...1,2,

Visiting Marie Curie Fellow at the Institute of Astronomy, Cambridge University.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... clusters

Part of the scatter also originates in the uncertainties of globular cluster mass computation, since it relies on model assumptions such as the isotropy of velocity distibution and a maxwellian distribution of velocities.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... plot

If we consider the Old Halo clusters with a metallicity larger than -1.8, the increase of the metallicity with decreasing mass is strengthened: the linear Pearson correlation coefficient is -0.64 corresponding to a probability of correlation of 99.93%. One can question about the validity of dividing the Old Halo group at a metallicity of -1.8. According to Zinn (1996) and Dinescu et al. (1999), the Old Halo group with $\rm [Fe/H]>-1.8$ is kinematically distinguishable from the Old Halo group with $\rm [Fe/H]<-1.8$ (their Metal-Poor component). However, more kinematical data are needed to clarify this tentative point.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.