Observations made at Las Campanas Observatory and at the Anglo-Australian Observatory have been used to determine line-of-sight velocities with an average accuracy of 3 kms−1 for 135 member stars in the globular cluster 47 Tucanae. The velocities were derived from cross-correlation techniques applied to 30 A/mm spectra obtained with digital sky-subtracting detectors. The spectra themselves have been used to analyze the cyanogen anomalies on the red giant branch in this cluster (Norris et al., 1984). When combined with the velocities published by the CORAVEL group (Mayor et al., 1983), these observations yield velocities for 212 stars with projected distances from the cluster center ranging from 3 to 68 core radii. After radial binning and analysis these observations yield the following results:
(i) The inner parts of the cluster show appreciable differential rotation with a maximum projected rotation velocity of approximately 6 kms−1 in the region 6–18 core radii. However, at larger radii the rotation declines rapidly and is essentially zero for radii greater than 30 core radii. This result is illustrated in Figure 1. To within the errors of the determinations, the position angle of the maximum rotation and that of the major axis of the stellar density distribution coincide.
(ii) In contrast to M3 (Gunn and Griffin 1979), “thermal equilibrium” multimass models (c.f. Da Costa and Freeman 1976) can ONLY reproduce the observed velocity dispersion values by including a substantial amount of “dark matter”; i.e. unlike M3, there is “missing mass” in 47 Tuc. In order to retain a fit to the surface brightness profile of the cluster, this “dark mass” (which provides perhaps 30 to 40 percent of the total cluster mass) cannot have a distribution much different from that of the cluster giants if it is in the form of stars and “thermal equilibrium” is maintained. In this case the obvious candidates for the dark matter are the white dwarf remnants of the stars more massive than the current turnoff mass, though many more such remnants are required than the number expected from extrapolating the present mass function. The difference between M3 and 47 Tuc in this case then implies that the 47 Tuc initial mass function had many more massive stars than did that for M3. The work of Freeman (1977), who demonstrated large IMF variations in the 8 − 1.5 solar mass range in young Magellanic Cloud clusters, provides observational support for this interpretation.