile. The most conspicuous feature in the whole field turned out to be a line of quasars going through NGC 520! In order to make absolutely sure of this result, I then asked Oscar Duhalde of the Las Campanas Observatory to take an ultraviolet/blue plate on a completely different telescope and the two of us examined independently the 2.1-squaredegree field which had been studied around NGC 520. The original search went to 20th apparent magnitude, but to be on the safe side for completeness, we restricted ourselves to quasars brighter than 19th magnitude. Seven quasars resulted from the Swings/Surdej candidates, five on the line and two off. Six additional quasars were found by Arp/Duhalde, one on the line and five off. Swings and Surdej originally were enthusiastic about the line, but after the Liege Conference in 1983, they renounced it, claiming that the association had been found from the Arp/Duhalde search!! The quasars found in this double-independent search of the area around NGC 520 are shown in Figure 4-9 where the reader will have to judge for himself the significance of the line. It is important to realize that just the distributed quasar density near NGC 520 is, by itself, much higher than in the expected Certain Galaxies with Many Quasars 55 • 6 3 S b u. •nut 1 J 0 £/.:• / . • >•• N '•• *:',;.• ..•• • * * •' • * - " •'• .• :• - i : ^ ••. / . - .v •v . v- •".•", • : . : • > • • » \ * . • % • Figure 4-9. All quasars brighter than 19th apparent magnitude around N G C 520 are shown, from two independent searches within the pictured area (from Arp and Duholde 1985). Is there a line of quasars going through N G C 520! Why did referees and editors of two major journals refuse publication of this result! background. It reaches 15 times expected density, about the same factor that we have encountered for excess density around galaxies previously discussed. Since all members of the line are undeniably quasars, in order to make the line "go away," one would have to discover many more quasars which would further increase the density above its already strong excess. The next piece of the puzzle fell in place, as the X-ray astronomers would say, serendipitously. I was scanning a list of targets which the Einstein X-ray observatory had observed. There, amongst a study entitled, "Normal Galaxies," was my old friend, NGC 520. How anyone could call NGC 520 a normal galaxy passes all understanding. But I understood somewhat after I wrote the Harvard/Smithsonian Center to try to get a copy of the observations. It turned out they had been made by a new acquaintance, an astronomer who had just advanced his career with an article in the magazine called The Sciences, in which he had dismissed as trash anything I had ever uttered about redshifts. You can imagine that getting my hands on this X-ray map was not easy. I have never been able to compare it with other fields taken under the same conditions. But the single map I did get showed a lot of Xray emission from the vicinity of NGC 520. It was apparent that whatever is producing the X-rays is elongated more or less along the line of the quasars. Figure 4-10 shows that the direction of the radio emission designated in 56 Certain Galaxies with Many Quasars RADIO SOURCESO967) m UJ X-RAY EMISSION CD 23' < r<37' OPTICAL QUASARS(1983) RADIO QUASARS(1970) a: 1 2UJ QUASARS r<36' Figure 4-10. The directions on the sky, from N G C 520, of the radio sources discovered in 1967, and the quasars discovered in 1970 (arrows). The X-raj material and additional quasars discovered in 40 120 200 280 3 6 0 1983 are rej>resented try upper and POSITION ANGLE FROM NGC 520 lower histograms respectively. 1967 coincides very closely with the direction of the radio quasar line discovered in 1970, which then coincides with the direction of the quasar line discovered in 1983, which in turn coincides with the peak direction of Xray emission discovered in 1983. As in the previous case of NGC 1097, we have in NGC 520 strong evidence for quasars ejected in a jet and counterjet direction and for this ejection to be accompanied by X-ray and radio material. In NGC 520, no optical jets are visible, but on the other hand, the galaxy is much more explosively torn up than NGC 1097, suggesting a very violent event. As to the interpretation of the structural appearance of NGC 520, there has been quite a fad in recent years to explain all asymmetrical galaxy forms as collisions or mergers. NGC 520 was interpreted in this fashion by my friends Alan Stockton and Francesco Bertola. The Armenian radio-astronomer Tovmassian, however, finds a single, compact radio source in the center of NGC 520 and favors the exploding-galaxy interpretation. My opinion is that the photographs taken with high-resolution telelscopes do not permit the interpretation of NGC 520 as two galaxies. Finally, the radio, X-ray, and quasar activity we have just discussed supports the interpretation of NGC 520 as an active galaxy. Two things have always bothered me about NGC 520. One is the large scale of the activity—the straight line of radio quasars stretches up to 7 degrees away from the galaxy—and the second is the relatively bright apparent-magnitudes of the quasars. The redshift of NGC 520 is about czo = 2272 km s"1, which, on the cosmological hypothesis, gives a distance about twice as great as the Virgo supercluster center. I have occasionally had a little thought: Is NGC 520 really this distant or it is closer than its redshift would indicate? In Chapter 5, we see the most recent evidence that the galaxy may be closer. Certain Galaxies with Many Quasars 57 Figure 4-11. The /ilaments emitting light from gaseous hydrogen-alpha are shown in this picture of M82, photographed by Allan Sandage. C. The Exploding Galaxy M82 This galaxy has always been one of the brightest and most peculiar known. (M stands for Charles Messier, the 18th century comet hunter.) In 1963, Allan Sandage took photographs of M82 revealing a twisted set of emission filaments emerging from either pole. M82 was interpreted as an exploding galaxy and became a prototype for objects in the universe exhibiting violent activity. (Some years later, there was an attempt to interpret M82 as a normal galaxy drifting through a cloud of dust, but in my opinion, the photographs patently rule that out. In addition, velocity spreads were later found in the gaseous filaments.) Sandage's photograph is shown in Figure 4-11. We will come back to M82 in later chapters because it is such a key object. But for the present it suffices to say that the galaxy is a companion to M81, an even larger, apparently normal spiral which dominates the M81 group of galaxies. From the results of finding quasars near companion galaxies described in Chapter 2, I was predicting that quasars should be found near companions like M82. The system lay in a direction passing too near our own galactic plane, and hence was too crowded with stars for me to use the ultraviolet-excess technique for quasar-discovery. But about this time, Arthur Hoag had invented the grism (grating prism) which would give small spectra of faint objects in a field. Because Art had been a long and valued friend, I urged him to search for quasars around companion galaxies such as 58 Certain Galaxies with Many Quasars N •• • • * • • • *• * * , &;-. I <••- • # •BjlinH it., * • ^; i•JHIM •H£ • • • m 1 ? * I 1 • M HUB Hmasmp 1 * ;, '."• ' " ' ' ' •* ^ .*•-•'•,» •"'':« • • ts4: • , !<•; 81| ^ . Figure 4-12. The four quasars which have been discovered near M82 are circled. The photograph is oxygen emission, showing the asymetricai/foments which appear to be associated with the explosion. A radio source (slightly extended contours) appears to have been ejected along with the quasars from the "notch" in the southeast side of M82. M82. I must say, the prospect of one of the sars back to an origin in M82. As Figure 4-12 genuinely good guys in astronomy making an shows, they can be all contained within a important discovery with his genuinely impor- cone emerging from the center of the galaxy. tant instrument pleased me also. But he in- The opening angle of this cone is just about sisted on observing so-called blank fields on the observed opening angle on the ejection the usual idea that this would give us the num- cones that we have seen emerging from NGC ber density of quasars at the edge of the uni- 1097 and NGC 520. Moreover, in M82, this verse. (When it comes to the universe, one ejection cone emerges from the galaxy on the edge is about as good as another, hence the same side as the puff of forbidden oxygen blank fields.) He was observing with Sandage emission. In fact, on the northern side of the one night and ran out of objects at the end of [O II] emission there is a notch cut out, into the night. To fill the time, they took a plate of which the orgin of the ejection cone natu- M82. Can you imagine, they found three qua- rally fits. Finally, if we look back at Sandage's sars about a galaxy diameter away to the south- original hydrogen-alpha picture in Figure 4- east, an order of magnitude closer to each 11, we see that the most conspicuous absorp- other than expected and all of about the same tion feature in the galaxy goes directly back redshift! to the center along the line of this postulated Of all the quasars known, this is a unique grouping. It could hardly be a coincidence ejection cone (see also Arp 1980 in references in Appendix to this chapter). that they fall so close to the unique galaxy The most recent developments concern- M82. It is natural, therefore, to trace the qua- ing this system again come from X-rays. Fig- Certain Galaxies with Many Quasars 59 M82 X-RAY MAP 69°55'20' 1 / - 69°55'00" — i 9 QUASARS 69°54'20" o 9h5f45s 40s 35s RIGHT ASCENSION Figure 4-13. The radio map (photographed) o / M 8 2 , with line contours o/X-ray emission superposed. From Kronberg, Bier- mann, and Schwab. See chapter 9 for further discussion of X-ray emission. ure 4-13 shows an X-ray map superposed on a be related to any other unorthodox results radio map of the system. The compact, varia- which had been previously known. But, as Fig- ble radio sources in a line down the spine of ure 4-12 shows, the patch of radio emission fits the galaxy are very peculiar, but the most sig- exactly into the quasar ejection cone just as it nificant result for our purposes is that the X- emerges from the explosive center of M82. ray material is again over on the southeast side of M82 and, in fact, extends out generally along the direction of the ejection cone which we had identified several years previously from the quasar and morphological data. Added to the association evidence for the quasars, this correspondence of detailed X-ray and morphological evidence with the quasar alignment would indicate in this one case alone that the association of the quasars cannot be accidental. To summarize the results of this chapter, I would say that we have analyzed over many years and in great detail, three of the galaxies with the best evidence for explosive, ejecting behavior. In each of these cases we have found very strong statistical evidence for density enhancements of quasars near these galaxies. In all three cases alignments of quasars towards the centers of these disturbed galaxies exist that can hardly have occurred by chance. Connected with these alignments we find evi- Radio ejection? Of course. When J. J. Condon was observing a number of bright galaxies with the Very Large Array telescope, he found an unusual radio source just to the southeast of M82. As is customary in extragalactic astronomy, this observation would not dence for ejection of radio material and X-ray material. There does not seem to me to be much doubt, in just these cases alone, and even without the evidence from the first three chapters, that quasars arise from some kind of ejection process within galaxies. 