JSTOR

1. REDSHIFTS OF GALAXIES AND QSOs

More than 70 years ago Hubble (1929) clearly showed that there was a correlation between redshift and distance for spiral nebulae. Using Lemaître’s solutions to Einstein’s equations (Friedmann’s earlier work of 1922 was not known in the west at the time), it was immediately deduced that the universe is expanding. This was probably the most important astronomical discovery of the 20th century.

From it the very strong belief developed that whenever a redshift is measured it must be cosmological in origin. As fainter and fainter galaxies were observed, Hubble, Humason, Mayall, and Sandage, and later others, showed that a smooth redshift versus apparent magnitude diagram emerged. However, a departure from the linear relation for might be able to give us cosmological information (cf. Sandage 1961). The very existence of Hubble’s law for galaxies strongly supports the view that redshifts of galaxies are largely cosmological in origin, although the quantization effects for small (Tifft 1976; Guthrie & Napier 1996), ignored by observers and theorists alike, suggest we do not understand these redshifts completely.

When the first few quasi‐stellar objects (QSOs) were discovered in the early 1960s, it became clear that they did not follow the simple Hubble law for galaxies. The apparent magnitude redshift plot for QSOs is practically a scatter diagram. Of course, this does not mean that the redshifts are not cosmological in origin, but it means that, if they are, there must be a very large spread in absolute magnitude at every redshift. Rapid variability of radio and optical fluxes of QSOs was also discovered, indicating that the sources are very small, and leading to a basic problem in understanding the physics of the radiation process (Hoyle, Burbidge, & Sargent 1966). This could be gotten around either by supposing that QSOs are much closer than their redshifts indicate or by appealing to highly relativistic motions (Woltjer 1966; Rees 1967). This latter idea became the preferred solution, in part because people believed in cosmological redshifts, but certainly not because they could easily make viable models involving “superluminal” motions. Hoyle & Burbidge (1966) showed that as good a case could be made for supposing that the brighter QSOs were ejected from comparatively nearby galaxies, rather than lying at the distance indicated by their (cosmological) redshifts. If they are comparatively local, then the bulk of their redshifts must be intrinsic (noncosmological), and they cannot be used to probe the distant universe.

2. STATISTICAL INVESTIGATIONS

An observational test of the noncosmological redshift hypothesis requires a demonstration that two extragalactic objects (one a QSO) with very different redshifts are physically associated. Since we have normally used the redshifts to measure distance, we can only test the cosmological redshift hypothesis by using statistical methods to prove a physical association or by finding luminous connections between objects with different redshifts. In the last 30 years, a large body of observational evidence has been published showing that many brighter QSOs may have a component of noncosmological redshifts. The following statistical investigations use well‐defined samples of QSOs and compare their positions with bright low‐z galaxies:

More recent studies (Norman & Impey 1999; Norman & Williams 2000) have confirmed earlier work that shows statistical associations between bright QSOs and faint galaxies (which they call background QSOs) and foreground galaxies, showing they tacitly assume QSO redshifts are cosmological. Many workers interpret the results as due to weak gravitational lensing by dark matter, sometimes requiring selection biases, dark matter overdensities, or dust. But none of it works well, and of course it doesn’t work for the brightest and nearest galaxies.

3. ASSOCIATIONS OF INDIVIDUAL QSOs WITH BRIGHT GALAXIES

QSOs are very rare objects when compared with galaxies, and we do not expect to see very many accidental configurations even if all ∼10,000 known bright galaxies were surveyed. When I compiled all of the close pairs of a bright nearby galaxy ( ) with a high‐z QSO, there were 47 pairs, 44 of them with separations ≤3 . NGC 470 and NGC 622 have two associated QSOs, and NGC 1073 and NGC 3842 have three. Since we have catalogs of all bright galaxies and good estimates of the QSO surface density as a function of magnitude (Goldschmidt et al. 1992; Boyle et al. 1990), the probability that any close pair is a result of an accidental configuration can be calculated in a straightforward way. If QSOs are distributed randomly on the sky, the number of pairs we expect to find by chance with separation of θ arcminutes is , where Γ is the surface density of QSOs per square degree and N is the number of cases investigated. Of the 47 pairs, a few were found serendipitously and some in the early work of Burbidge et al. (1971). Apart from this, only Arp looked carefully at a fraction of the bright galaxies, and he only examined ≲200 before his observing program was abruptly stopped (Arp 1987). A comparison between the number of pairs expected by accident and those actually found is shown in Table 1. We have conservatively set . In addition, luminous connections have been found in a few cases (e.g., NGC 4319+Mrk 205; Sulentic & Arp 1987) showing a physical association between active galaxies and QSOs with very different redshifts.

TABLE 1 Comparison of Expected and Found Number of Pairs

Open New Window

Is there any way in which the close proximity of high‐z QSOs with low‐z galaxies could be understood while still believing that they both have cosmological redshifts? The only possible mechanism suggested so far is gravitational lensing, invoked by Bartelmann & Schneider (1994) for faint clusters and higher redshift QSOs. They suppose that very large amounts of dark matter are present in all of the galaxy clusters which show strong positional correlations with much higher redshift QSOs. But this argument won’t work for the correlations found with the brighter galaxies and QSOs. For very close pairs involving comparatively nearby galaxies, Canizares (1981) proposed that amplification of the QSO luminosity would be possible by microlensing if the halos of the nearby galaxies contained a large number of dark condensed objects. But this would require a very high surface density of faint QSOs (Ostriker 1989; Schneider 1994), far in excess of the measured value.

