In my view, Neta Bahcall's obituary of Vera Rubin oversimplifies the dark matter problem (Nature 542, 32; 2017). Many diverse observations have contributed to the current picture, and the unknown nature of dark matter shows that this is still far from complete.
Rubin investigated rotational velocities in spiral galaxies using ionized gas regions that she was able to observe at optical wavelengths. In the 1970s, such regions could not be detected beyond a galaxy's optical image. The observed rotation curves therefore did not extend far enough to demonstrate the presence of dark matter convincingly, even though the flatness of their outer parts might convey that impression.
At that time, I and several other astronomers used the 21-centimetre radio wavelength of neutral hydrogen to determine rotation curves that often went well beyond the optical image, thereby probing the dark matter regime more effectively.
Such observations from several galaxies, coupled with optical surface photometry, permitted the calculation of local mass-to-light ratios in the outer parts of spirals (A. Bosma PhD thesis 1978; see http://go.nature.com/2mkodu5). This clinched the evidence for dark matter and still constitutes the primary source of rotation-curve data for exploring its mysteries.