60 Certain Galaxies with Many Quasars Appendix to Chapter 4 1984, Arp. H., Wolstencroft, R. D., and He, X. T. Astrophys. Journ., 285. p. 44. This is the latest paper on NGC 1097 and contains references to earlier work. NGC 1097 is also discussed below. 1983, Arp, H. in "Liege Symposium on Quasars and Gravitational Lenses," Institut d'Astrophysique. Universite de Liege, June 1983, Paper 47. A number of results on "Groups, Concentrations, and Associations of Quasars," appear in the proceedings of this conference which are not published elsewhere. Discussion recorded at the end of this paper includes some obviously inaccurate statements about quasars of redshift near z = 1 containing only one visible emission line in their spectrum and selection of quasars near NGC 520. 1980, Arp, H., Annals of the New York Academy of Science, Vol. 336, pages 94-112. This is the "Texas Symposium" in Munich. Results for association of quasars with galaxies, particularly with companion galaxies, are reviewed to that date. The evidence for an ejection cone of quasars from M82 is discussed only in this publication. The fourth quasar southeast of M82 is shown in Astrophys Joum., 271, p. 479. 1983, Condon, J.J., Astrophys. Journ., Supp., 53, p. 459. Radio observations of M82. 1985, Arp, H. and Duhalde, O., Publ. Astron. Soc. Pac, 97, p. 1149. Observations of quasars near NGC 520. Certain Galaxies with Many Quasars 61 DISTRIBUTION 5 OF QUASARS IN SPACE The conventional view of quasars is that they are normal galaxies which have, for some reason, superluminous nuclei which enable them to be seen at great distances in the universe. But if quasars really were these kinds of galaxies, we should expect to see them clumping into the clusters or superclusters that characterize the distribution of galaxies on the largest scales. Attempts have been made to relate some quasars with faint, adjacent galaxies of the same redshift. But no conspicuous clusters are evident. Moreover, it is completely clear that we do not see clusters or groups of quasars all having closely the same redshift. The conclusion forced on the conventional believers is that quasars are so rare that we seldom see a cluster of galaxies with one; that is, far less than one quasar exists per average supercluster. But if we look around the quasars we do see, to a faint enough level, we should see the galaxies that accompany them in their clusters and superclusters. Wide-field Schmidt telescopes, since the invention of high-detectivity emulsions, can routinely register galaxies to a limiting apparent magnitude fainter than 23. That corresponds to a redshift for a normal galaxy of at least z = 0 . 5 . We should be able to easily see faint, rich clusters of galaxies around quasars out to this redshift and beyond. We do not. (You can believe that if we did we would have heard an enormous amount about it!) Clearly, this is an outstanding violation of the cosmological assumptions. The only way to place the quasars into clusters, where they "belong", is to move them in much closer than their redshift distance. In fact, throughout this book we shall see that if the objects with anomalous redshifts are assigned closer distances they join nearby groups and clusters of galaxies. This is the way the universe is observed to be structured. But if these objects are left out at their putative redshift distances they are left hanging isolated in space. Of course an inverse investigation could be made. The faintest, richest superclusters in the sky could be identified and the area which they define summed. On the cosmological assumption these areas should contain most of the known quasars of the same redshift. My impression, from what I have seen of the distribution of faint galaxies with re- Distribution of Quasars in Space 63 TABLE 5-1 Quasars in Dense Groups No. Group Name 1. Hazard 1146+ 1112 (~2°SEof NGC3810) 2. NGC 450 SW (-2° SW of NGC 450) 3. NGC 2639 SE (~30'SEof NGC 2639) 4. NGC 1097 NE (within 24' of NGC 1097) 5. NGC 520 (within 28' of NGC 520) z. 1.01 0.955 1.18 3.1 0.33 Redshifts of Quasars Z; Zj Z< Z5 1.01 1.10 0.86 2.12 0.960 0.69 1.23 1.89 1.11 1.52 (0.78) 0.53 1.00 0.34 0.89 0.92 1.20 0.63 1.41 Factor of Density Area Over It, (sq. deg.) Average 0.014 60 0.013 64 0.013 51 (1.1) 0.04-0.02 21-50 1.47 0.05 60 spect to quasars, is that this test would fail spectacularly and that this is why the researchers of cosmological persuasion have not performed it. On the other hand, examining the distribution of quasars on the sky does reveal conspicuous clumps and groups of quasars. The trouble is that the quasars within each group have dissimilar, or only moderately similar, redshifts. If thes< groupings are real, and if the redshifts were distance indicators, each group, with its range of redshift, would represent an elongated "finger" of quasars pointing just at our position in space. The Copernican principle, namely that the odds are overwhelmingly against our occupying a special position in the universe, would then require that the redshifts of these quasars did not indicate their distances. The fingers pointing at us are telling us our assumptions about redshift are foolish. A. The Densest Groups of Quasars In order to investigate this question of grouping of quasars without prejudice as to redshift, I list in Table 5-1 the densest groups of four or more quasars that I have encountered in my 20 years of quasar research. Some of these groups, such as the ones shown in Figures 5-1 and 5-2 are so compact and isolated that there could be no question that they are a physically associated group of quasars. I make no attempt to prove this statistically in the sense of computing the chance of finding these configurations in a sample of random quasars in the sky. Since most astronomers pay no attention to proofs that quasars are local anyway, I might as well present an easy proof rather than a hard one. Instead of doing complicated statistics on heterogeneous searches, I simply reason that if the five dens- 64 Distribution of Quasars in Space N' . . 10' Figure 5-1. laenti/icanon of all quasars mthin an area heated about 2° SE o/NGC 3810, found from objective-prism spectral searches bj Arp and Hazard. Redshifu are written next to position of quasars. NGC 4 5 0 SW -2°/30' DEC 3'/OO' Figure 5-2. A plot of all quasars found within an area southwest of N G C 450 searcKed for ultTavioletexcess candidates by Surdej, Swings, and Arp. 20 - N 10 i i i i Radio quasars V>18.0 — rn 0 6 N U_ 2 1 L__ 1 1 i 1 1 1 Densest groups V2:18 - •• • •• • I I.• • • • • i * i .8 1.2 1.6 2.0 2A i • 2.8 3.2 figure 5-3. The distribution of quasar redshifts for the densest groups compared to the distribution of redshifts for radio quasars from all over the sky. est groups in Table 5-1 were random associations, they should have the same properties as the average quasars in the sky. They have, in fact, outstandingly different properties. For example, Figure 5-3 and Table 5-1 reveal that there is a clear preference for redshifts between 0.8 < z < 1.2, whereas radio quasars from all over the sky are rather evenly distributed between 0.4 < z < 2.2. There is also a tendency to find pairs of quasars within each group. (Pairing tendencies for quasars are discussed further in the Appendix to this chapter.) On the average, the pairs within these dense groups are separated by only 4.6 arcmin on the sky and only 0.07 in redshift. The chance of finding such pairs by chance within the general population of quasars is only about 10~3 per pair. Yet we find eight such pairs in the five dense groups. Even allowing for the fact that these are selected dense groupings of quasars, there is clearly a significant physical association of pairs. Their average difference in redshift of A z = 0.07, however, translates to a velocity difference of 21,000 km s'1, a clearly impossible value for objects belonging, on the cosmological interpretation, to the same cluster or supercluster of galaxies* The physical reality of these dense groups destroys the possibility that quasars are at T h e one pair among the 26 quasars listed in Table 5-1 which has very closely similar redshifts was recently pounced upon by advocates of conventional quasar distances. They proclaimed this pair to be the result of a gravitational lens of enormous and unprecedented mass sitting invisibly out somewhere in this direction in the universe (Nature 321, p. 142, 1986). Embarrassingly enough, further observations revealed almost immediately that the spectra were in fact enough different so that the quasars had to be two separate, albeit quite similar quasars. This is just the pairing tendency evidennt in the quasars discussed in this chapter and evident in the compact, excess-redshift ejecta of active galaxies discussed in following chapters. The above, widely publicized incident of marvelous discovery and almost instant refutation, however, underscores two points: (1) no result in the field is absurd enough to provoke reexamination of basic assumptions, and (2) selecting one aspect of data which supports a hypothesis and ignoring other aspects which contradict that hypothesis is a form of "altering data". 