4. THE CONNECTION BETWEEN ACTIVE GALAXIES AND X‐RAY–EMITTING QSOs

Using the ROSAT survey of compact X‐ray sources, Radecke (1997) showed these sources tend to cluster about active galaxies. The significance of the association is as high as for sources 10 –40 from a galaxy. Arp (1997) also showed that many of these sources could be identified with blue stellar objects which turn out to be high‐z QSOs (Burbidge 1995, 1997, 1999; Chu et al. 1998; Arp et al. 2001). More than 40 QSOs around 15 galaxies have now been found in this way. In many cases the QSOs are aligned along the minor axis of the galaxy, suggesting that they may be ejected from the galaxy.

5. PEAKS AND PERIODICITIES IN THE REDSHIFT DISTRIBUTION OF QSOs

One final kind of evidence comes from the peaks and periodicities in the distribution of the observed redshifts. So far, I have shown from statistical and morphological evidence that part of the QSOs redshift cannot be due to cosmological expansion. We call this the intrinsic component . Then, the observed redshift is given by where and are the cosmological and Doppler components, respectively.

There is no evidence for peaks and periodicities in the velocities of normal galaxies, whose random motions indicate . Thus, if peaks are found, they must be largely due to . In other words , meaning is very small (i.e., these QSOs have small cosmological redshifts). Peaks in the redshift distribution for QSOs were found at and 0.061 (Burbidge 1968). Several more were found at , 0.60, 0.96 and 1.41, and Karlsson (1971) showed that these fit a periodic sequence with . Burbidge (1978) gave a histogram for ∼600 QSOs clearly showing these peaks and demonstrated that these were not the result of spectroscopic selection or the shifting of key emission lines. Samples which were used to disprove the Karlsson effect did not fulfill the criteria originally specified (spectra with broad emission lines and nonthermal continua).

Later Karlsson (1990) found the same periodicity in a sample of QSOs statistically associated with bright galaxies. Karlsson’s formula predicts that if higher redshift QSOs are present, peaks beyond 1.96 should appear at , 3.44, 4.47, 5.71, etc. Two samples chosen to avoid any optical selection effects are (1) X‐ray–emitting QSOs associated with active galaxies and (2) all the very close QSO pairs or multiples with separations ≤10 . A histogram of these samples is shown in Figure 8a of Burbidge & Napier (2001), where the peaks predicted by Karlsson's formula beyond 1.96 can be seen. For the X‐ray–emitting QSOs, we know the redshift of each parent galaxy. If we suppose that the intrinsic redshift component for each QSO is exactly the value at the nearest peak, we can calculate . We find that km s⁻¹, suggesting that the QSOs may be ejected from the galaxies at speeds of ∼0.03–0.1c. For both samples most of the QSOs have been found since the original periodicity was identified over 20 years ago.

6. CONCLUSIONS

The observational evidence described here shows that QSOs exist which have large redshift components not associated with the expanding universe. However, there is also evidence that some QSOs have redshifts which are largely cosmological (e.g., those residing in galaxies where ). For example, evidence comes from J. Miller’s spectroscopic work and also from imaging studies made with the Hubble Space Telescope.3 Damped Lyα absorption ( ) may also be a manifestation of a high‐z parent galaxy, so that (cf. Burbidge 1996). Clearly, the intrinsic redshift component can range from very small to very large values.

But, whatever is said or done, the evidence described here cannot be avoided. It tells us that noncosmological redshifts exist and that they show periodic effects. Hoyle & Burbidge (1966) attempted a first explanation. Since much of the observational evidence is in the literature, we may ask why it is being ignored. Here the problem appears to be sociological, not scientific. Early on, the community showed by its response to the work of Arp and others that it was not prepared to treat the evidence on its merits. This is even more true today. Many observational programs concerning the early universe are based on the assumption that redshifts always measure distance, so we can use QSOs as cosmological probes. It is not surprising that the few astronomers who believe there is a case for noncosmological redshifts cannot get research support or telescope time or that young people are afraid to touch the subject. As in other fields, the peer review process is overshadowed if not completely broken by the need to conform.

Progress has been held up before because authority would not accept new ideas. Wegener’s 1910 proposal that continental drift is taking place was buried by the geological establishment for more than 40 years until Blackett, Runcorn, and others found evidence for a wandering magnetic pole. Russell and Eddington could not face up to Cecilia Payne’s demonstration in 1925 that hydrogen is the most abundant element, and for 5 or 6 years they resisted the inevitable, and the community followed them. In 1938 Eddington would not accept Chandrasekhar’s result based on the “new” physics, although all quantum physicists knew it was correct. Now we have an even harder problem—anomalous redshifts which not only are highly destructive to many ongoing cosmological investigations but for which we have no good theory. Thus, both theorists and observers find it easier to censor, ignore, or bury the observational evidence. The end is not yet in sight.