Early Hi work on external galaxies was done with single-dish telescopes, often prone to side lobe effects (cf. discussion after Salpeter 1978). Progress was slow, and hampered by low spatial resolution — a small ratio of Hi radius to beam size (Bosma 1978, his Chapter 3.4) — of the observations. With the Dwingeloo telescope, work on M31 was followed by major axis measurements of M33 and M101 (Volders 1959). The Parkes telescope was used to image the LMC (McGee and Milton 1966), the SMC (Hindman 1967), NGC 300 (Shobbrook and Robinson 1967), and M83 (Lewis 1968). In the northern hemisphere, data were reported for several galaxies by Roberts (1966, for M31), Roberts (1972) and Davies (1974). In an Appendix, Freeman (1970) discusses rotation data for the LMC, SMC, NGC 300 and M33, and, for the latter two galaxies, found that the turnover velocity indicated by the low-resolution Hi data is larger than the one indicated by fitting an exponential disk to the optical surface photometry, suggesting the presence of matter in the outer parts with a different distribution than that in the inner parts.
The quest for higher spatial resolution was first achieved routinely with the Caltech interferometer, (e.g., for M101, Rogstad and Shostak 1971; Rogstad 1971), and results for five Scd galaxies were summarized in Rogstad and Shostak 1972, who found high M / L material in the outer parts of these galaxies. Work began on M31 with the Cambridge half-mile telescope with a first report (Emerson and Baldwin 1973) showing “normal” M / L values. Their data were in disagreement with data by Roberts on the same galaxy, as debated in a meeting in Besançon (Roberts 1975; Baldwin 1975). This problem was settled by new data reported by Newton and Emerson (1977), who by and large confirmed the data of Roberts and Whitehurst (1975). Meanwhile, Hi work with the Westerbork Synthesis Radio Telescope (WSRT) had started, with as first target M81. The resulting rotation data, extending beyond the optical image, were discussed in Roberts and Rots (1973), with curves for M31 (300 ft data), M81 (WSRT and 300 ft data), M101 (from Rogstad's Caltech data) and the Milky Way. These data, in particular for M31, also suggested that there could be more than meets the eye in the outer parts of galaxies, i.e., material with high M / L ratio (cf. Fig. 1, left panel).
 |
Figure 1. Left panel: rotation curves of M31, as determined by Babcock (1939, purple points), van de Hulst et al. (1957, orange points), Rubin and Ford (1970, black points), and Roberts and Whitehurst (1975, red points). The blue line indicates the expected maximum disk rotation curve based on an exponential disk with the scale length given in Freeman (1970) based on the study of de Vaucouleurs (1958). The arrow indicates the optical radius. Right panel: modern picture of the outskirts of M31, as determined from the PAndAS survey (reproduced with permission from Ferguson and Mackey 2016), where the outline of the optical image is in white. Note the change in scale. |
A second line of argument for material with high M / L ratio in the outer parts of galaxies came from theory. Early numerical experiments, e.g., by Hohl (1971), showed that simulating a galaxy disk with N-body particles and letting it evolve generated the formation of a bar-like structure with relatively high velocity dispersion. Since strong bars are present only in about 30% of the disk galaxies, bar instability in disk galaxies Ostriker and Peebles (1973) proposed to stabilize the disk by immersing it in a halo of “dark matter”, in such a way that the gravitational forces of the halo material acting on the disk could prohibit the bar to form. This requires that out to the radius of the “optical” disk, the mass of the halo is 1 − 2.5 times the disk mass. Once hypothesized, it followed that the masses of galaxies exterior to this radius could be extremely large. This was subsequently investigated by Einasto et al. (1974), as well as Ostriker et al. (1974). While the former considered five galaxies, amongst which IC 342, and data on binary galaxies, the latter put forward at least half a dozen probes (the rotation curve data discussed above, the Local Group timing argument already considered in Kahn and Woltjer (1959), binary galaxy samples, etc.). Both papers concluded that there must be additional, dark, matter beyond the optical radius of a galaxy, since the mass increases almost linearly with radius, and the light converges asymptotically.
Athanassoula (2002, 2003), using much-improved N-body simulations with live haloes (i.e., haloes responding to gravitational forces), confirmed the above results for the initial phases of the evolution. She found, however, that at later times, when the bar evolves and increases in strength, the halo material at resonance with the bar will actually help the bar grow and become stronger. Hence the theoretical picture in Ostriker and Peebles (1973) is now superseded. Likewise, Athanassoula (2008) showed that the Efstathiou-Lake-Negroponte global stability criterion (Efstathiou et al. 1982), popular in semi-analytic galaxy formation models of galaxies, is not really valid.
Although some cosmologists developed the theory of galaxy formation, and came up with a two-stage model (e.g., White and Rees 1978), the debate about the validity of the data, and even the notion of dark matter itself, continued for several years, witness papers disputing the idea (e.g., Burbidge 1975; Materne and Tammann 1976). New data on this topic came from efforts improving the statistics of binary galaxies and clusters of galaxies, but the most convincing evidence came from fresh data on the rotation curves of spiral galaxies, discussed in Sect. 4.