66 Distribution of Quasars in Space their redshift distances, because, as just ex- optically faint radio quasars in the North Ga- plained, the redshift range of the associated lactic Hemisphere (NGH radio quasars in Fig. quasars is too large by a wide margin. A typi- 5-4) are associated with the Virgo cluster of cal cluster of galaxies near redshift z = 1 galaxies, i.e., the center of our Local Super- would have redshifts ranging, at most, be- cluster. This 1970 study is now translated into tween 0.99 < z < 1.01. the upper-left panel in Figure 5-4 where we But at the same time, the existence of these see that the brightest quasars belonging to the groups is the key that unlocks a more detailed Virgo cluster are concentrated in redshift understanding of the puzzling data on quasars. near z ~ 1. There are almost no quasars near z The reason that it is such a key is that it al- ~ 2. Presumably, they are too faint in appar- lows us to ask a crucial question. (It might be ent magnitude to be observed in Virgo! said that the most difficult part of research is In contrast, the same sample of quasars in not to get the right answer, but to ask the the opposite direction of the sky (SGH radio right question.) The question is: quasars) show a strong concentration near re- "Why do the densest groups of quasars have dshifts z ~ 2. This represents gross differences redshifts near z = 1?" in the quasar distribution within a complete and homogeneous quasar sample over the sky. B. The Intrinsic Luminosities of Quasars of Different Redshifts On the cosmological interpretation of quasars this would require an enormous violation of the usually assumed cosmological principle One obvious answer to the question of why the densest groups of quasars have z ~ 1 is: "The luminosity of quasars with z = 1 is greater than the luminosity of quasars of other redshifts." In that case, the z = 1 qua- that on large distance scale the universe is homogeneous. It is astonishing that this clear evidence contradicting the cosmological assumption has lain ignored and uninvestigated for over 15 years! sars could be seen at a greater distance, where To return to the luminosity-redshift rela- the scale of their separation appears smaller tion for quasars, we can easily draw in the line and the groups therefore appear denser. satisfying the average values for the NGH ra- Can this be tested? Yes, a straightforward dio quasars in Figure 5-4. This "roof-shaped" way to test this is by plotting the apparent relation is shown as a line in the two panels magnitudes versus the redshifts for all the var- in the middle of the figure. Even though this ious groups of quasars that I have, over the roof-shaped line is derived from the NGH years, come to believe are physically associ- quasars in the upper-left-hand panel, we do ated. This has been done in Figure 5-4. For not draw it in that panel in order to avoid these quasars at the same distance, those of prejudicing the eye with the line. One can brightest apparent magnitude must be the thereby see that the points just representing most intrinsically luminous. We realize from the NGH quasars by themselves clearly de- the upper-left-hand panel alone in Figure 5-4 fine the adopted relation. (A roof-shaped rela- what we should have seen in 1970, that the tion implies that the quasars with redshift z ~ quasars near z = 1 are the intrinsically bright- 1 are the intrinsically most luminous and that est quasars. (Actually, this point was first quasars with both lower and higher redshifts stressed in my contribution to the 1972 Kra- are less luminous.) kow and Australian symposia referenced in The fit of this luminosity-redshift relation the Appendix of Chapter 2.) In that same to the quasars studied in the NGC 1097 field Chapter 2, we discussed the 1970 paper in the is shown in the upper middle panel of Figure Astronomical Journal which showed that the 5-4. From the study of NGC 1097 discussed Distribution of Quasars in Space 67 17 V mag. - 19 I 3° $* °e° ° oo° ° o ° °o o o I I NGH RADIO aUASARS_ o _ o 21 17 - . • • V mag. *./ I I I I I I SGH RADIO . aUASARS • ' 19 I i i 21 _ + NGC 1097 + - N \ + \ + •1 + • + °°''Sx , 6 ' * ONo p 0 0 0\ 0 00 . 00 0 o° ^^ \ \ 0 . c o* • 0 • • — SCULPTOR AREA i r NGC 520 17 K XX X X 19 21 DENSE GROUPS 17 o+ 19 21 1 2 z Figure 5-4. The apparent brightness (V mag.) plotted against the redshifl (z) for various groups of quasars believed to be physically associated. From Arf>, Liege Conference. in Chapter 4, we would expect 12 to 15 more tion involving bright apparent magnitude quasars than expected from average back- quasars. This panel cannot be taken too rigor- ground to represent the quasars actually phys- ously because the size of the regions have ically associated with this jet galaxy. Possibly been arbitrarily adjusted. But the Sculptor re- some of the brighter and certainly some of the gion will be considered in much more detail fainter quasars represent projected foreground later and the association of the quasars with and background objects. In that case, we find the nearby Sculptor group galaxies justified in just about a dozen or so quasars which outline detail. the expected relation fairly well. The final panel in the lower left of Figure The other association of quasars discussed 5-4 shows that in the South Galactic Hemi- in Chapter 4 was with the galaxy NGC 520. sphere (SGH radio quasars) the quasars might Those quasars are shown in the upper-right- be fitted by the brightest (nearest to us) roof- hand panel of Figure 5-4. They also appear to relation of all. The direction of these quasars outline a roof relation quite well, particularly defines essentially the direction of the Local the brighter envelope of points. The actual Group of galaxies, the nearest galaxies to us. line has not been drawn in the NGC 520 panel because that would commit us to a zero point for the system, i.e., it would say that we actually believe a certain relative distance for the system. Later, we will encounter confirmatory evidence for NGC 520 being a member of the Local Group of galaxies despite the rather large redshift of the central, disturbed galaxy (z = 2200 km s~'). (We are actually using quasars here as distance indicators for peculiar galaxies!) The establishment and verification of this luminosity-redshift relation for quasars forces us to consider the surprising conclusion that the highest redshift quasars (z ~ 2), instead of being the most luminous objects in the universe as has been heretofore supposed, are actually the least intrinsically luminous quasars. This is a startling development but it enables us to pose the most crucial question of all. It is: "If the quasars with redshifts near z = 2 Quasars from three of the densest groups have the intrinsically lowest luminosities, on the sky as discussed earlier and illustrated then those that have the brightest apparent in Figures 5-1 and 5-2 are now plotted in the magnitudes are the closest quasars to us in bottom right panel of Figure 5-4. It is evident space. Where are they located on the sky?" that they concentrate around redshift z = 1 as if they were more distant objects that had just C. Quasars in the poked above the limiting threshold of discov- ery at V = 19th to 20th magnitude. Local Group of Galaxies Next we come to an area in the sky near the constellation of Sculptor. Later in this The answer to the question of where the chapter, we discuss the evidence for a large lowest-luminosity quasars are located in the physical association of quasars in this area. In sky breaks open the box of contradictory in- the lower middle panel of Figure 5-4 we plot ferences into which the establishment has so the many high-redshift quasars in this group far succeeded in locking this issue. Figure 5-5. found by objective-prism techniques (filled shows the plot of all apparently bright radio circles). In order to get an idea where the quasars, found from radio surveys over the sky, brighter quasars in this area fall we also plot with redshift near z — 2. It is immediately ob- radio quasars (open circles). The quasars in vious that there are three to four times as this region could therefore satisfy a roof rela- many of these high-z quasars on the half of tion moved rather close to us—that is, a rela- the sky toward the Local Group of galaxies Distribution of Quasars in Space 69 than there are in the side of the sky toward the more distant, Virgo supercluster. There really is no way around this result. It requires that many of the known quasars come from galaxies that are among the closest to us. Moreover, this one group of quasars and nearby galaxies spreads over areas of the order of one-third of the visible sky. In some sense, we are imbedded in the edge of the Local Group and would expect to see some of it, at least thinly, in all directions. Since quasars with smaller redshifts are generally intrinsically more luminous and can be seen in more distant, apparently smaller groups in different directions on the sky, and since all this depends on the apparent magnitude level we are looking at, it is not surprising that conventional, nondiscriminating analyses can make, and have made, almost any statement they please about quasar distribution on the sky. In contrast, the lesson we have learned so far in this section is that the intrinsic luminosity of a quasar depends on its redshift, and that at any given redshift, its apparent magnitude depends on its distance. This is strong stuff. Even though I feel it follows ineluctably from the observational evidence, it is still necessary, in the old-time scientific spirit, to test it against all the evidence we can get our hands on. The first test we make is to look more closely at that concentration of quasars in the direction of the center of the Local Group. Figure 5-6 shows a region of the sky more directly centered on the Local Group. There we see an obvious line of quasars extending from mid-upper left to lower right. The line originates from the Local Group companion galaxy, M33! We saw in Chapter 2 that, statistically, quasars tend to be associated with companion galaxies, ostensibly because companion galaxies are a younger and more active variety of galaxies. M33 is the famous spiral in Triangulum with spiral arms composed of young, blue stars and glowing hydrogen gas. It is the most conspicious companion to the dominant galaxy in our own Local Group. M33 is the nearest spiral galaxy of this type to us. Now, we see the nearest quasars to us emerging on a line from this object. I really do not know which is the more exciting, seeing this much maligned idea of local quasars exonerated so dramatically, or the shock of confronting this new and greater mystery of what the quasars are and how they originate from M33. There is, of course, the naturally following question of what else is associated with M33. A similar graphical analysis to that of Figure 5-5 and 5-6 shows that there is also a concentration of low-redshift (0.27 < z < 0.47) quasars near M33. But they are all brighter, about 2 magnitudes brighter, than the high-redshift quasars that we have just seen associated with M33. Now we check against the "roof relation derived from Figure 5-4 and see that the luminosities of these lower-redshift quasars are required to be just about 2 magnitudes brighter! (See Fig. 5-8.) So the quasars found associated with M33 confirm this relation. Not only do we see quasars of low redshift in this region southeast of M33, but we also see certain radio galaxies of the same redshift. Since many of the low-redshift quasars have fuzzy edges when closely inspected on good photographs, these radio galaxies, which have somewhat fuzzier edges, are physically similar and form a continuous class. (The naive insistence of the cosmological group that any spot in the sky that is fuzzy has to be at its redshift distance will be discussed in forthcoming chapters.) But as Figure 5-7 shows, the exciting fact about the distribution of low-redshift quasars and radio galaxies is that not only do they also form an elongated group southwest of M33, but also that the direction of that elongation is rotated slightly counterclockwise from the line of high-redshift quasars 70 Distribution of Quasars in Space 30 15 Dec 0- 2.38 o o 2.05 O229 3 2 0 23 22 21 20 236 % 203 O207 no o° Q180 Q 233 o 196 0 76 81 o2i i" o - O2 11 59 °15S 09 198 O 0 88 -15 - 217 o o'56 -30 1 30 o'72 15 - Dec 0 - ' O2.I8 'o 2.03 9,.6O 1 \ i ' °2.H o>52 -15 o222 -30 o220 1 18 17 16 15 14 13 12 R.A. 212 K 196' c / 12 0 2.35 - o Figure 5-5. A plot o/higfi-reds/u/t quasars all around the sky. Redshifts are written next to the position of each quasar. The center i of the Local Group of galaxies is at 10 rougWy R.A. = 0h 40m, Dec. = +41° What this must mean is that the quasars are indeed ejected from galaxies, as we saw in Chapters 3 and 4, but that ejection direction does not stay fixed in space. As a function of time we would expect it to rotate because the ejecting galaxies rotate. Quasars ejected in one direction should also be older than quasars ejected in a later direction. This implies that what are now the lower-redshift quasars were ejected earlier and that as time progressed they became more luminous and evolved from high redshift to their present low redshift. That is, the quasars became more like peculiar, high-redshift, companion galaxies. This represents the most provocative direction in which to follow up these results, as will be discussed later. But at this point I would like to cement absolutely the QUASARS 0.3^ S,, «7.0 U0 \ . — — -« 50 45 - 0 ' 0 ~~ ^ ^ \ s . b=-10°/' 30 DEC. 15 - / / X 0 ^M3V X •% o ' v- X x o 0 0 o /A \ 0 0 >«. '* o* 8 o°°, 0 o o "b X Figure 5-7. The distribution 0 0 * 00 0 0 o 0 0 0 0 - toward M33 of high-redshift quasars (open circles), bright, low-redshift quasars (filled -15 1 0 O| I I 1 1 1 ° I circles) and low-redshift radio 5 4 3 2 1 0h 23 22 21 20 galaxies (crosses). unique nature of the distribution of the quasars around M33 and their association with M33. First of all, there is the point about the reality of the concentration and alignment of quasars southeast of M33. Accepting its reality, as we shall see in a moment, leads to disaster for the conventional viewpoint. Given this end result, the usual procedure for the establishment, as in the precedent of previous events, would be to perform a statistical analysis on the distribution in which boundary assumptions would be adjusted until they yielded a nonsignificant result. Fortunately, this has been forestalled by a sophisticated and thorough statistical analysis performed by the astrophysicists J. Narlikar and K. Subramanian at the Tata Institute in Bombay, India. They show that the quasars in Figures 5-5 are significant concentrations, with probabilities only about 10"4 of being random distributions, and that the quasars are also significantly aligned by the same large factors of significance. This is an important piece of mathematical analysis, but for myself, and I suspect for many readers, simply the visual evidence in Figures 5-5 and 5-6 demonstrate quite forcefully that there is a real alignment of quasars emerging to the southwest from M33. In order to see exactly what this quasar alignment consists of, we outline the greatest concentration of high-redshift quasars in the vicinity of M33. We then display all the radio quasars present inside this region in the apparent magnitude-redshift diagram in the top panel of Figure 5-8. We use radio quasars throughout because they are drawn from radio surveys, which are generally homogeneous all over the sky at any given declination. In the bottom diagram we display all the radio quasars in a much larger comparison area in the opposite direction in the sky (the Virgo supercluster region). We see that the distributions are completely different. The most important difference is the large clump of high-redshift quasars which are relatively bright in apparent magnitude (around V ~ 18 mag.) which are present southwest of M33 but are essentially absent in the opposite quadrant of the sky. These high-redshift quasars are simply the closest large group of quasars to us, associated with the Local Group galaxy M33. Given the reality and uniqueness of the quasar alignment with M33, the last remaining escape for the cosmological adherents is to claim, "selection effects." The disproof of this possibility is very cutting. The reason is 72 Distribution of Quasars in Space 15 16 V mag. 17 18 19 M33 SW oo o o a O O O Q O % Q D OO 15 16 V mag. 1 7 18 19 o o o ego 0 O o oo oo NGH Figure 5-8. Inside the area southwest of M33 are shown all radio quasars in an apparent magnitude^edshift diagram. Below is shown a larger comparison are in the opposite direction of the sky. that all the quasars we are dealing with are undeniably real quasars—spectra which identify the redshifts exist for all of them. Therefore the only possible way to make the concentration southwest of M33 disappear is to discover many more high-redshift radio quasars in other directions all over the sky. But, it is preposterous to suppose that only radio quasars of high redshift were selectively not observed in other regions of the sky. Yet this is what the conventional interpretation must claim in order to save the situation. It should be made very clear that they have the responsibility of producing, and publishing, the spectra of these missing quasars near z— 2, either that or admit their distribu- tion is anomalous. If they admit the latter they will have a grotesque inhomogenity on the largest scale of the universe pointing at the observer (because the inhomogenity contains a range of redshifts around z — 2). They would also have to ascribe the alignment with M33 as an accident. Two further interesting comments can be made about Figure 5-8: One is that the quasars in the top panel are all very bright in apparent magnitude. That is as required by their belonging to a galaxy as close to us in space as M33. In fact, the "roof relation derived from the analysis earlier in this chapter is confirmed in Figure 5-8 by the fact that the quasars of redshift z ~ 1 or less average about two Distribution of Quasars in Space 73 magnitudes intrinsically brighter than quasars around z = 2. The other interesting feature of the quasars southwest of M33 is that very few quasars of faint apparent magnitude appear in this area. It is as if we were seeing a cloud at the distance of the Local Group of galaxies and then a relative void beyond. The "cloud" distribution is a necessity on both the cosmological and local interpretation of quasars. That is because Olbers' paradox would demand an infinite sky brightness if they' were not in clouds. (Olbers merely pointed out that a uniform distribution of luminous objects extending to an indefinitely large radius in space would necessarily lead to an indefinitely bright sky instead of the dark night sky we observe.) In the cosmological interpretation, the catastrophe is avoided by invoking "evolution" of quasars. That is, beyond a certain distance, the conventional viewpoint resorts to simply turning out the quasars (saying they are too young to have formed). So, they are dealing with a limited cloud, but one of very large dimensions. The local- interpretation, we see, simply has smaller dimensions to its clouds of quasars. Since these local quasars are generally less luminous than the galaxies they are grouped with, they avoid an infinite sky-brightness in the same natural way as do the groups and clusters of galaxies to which they belong. But there is an added fillip to this necessary cloud picture: Fred Hoyle pointed out an absolute mathematical disproof of the cosmological nature of quasars. Mathematically he showed that in order to reproduce, on the cosmological assumption, the observed numbers of quasars as a function of redshift, their luminosity function must be very steep. (That is, at a given distance, or redshift, the number of quasars must increase rapidly as their luminosity decreases.) All the conventional analyses indeed require this very steep luminosity function. Yet as Hoyle points out, the observations strongly violate this requirement. You can see a dramatic example of this in the top of Figure 5-8. Between redshifts 0.2 :£ z S 1.0 there are a number of quasars between apparent magnitude 16 ^ V S 17 mag. If the conventional luminosity function were really valid, then we should observe ten times as many quasars in the redshift interval between 18.5 :£ V ^ 19.5 mag. Actually, we observe practically none. There could scarcely be a more clear-cut observational contradiction of the cosmological requirement. It may or may not be hard to believe, but many astronomical research centers do not even have the Hoyle publication. One center where I brought it to their attention made the reply that: "Well, the mathematics are correct but the observations are not good enough to test the claim." It would be a marvelous confirmation of what we have learned about quasars in the Local Group if we could look at the next most distant group of galaxies and observe something like the same phenomena at a correspondingly smaller scale and fainter apparent magnitude. That opportunity is presented to us by a group of galaxies located in the constellation of Sculptor. D. The Sculptor Group of Galaxies A group of galaxies in the southern hemisphere of the sky lies roughly 2-3 times farther from us than the Local Group galaxies such as M33 which we have just discussed. Two dominant galaxies define the Sculptor Group. One is the impressive spiral galaxy NGC 300, which is of much the same type as M33 yet still close enough to see distance-indicating Cepheid variable stars in the arms. The other, NGC 55, has comparable size but more irregular shape. Now an extraordinary stroke of good luck occurred when two astronomers at the U.S. National Observatory in Chile, decided to observe a sample of quasars. They picked a declination zone that ran high overhead for them (Dec = -40°) and observed a 74 Distribution of Quasars in Space STRONG LINE WEAK LINE TOTAL- 116 STRONC LINE QUASARS TOTAL • «« Figure 5-9. The numbers of WEAK LINE OUASARS TOTAL* 32 objective prism quasars in the Dec = -40° zone. Note that the proportion of strong-line quasars rises just at the position of the Sculptor group galaxies NGC RIGHT ASCENSION (HRS) 300 and NGC 55. long, narrow strip of sky, 5 degrees wide, run- But I noticed that the quasars at the ends of ning from west to east. The good luck was the strip contained proportionally more that this strip runs right across our Sculptor "weak-lined" quasars. That is, the emission galaxies NGC 300 and NGC 55. The begin- lines identifying them as quasars were fainter ning and the end of the strip lies outside the and they were consequently more difficult to Sculptor group and can be used to compare to discover. If the photographic emulsions were the results in the center of the strip. really less sensitive at the ends of the strip, An uncomfortable result became appar- then a proportionally smaller number of these ent as soon as they plotted their results. A quasars should have been found rather than a good many more quasars were found in the larger number. When I tried to publish this center of the strip than at the edges. Since result in the British journal, Monthly Notices these two astronomers accepted unquestion- of the Royal Astronomical Society, it was sent to ingly that the quasars were out at the far one of the original two authors to referee. reaches of the universe, this result obviously Needless to say, it was not published. It was could not be correct. Therefore, after the fact, almost stopped again when I sent a short ar- they decided that the photographic emulsions ticle to Nature, but, thanks to the last-minute they had used were less sensitive on either intervention of an editor, it finally appeared. end of the strip than they were in the middle The diagram shown in Figure 5-9 is from that of the strip! article and shows how the concentration of This was duly published and accepted strong-line quasars rises dramatically at just Distribution of Quasars in Space 75 -32°DEC -36°- -40°- -44C OBJ.PRISM QUASARS U ; ° O OO o\ u ° ° o \ o ° ° °° o o 00 23 R.A Figure 5-10. Plot of the high-redshift quasars in the region of the Sculptor Group galaxies. the position of the Sculptor group galaxies, NGC 300 and NGC 55. It is interesting that we have here more than just a concentration of quasars in this region. We have a concentration of a particular kind of quasar. (The Appendix to Chapter 1 references articles which discuss factors of 10 density discrepancies across this region.) I was able to measure more quasars north of the Dec = -40° strip, and thus extend the areas of homogenous quasar discovery to the next strip around NGC 300 and NGC 55. These results are shown in Figure 5-10. The quasars in this entire area are now seen to have a very interesting distribution. Of course, there is the general excess of quasars demonstrated by Figure 5-9. But in addition these excess quasars tend to group around the major galaxies, NGC 300 and NGC 55. The densest distribution in Figure 5-10 forms a line southeast of NGC 300 about 9 degrees long. In Figure 5-7 we saw a similar elongated alignment of quasars emerging from M33. Of course, the line from M33 was about 5 times longer in angular extent, but M33 is less than half as distant. These lines also must have arbitrary projection angles in space. Finally we can check the apparent magnitudes of the quasars in these lines, and as Figure 5-11 shows, the NGC 300 quasars are just about 1.5 magnitudes (a factor of two in distance) fainter. Therefore we see that in both the closest and next closest groups of galaxies to us in space we identify elongated distributions of quasars emerging from the major spiral galaxies in each group. Moreover, the scale of the distribution on the sky and apparent magnitude of the quasars supports the known fact that the second group of galaxies is just about twice as distant. Of course, lines of quasars are exactly what we observed in the ejecting galaxies NGC 1097, NGC 520, and M82 in the preceding chapter. • E. The Quasars Belonging to M82 It is more than interesting now to re-examine the third most distant group of galaxies from us. That is the M81 group at about 1 magnitude greater distance modulus than NGC 300. The major active companion in that group, M82, was found to have an "ejection cone" of three quasars with redshifts around z ~ 2 emerging from it (Chapter 4). This is the beginning of a line, or elongated distribution. These quasars are very faint, between 20th and 21st magnitude, as shown in Figure 76 Distribution of Quasars in Space 17 18 19 20 21 QUASAR APPARENT MAGNITUDE 5-11. So we apparently have a glimpse, at apparent magnitudes just above the plate limit, of a similar line of high-redshift quasars emerging from an active galaxy in also the third most distant group of galaxies. Figure 5-11 demonstrates that the apparent brightness of their associated quasars scales accurately as would be required of galaxies at these three different distances. Summary What we have done in this chapter is to account for large numbers of quasars. Referring to the numerous quasars mentioned in the beginning of the chapter that were not obviously associated with any galaxy, it turns out now that many are, in fact, associated with galaxies—it is just that the galaxies are so close that the associations stretch over large areas of the sky. It also turns out that the high-redshift quasars around z « 2 are not generally seen beyond the relatively nearby M81 group. That means that we have highredshift quasars which belong to M33, and possibly to other members of the Local Group, projected over large areas of the sky, probably quasars that belong to our own galaxy which could be projected in almost any direction in the sky, plus various quasars that are contributed by more distant galaxies and groups like Sculptor and M81. It will require some study and discrimina- Figure 5-11. Comparison of op/wrent magnitudes of high redshift quasars in M33, NGC 300, and M82 lines. See Arp (1984, Fig. 4) for details. Note that the high redshift quasars associated with the most distant of the three galaxies, M82, are almost at the discovery limit. tion to sort out which quasars belong to which groups; possibly, in a number of cases, we can never be completely certain. The picture is further complicated when we consider quasars with redshifts around z = 1. These are intrinsically brighter and can be seen at greater distances. Therefore they contribute some very bright apparent magnitude quasars which belong to extended nearby groups and in addition contribute fainter, smaller-scale groupings. But the smaller-scale groupings appear in regions of the sky which do not have much relation to the nearby, larger groupings. What is clearly needed now are careful, homogeneous quasar searches with various techniques to uniform limiting magnitudes all over the sky. Then detailed interpretation can be made. This is the necessary, hard scientific work that should have been undertaken long ago. It is simply irresponsible to say, "I know all the important things about the universe—I do not have to study it. I need only to take a sample here and a sample there and announce that my model is now correct to several decimal places." This approach reminds us of the blind men feeling the elephant. One feels the leg and says an elephant is a tree, another feels the trunk and says it is a snake. In current extragalactic astronomy, one group has gone a step further by trying to eliminate all the others and to be left proudly waving the tail and proclaiming, "There is no uncertainty about the answer now." Distribution of Quasars in Space 77 Before we go on to the next chapter, however, we have to face a difficult and challenging problem. The problem is, namely, that associating the quasars with nearby galaxies means that their redshifts cannot be due to the Doppler effect of a recession velocity at great distances in the universe. Some other explanation for the high redshifts of the quasars must exist. What this explanation is, is a matter of spirited debate, as it should be, among that small band of astronomers who believe in the noncosmological redshifts of quasars. We will come to this animated discussion in a few chapters, but before that I would like to talk about nonvelocity redshifts in galaxies. That's right, galaxies. The reason for this is that quasars are small, mysterious objects that invite speculative theories. Perhaps they are redshifted by a strong gravitational field, perhaps by high ejection velocities, perhaps by robbing the photons of Appendix to Chapter 5 their energy by some scattering process. Some theorists have been intrigued with the idea that quasars are distant objects which are gravitationally lensed, and amplified in brightness, by galaxies near their light path. But evidence points to the fact that galaxies can also have nonvelocity redshifts. If this is true it presents us with objects we know vastly more about. We can actually study rotation, dynamics, and chemical composition of the constitutent stars in many galaxies. How could an entire galaxy have a nonvelocity redshift? The answer to this may be even more far-reaching and staggering than the answer to the same question about quasars. And, if the more energetic, compact forms of galaxies can be shown to be physically continuous with quasars then the answer to the redshift riddle for quasars may be the same for quasars as for galaxies. PAIRS OF QUASARS The tendency of quasars to pair has been obvious for a long time. In 1970(Astron. Joum. Vol. 75, p. 1), I pointed out a number of quasars which fell conspicuously closer to each other on the sky than average and had a number of properties such as apparent magnitude, radio properties, and redshifts which resembled each other more than one would expect for randomly occurring objects. Of course, the redshifts were, typically, enough different so that they would invalidate the cosmological redshift assumption if the quasars were actually physically associated. I remember, in the early days, Fred Hoyle discussing the obvious similarities between 3C286 and 3C287, two quasars close together in the sky. Today, it is impressive to run your eye down modem lists of quasars and see how many obvious pairs stand out. This phenomenon was quantitatively investigated by G. R. Burbidge, E. M. Burbidge, and S. L. O'Dell in 1974. They demonstrated, using only the few very close quasar pairs known at the time, that the redshift differences could not be reconciled with redshifts indicative of distance. Poof! There went the cosmological hypothesis! Well, one would have thought so, but it was privately stated with calm assurance that this calculation did not count because the test was made after the quasars had been discovered. This is the old "a posteriori" argument which was now further deformed to say, "You cannot test any data that already exists." Undaunted? Well, I cannot say, but Burbidge and Narlikar nevertheless recently went on to make the calculation using all the quasars discovered after the time of their first analysis. They now obtain a probability ~ 1O\ less than one chance in ten thousand that these pairs can be accidental (see references following). An illustration of how things work in this game was accidentally revealed to me shortly before this last Burbidge and Narlikar paper. An astronomer analyzing this quasar data found a disturbingly significant excess of these same pairs. He said, "Well, this is obviously a selection effect caused by astronomers looking in the vicinity of radio quasars for nonradio quasars." (Of course, these cases are minuscule in number and could be easily identified.) But as he "normalized out" this effect and sent his paper proclaiming another proof of the cosmological nature of quasars off for instant publication, he smiled at me and said, "Gee, Chip, I really would love to find some hard evidence for noncosmological quasars." The latest paper on this subject (European Southern Observatory Preprint No. 422) purports to refute the physical pairing of quasars of different redshift:. But the analysis mixes physical groupings of widely different distances and hence widely different characteristic separations. Even so, close inspection of the graphical results shows consistently more close separations than expected. The effect would be even more conspicuous if the fitted line had been drawn accurately through the points representing large separation. 78 Distribution of Quasars in Space Redshift Periodicities in Different Groups of Quasars Now that we have established the existence of different physical groups of quasars, we can take another, more illu- minating look at the preferred values of quasar redshifts that are discussed toward the end of the first chapter. We saw there that the three quasars in NGC 1073 fit almost exactly the mean periodicity of all quasars taken as a whole. But other physical groups of quasars can have slightly different periodicites. Because the spacing between periods follows the rule that the intervals in the logarithm of (1 + z) are constant A (tog 1 + z) = const.), a group with a slightly different constant will have intervals which become progressively1 larger as larger redshifts are considered. An example of this is shown in Figure 5-12. i i 0.35 1 f 1 o tt 0.30 0.60 1.01 1 t 0.96 1.56 1 t 1.41 i i 0.4 0.8 1.2 1.6 z=REDSHIFT t 1.96 1 2.0 2.25 i } OBJECTIVE PRISM NGC 3842 NGC 1073 } ALL QUASARS 2.4 Figure 5-12. The preferred periods for all quasars as a whole are shown along the bottom of the strip. Preferred periods for the objective prism quasars (which are dominated by the Sculptor group quasars) are shown along the top. The points indicate how well the triplets of quasars discussed in Chapter 1 fit these preferred periods. The quasars selected by objective prism searches (Box and Roeder, see Appendix to Chapter 1) are dominated by the large group of quasars associated with the Sculptor group of galaxies discussed earlier in this chapter. We see that the periodicity of this objective-prism selected group of quasars has a slightly larger constant in the logarithm. Another physical group, the three quasars around NGC 3842 shown in the first chapter, fit this larger spacing. In general, each group of quasars we have identified as physically belonging together tends to have either a slightly smaller or slightly larger spacing than the mean of all quasars. This confirms the periodicity, but gives a broadness to the values of preferred redshift for all groups lumped together. Each individual group, however, tends to have more exactly defined peaks of preferred redshift. The final chapter in this book makes some suggestions as to what might cause the fundamental periodicity. What causes the slight difference from group to group? There is only a hint, which will be discussed later in Chapter 9 when we examine Virgo cluster quasars. It is clear that further study and analysis of these periodicities could furnish invaluable insight into these most basic physical properties of matter. References to Papers Covering Material in Text 1981, Oort, J. H., Arp, H., and de Ruiter, H., Astron. and Astrophys., 95, p. 7. This paper investigates the properties that quasars would have if they were at cosmological distances and members of superclusters. I thought I might initiate a healthy precedent in the field by looking at the data from a standpoint different from my own personal belief. The paper demonstrates the masterful astronomical knowledge of Prof. Oort and it was a privilege to write the paper with him, but I must admit that the results which I felt to be important about the separations of high-z quasar pairs being too large to be accommodated by the cosmological model were not included in the paper. 1983, Arp, H., "Groups, Concentrations, and Associations of Quasars' in "Quasars and Gravitational Lenses," 24th Liege Astrophysical Colloquium, Institut d'Astrophysique, June 1983. This review paper developed the data on luminosities of quasars in groups at different distances. The discovery of the line of quasars from M33 first emerged in this presentation. The conference is also interesting to read for the number of papers critical of my general viewpoint about quasars. Distribution of Quasars in Space 79 1983, Hoyle, E, "A Gravitational Model for QSO's," Preprint Series No. 88> Department of Applied Mathematics and Astronomy, University College, P.O. Box 78, Cardiff, England. This refreshingly candid discussion of the quasar problem argues for a gravitational mechanism as a cause of the redshift. I do not personally subscribe to this explanation, but the discussion of quasar phenomena is rigorous and informed and the science is top-notch. This monograph is in very few libraries and must be obtained from the University of Cardiff. 1984, Arp, H., "The Nearest Quasars," Publ. Astron. Society of the Pacific, 96, p. 148. This paper investigates the line of quasars southwestof M33 and shows how low-redshift quasars and radio galaxies are rotated slightly from the line of high-redshift quasars. Of great importance in this paper is the discussion, the most fundamental I could make, of the arguments previously advanced to prove that radio galaxies had to be at their redshift distances. 1984, Arp, H., "A Large Quasar Inhomogeneity on the Sky," Astrophys. Joum. (Letters), 277, p. L27. This paper offers additional evidence that the quasars southwest of M33 cannot be an accidental fluctuation in a uniform distribution and that this observed distribution makes no sense on the cosmological model. A sad by-note is that George Abell, who was a classmate of mine in my graduate-school days, was refereeing this paper. One of the last letters he wrote before his untimely death said that he could not explain how it occurred, but he was sure the alignment was some sort of selection effect. 1984, Arp, H., "Distribution of Quasars on the Sky," Journ. of Astrophys. and Astronomy, 5, p. 31. This was an invited paper by V. Radakrishnan for the Jubilee Issue of the Journal of Astrophysics and Astronomy. It developed further the picture of the Local Group, quasars with z~2 and established consistency between the line of quasars in the Local Group and the next nearest Sculptor group. It showed that inhomogeneities of distribution of the quasars were reflected in inhomogeneous distribution of radio sources. The radio sources have, in the past, been supposed to be uniformly distributed and it is very important to now reinvestigate this important question. 1985, Narlikar. J. V. and Subramanian, "A Statistical Significance of a Large Quasar Inhomogeneity in the Sky," Astron. Astrophys., 151. p. 264. This shows the concentration of quasars southwest of M33 to be highly significant. 1985, Burbidge, G. R.. Narlikar. J. P., and Hewitt, A., Nature, 317, p. 413. This paper gives the most recent calculation of the chance of accidentally observing the known number of apparent quasar pairs with discordant redshifts. The probability is now less than one in ten thousand, so that the cosmological hypothesis is dropping back badly from an already very bad, earlier position. See the text here and Nature, 323, p. 185 for the most recent exchange of viewpoints. 80 Distribution of Quasars in Space GALAXIES WITH 6 EXCESS REDSHIFT In the normal course of observing the sky with telescopes, we expect to see galaxies near to each other in groups. When we measure the displacement of the absorption and emission lines in their spectra, we expect to find the redshifts of these galaxies to be very close, differing by only a few hundred kilometers per second (km s"1)- When we do see a much larger redshift, we instinctively feel that it is an unrelated object at a much greater distance in the far background where the expansion velocity of the universe is carrying it away from us more rapidly. It is an enormous shock therefore when we measure two galaxies that are interacting, or connected together, and find that they have vastly different redshifts. That is what happened when I measured the redshifts of the two galaxies pictured in Figure 6-1. It was 1970 and Palomar observers still had to ride all night in the cage of the 200-inch telescope in order to obtain direct photographs and spectra of astronomical objects. An observer was usually lucky to get two spectra in a night of objects as faint as the ones in Figure 6-1. But I was following up interesting objects from my Atlas of Peculiar Galaxies, and I was interested in that class of objects where companion galaxies were found on the end of spiral arms. As in the case of the quasars, this study led to big trouble when I discovered the redshifts of the two connected objects differed by Az = 8,300 km s'1. The trouble lies in the fact that one cannot even postulate that by some freak accident two galaxies are in the same region of space and that the smaller galaxy is running past the larger galaxy with a relative velocity of 8,300 km s"'. At that rate of passage the companion would be unable to pull out a filament from the larger galaxy. The gravitational pull needed to shear stars out of their normal orbits cannot build up in the relatively short time that such a rapid encounter would allow. We therefore conclude that the objects cannot have this velocity difference and we are back to an object with an intrinsic redshift. Only this time, it is not just a compact object like a quasar, but a whole galaxy of stars, each star of which must share, for some mysterious reason, a redshift much elevated above the normal. Of course, the first thing one considers in a case like this is whether this could possibly Galaxies with Excess Redshifts 81 Figure 6-1. The large, disturbed Seyfert galaxy, NGC 7603, with a companion galaxy apparently attached by a filament. The redshifi of the larger galaxy is 8,700 km s-' and of the smaller is 17,000 km s'. These pictures were when by Roger Lynds with the Kitt Peak National Observatory 4-meter telescope in 197i and have recently become available through the information adding of separate pictures by Nigel Sharp. represent a background object that just accidentally happens to appear projected onto the end of a luminous filament belonging to a more nearby galaxy. So back I went into the cage on my next observing run in order to obtain the longest, deepest and best possible photograph of the object. It showed the connection strongly, as did all other pictures like the one shown in Figure 6-1. Unlike the connection between NGC 4319 and Markarian 205 discussed in Chapter 3, nobody ever tried to question the reality of the luminous arm emerging from NGC 7603. What little debate took place in this case hinged on whether its connection to the companion was real or only apparent. A number of arguments attested to a real connection. First of all, the larger galaxy, NGC 7603, is a Seyfert galaxy, a fairly rare kind of galaxy with an active nucleus. Secondly, NGC 7603 is all torn up inside and nothing else is around except the companion to account for this disruption. Thirdly, only one luminous arm or filament emerges from NGC 7603 in such a way as to make it an almost unique object among galaxies. This unusual arm ends right on the companion! All this, naturally, is extremely unlikely to 82 Galaxies with Excess Redshifts be an accidental occurrence. But still I was not satisfied and I studied the companion closely. The original plate showed that the form of the companion, with its broad, bright nucleus and a discretely lower surface brightness disk, is unusual. But the bright rim on the outer edge of the companion, just where the arm from NGC 7603 connects, proved to me that there is actual physical interaction between the two. * But the spectrum of the companion puzzled me. It had only the usual absorption lines found in ordinary galaxies of older stellar population type. As we will see, a number of other examples of these anomalously redshifted companion galaxies tend instead to show emission lines and absorption lines that are more characteristic of a younger stellar population. The evidence from the nature of the connection and the peculiarities of the galaxies establish rather conclusively that the present two are connected, but the spectrum of the companion so far offers no clues as to unusual physical conditions that might account for its 8,300 km s"' higher redshift. This was the only example of a strikingly discordant redshift that turned up this small class of connected companions in the Atlas of Peculiar Galaxies. But the much larger Catalogue of Southern Peculiar Galaxies and Associations, initiated a few years later by Arp and Madore, furnished more examples of smaller companions connected to, or interacting with, larger galaxies. These were the productive days when I used the 2.5-meter Carnegie Institution telescope in Chile to get further photographs and measure spectra of these objects. In the first batch of seven candidates, I found three to have highly discrepant redshift values. Later more turned up. A sampling of these objects is shown in Figures 6-2 and 6-3 here. The luxury of having long runs on the telescope allowed the spectra of these objects to be studied in some detail. There are many extraordinary peculiarities discovered in these spectra which should be followed up. But as a preliminary classification I have noted in the sample of them that are listed in Table 6-1 whether the spectra have emission lines and whether the absorption lines are characteristic of a young or old stellar population. We should understand that if we just go around picking galaxies at random in the sky and studying their spectra that by far the most common spectrum we will find is nonemission with late-type stellar absorption. Table 6 shows, however, that the discordant redshift companions characteristically exhibit excited, emission lines and young stellar population absorption lines. If they were just accidental projections of ordinary background galaxies they would have the characteristic spectra of background galaxies. This is a clinching proof that these excess redshift companions are really physically associated with the lower redshift galaxies. The way in which I would think a rational astronomer would have to handle this disturbing situation is to look carefully at objects like NGC 7603 in Figure 6-1, or at some of the examples in Figures 6-2, 6-3, and 6-4, and say: "Well, these are certainly connected objects, the eye readily tells us they could not be anything else." Then, he puts aside the pictures and says, "This goes against everything I have been taught about redshift-distance laws. They must be just accidents." But then he thinks about the confirmatory evidence from their spectra (and the abundant other evidence discussed in this book) that demonstrates they are not accidents. He then goes back to the pictures and says, "Well, it seems that they are not accidents and when I actually look at the pictures I can see many more confirmatory details which demonstrate that this is just simply the surprising, but actual •Photographs taken shortly afteward in 1973, with the Kitt Peak National Observator's 4-meter telescope confirmed the peculiar bright rim on the companion, its deformed shape and the brightening of this rim near its point of contact with the arm from NGC 7603- The existence of these confirming pictures not known to me until recently. These latter photographs are shown in Figure 6-1. Galaxies with Excess Redshifts 83 • •* - • •-• • * ^t'-.-tj '*•''&'- ••"THP toi-* * *' * '•i 1 * * * * • •HE'*' • " * * * * H% S..7 •*; *' ^i / - ^ "'-'^•'^ # • (&>•. ::-;t*r-3BBi * B L i * '• • • * CO in O §11111 # Oz %.$^ " \,*-, V- '* • > • ;j*K'; * SKf-i - . -< * y * • /«•.. - , ' s i , f'*1;' *<&^* *•"% *••*" *•• * wi' ••*•*'.'.*« . ^ , - : ; d a * . / . . *• •T'. . •_ ; . *• * * • *'.•'".•"• » *>.«•*.'- IMSt*'4£! .1*-.- " ' - •' * * * * * * AM 2054-221 _:-"„. •§swTm Figure 6-2. Four examples of interacting high-redshift companions (marked by arrows). See Table 6-1 for data. Figure 6-3. A particular!)! interesting example of a high rahhift companion (AM2006-295). Arrow points to condensation in arm with 22,150 fern s 'excess redshift. TABLE 6-1 Sample of Connected or Interacting Galaxies with Large Discordant Redshifts* Main Galaxy Companion Type of Spectrum Excess Redshift i (km s) Illustrated in Figure No. NGC 7603 AM0059-402 AM0213-283 AM0328-222 AM2006-295 NGC 1232 NGC 53 AM2O54-221 Comp SE Comp S Comp N Comp S KNSW GalB Comp N Comp E late absorption late absorption strong emission early absorption emission, early absorption weak emission, peculiar absorption emission, early absorption emission, early absorption emission late absorption +8,300 +9,695 + 14,021 +17,925 + 22,350 + 26,210 + 32,774 + 36,460 6-1 Arp. J., 239, 471 6-2 6-2 6-3 6-4 6-2 6-2 *As pictured in Figures 6-2, 6-3, and 6-4. Ft r a more complete listing of cases see Astruphysical Journal. 263, p. 70. "Late absorption" described a spectrum typical of low-iuminos ity, older stars, "Early absorption" is a spectrum typical of high-luminosity, younger stars. 86 Galaxies with Excess Redshifts way things are." It is a rare occasion when a person, even a scientist, is able to really look at a picture without forcing it into a frame of prior reference. It is, in fact, instructive to look more closely at these pictures. For example, in Figure 6-1 one sees two (stellar appearing?) condensations in the filament leading to the companion. This is even stranger behavior for a part of a galaxy to exhibit. What are they? What would spectra with a large telescope and the newest spectrographs reveal about them? Then there is the three-armed spiral galaxy shown in Figure 6-3. Do you realize how rare three-armed spirals are? And this third arm originates from a point midway out along one of the two symmetrical arms! The discrepant, redshift object occurs in the middle of this third arm. How did it get there? These questions call for a working hypothesis within which we can try to organize the disparate facts presented by the observations. But we have already described an ejection hypothesis that was needed to explain the origin of quasars in the earlier chapters of this book. There we saw that the compact objects that were to become quasars, the protoquasars, must emerge from the galaxy's nucleus as relatively small, high-redshift objects and later expand, with their redshifts decaying and the objects becoming more like compact, peculiar companion galaxies. There are several advantages to this hypothesis for the discordant companions described here. (I stress that this is only an empirical hypothesis at this stage— a working scheme to connect the various observations which cannot be explained by the current theories about galaxies.) One advantage is that only one explanation would have to be invented to explain the excess redshifts of both quasars and companion galaxies. If there is a continuous physical evolution between the two, as there appears to be a continuous range of physical properties between them, then the same mechanism for nonvelocity redshifts, in differing amounts, could explain both. At this point, two comments should be made: (1) The companions with the highest redshifts which are discordant (45,000 km s1 corresponds to a z =0.15) are in the same redshift range as the smaller redshift quasars. (The brightest apparent magnitude quasar in the sky, the famous 3C 273, has z =0.16.) (2) The companions with the highest excess redshifts are spectroscopically the most like quasars. The puzzle I encountered in the spectrum of the companion to NGC 7603 could be resolved if the spectra with relatively small excess redshift are relatively normal, but the spectra become more abnormal as the excess redshift grows. A glance down Table 6-1 shows that with increasing excess redshift, the spectra contain more excited emission lines and younger stellar absorption lines. Quasars, are of course, characterized by high temperatures and conspicuous emission lines, that is, they are the most extreme of these objects. As for the young stellar absorption spctra, extensive observations of the prototype quasar, 3C 48, enabled the astronomers J.B. Oke and T. Boroson to proudly announce that if had the underlying absorption spectrum of a galaxy—but the absorption spectrum is of the young stellar type! This last point about underlying galaxies is an interesting and much belabored one. A number of astronomers who have attempted to defend the status quo have tried very hard to prove that underlying all quasars are galaxies. The idea was that there might be some uncertainty about quasars because they are exotic objects, but that galaxies are familiar kinds of objects which must be at their conventional redshift distances. Proving quasars to be galaxies, they felt, was proving quasars to be at their redshift distances Oust in case it needed to be proven, which, of course, they claimed was unnecessary). But what has been apparent from the observations all along— even before many of the present investigators started their work—is that quasars and compact, active galaxies are continuous in phyisical properties. So, by proving again that they Galaxies with Excess Redshifts 87 are related kinds of objects they have proven that the evidence for nonvelocity redshifts in quasars is supported by the evidence for nonvelocity redshifts in galaxies—and vice versa. It is a cruel fact of life that whatever the current, official theory is, it must explain all the observed facts. A single well-founded, contradictory observation will suffice to topple the whole edifice. But we have seen that the conventional theory that galaxy redshifts can only be due to Doppler velocity has been violated not just once, but in numerous, independent instances. We will continue to see these violations accumulate. Figure 6-4. a) The spiral galaxy NGC 1232 with a companion near the end of one spiral arm (gal A) and a small companion near in inner spiral arm (gal B). b) Insert of photograph in infrared wavelength sliou>ing compact blue companion and disturbed arm nearby. As of 1982,38 examples of these discordant redshift companions around 24 main galaxies had been published. We cannot discuss them all here but the references for this data are given in the Appendix. One example is so interesting, however, that I cannot resist devoting a few pages to it. A. The Large Spiral Galaxy NGC 1232 and its Two Discordant Redshift Companions As Figure 6-4 shows, NGC 1232 is a large, beautiful spiral. The companion galaxy near the end of the spiral arm shows the same resolution of knots and features as the main 88 Galaxies with Excess Redshifts galaxy. It is the usual kind of galaxy that large spirals typically have as physical companions. One can even trace the arm which ends near the companion back to the main spiral where it splits, strangely, into a channel about the width of the companion. I would suggest this might be evidence for the companion to have originated within the main spiral and have traveled outward along this arm. Be that as it may, there was never any hesitancy about accepting this galaxy as a run-of-the-mill companion to a large spiral galaxy. Scarce heed was even paid to the fact that the redshifts of both galaxies were cataloged as essentially equal. But then an unpredictable event occurred. The cataloged redshift for the companion was found to be in error. The redshift of the companion was really 4,776 km s"1 greater than the main galaxy. One of the foremost galaxy experts of this era, Gerard de Vaucouleurs, who has been rather more open to discrepant evidence than most, nevertheless had the following comment about this development: "Until recently I was convinced from appearance and resolution that this was a physical pair, in fact rather similar to our own galaxy and the Large Magellanic Cloud. However, the differential velocity, A V = +4776, forces us to conclude that this must be an optical pair unless you can offer compelling proof that the two are physically connected." My own argument was that the companion was not the kind that was found isolated in space, but was of the kind found with larger spirals such as NGC 1232. Nevertheless, I did take very deep photographic plates of the system in order to search for "compelling proof," perhaps in the form of deformations or extensions of the outer regions of NGC 1232 in the region of the companion. I found no further evidence for the association of the companion than was available originally. But I did find something else that turned out to be enormously interesting. Tracing back along the same spiral arm which ends on the companion I noticed an anomalous thickening and deformation, as if something was perturbing the arm at that point. Next to this disturbed part of the arm, I noticed an object. Purely out of curiosity, and fully expecting an H II (gaseous emission) region at the redshift of the main galaxy, I took its spectrum. There was thereupon one of those rare and thrilling moments in research when you can look down a long corridor into the future. The spectrum showed a redshift of over 28,000 km s"1 (almost onetenth the velocity of light!), far exceeding the mere 1,776 km s"1 of the main galaxy. Studying the spectrum was fascinating. There were six separate kinds of peculiarities that indicated the object cannot be any normal kind of background galaxy. Among them the fact of the narrow calcium (K) absorption line implied it was a low luminosity galaxy. But the most compelling argument that the object was at the distance of NGC 1232 was simply that one does not see background galaxies through the disk of a spiral galaxy. The dust and obscuration in the disk of a spiral galaxy, particularly near the arms, simply forms an impenetrable screen. If, by some strange quirk, we were able to see through a thin part of this screen, we would certainly expect a background object to be heavily reddened. But galaxy B, as we call this object, is extremely blue! It is so blue, in fact, that there is no way that it can be any kind of normal galaxy. This evidence, even though of the most detailed and quantitative kind, has always been simply ignored. One of the purposes of this book is to bring all this evidence together, to show that it massively contradicts the accepted paradigm, and to challenge the establishment to deal in a responsible scientific manner with the observations. In view of our empirical, working hypothesis we can point out that one way an ob- Galaxies with Excess Redshifts 89 ject like B could arrive at a point in a spiral arm, as observed, is to travel along the spiral arm from the nucleus. It is interesting that this is the same arm along which it was speculated that companion A may have emerged. Perhaps even more suggestive is the knot (KNSW) in the three-armed spiral pictured in Figure 6-3. (That was another of those exciting moments where I took the spectrum expecting a low-redshift emission region.) But that third arm, in my opinion, could only have been formed by an ejection phenomenon along the original track of one of the two main arms. The discordant redshift object then lies along the trajectory of that ejection. There is nothing sacred about this hypothesis, but at least it explains why these discordant redshift objects are found in these arms, an explanation which would be rather difficult otherwise. The subject of companion galaxies has already been an important topic in this book. We have seen evidence for the origin of highredshift quasars associated with companions in nearby groups of galaxies. Now we have seen evidence for companion galaxies themselves to be peculiar and to have high nonvelocity redshifts with respect to their main galaxy. The subject of companion galaxies will continue to be an important subject in this book because, as we shall see, the structure of the universe we live in is typically that of groups of galaxies here and there, each dominated by one or two large galaxies with the rest of the galaxies forming a group of smaller companion galaxies around them. It is these smaller, compact objects and companion galaxies in groups which again and again furnish the most abnormal and most excessively redshifted objects that we encounter. B. The Active Central Galaxy, NGC 4151 In most groups, the central galaxy is not particularly active at the moment. For example, the central galaxy in our own Local Group is M31, a prototypical Sb spiral. (An Sb has a moderate central bulge of stars and not too conspicuous spiral arms.) It contains the majority of mass in the group and is apparently undergoing fairly smooth and regular rotation of its gas, dust, and stars around its center. M81, in the third most distant, major group is an Sb spiral almost identical to M31. Such Sb spirals are typically seen throughout space with their retinue of smaller companions grouped around them. But NGC 4151 is an Sb spiral with a very active nucleus. How active, you may ask! The brilliant point of light in its nucleus is variable, is a source of radio emission and X-rays, and shows broad, high-excitation emission lines. Altogether, the nucleus acts in many ways like a quasar. The person who has studied this nucleus most intensively, and with whom I am well acquainted, has attempted to find clues to the energy sources and energy mechanisms within quasars and active galaxies from largescale programs of observing and analysis of this nucleus. But I was always interested in the outer regions of NGC 4151. It is surrounded by a retinue of smaller galaxies. The trouble is that all of these smaller galaxies are of considerably higher redshift. The usual response would be to say that NGC 4151 is accidentally centered on a more distant group of galaxies. Unfortunately for this hypothesis, these smaller galaxies have generally different redshifts between themselves, ruling out their membership in the same group by conventional criteria. As usual, the first step was to get the best possible plate of the field. On a very dark night at Palomar, with exquisite seeing, I obtained the three-hour exposure shown in Figure 6-5. It shows the arm on the north going out and around almost to NGC 4156, the high surface brightness galaxy to the northeast. The other arm of NGC 4151 goes around and out to the south, where it appears to join a smaller galaxy just at the southwest edge of the picture. Since both of these 90 Galaxies with Excess Redshifts Figure 6-5. Deep exposure o/NGC 4151 by Arp until 200-inch Palomar reflector. smaller galaxies are of considerably higher redshift than NGC 4151 itself, I made a special effort to follow this up. Eventually, I obtained a number of plates and, with Jean Lorre's help, informationadded them to give the high-contrast picture shown in Figure 6-6. I think this picture shows convincingly that a high-redshift galaxy is attached to the end of each NGC 4151 arm. This result seemed so startling that I simply stated it and did not especially emphasize it with respect to the other evidence for the existence of discordant redshifts. But now that further evidence, as discussed earlier in this chapter, shows that objects of excess redshift occur in or along spiral arms, it can be put forward as strong corroborative proof of the phenomenon. It should also be noticed that the values of the redshifts involved are very similar to those in the Stephan's Quintet system discussed later, in Section E of this chapter. I have to add a note about B. A. Vorontsov-Velyaminov, a Russian astronomer, who in 1957 searched the Palomar Sky Survey paper prints for peculiar galaxies. One of his favorite kinds was companions on the ends of spiral arms. I learned the English word "gemmation" from him—a botanical term meaning budding or outgrowth. He was always searching for a spiral galaxy with companions on the ends of both arms to demonstrate his theory of galaxy formation. I should dedicate Figure 6-6 to him. There is a lot more of interest in the NGC 4151 region. It is a very active region in that all sorts of odd objects are found there. The knotty spiral due east of NGC 4151 has H II regions of much larger apparent diameter than those in NGC 4151, even though it is Galaxies with Excess Redshifts 91