We develop, implement and characterise an enhanced data reduction approach which delivers precise, accurate, radial velocities from moderate resolution spectroscopy with the fibre-fed VLT/FLAMES+GIRAFFE facility. This facility, with appropriate care,
delivers radial velocities adequate to resolve the intrinsic velocity dispersions of the very faint dSph dwarf galaxies. Importantly, repeated measurements let us reliably calibrate our individual velocity errors ($0.2 leq delta_Vleq 5$ km s$^{-1}$) and directly detect stars with variable radial velocities. We show, by application to the Bootes-1 dwarf spheroidal, that the intrinsic velocity dispersion of this system is significantly below 6.5,km/s reported by previous studies. Our data favor a two-population model of Bootes-1, consisting of a majority `cold stellar component, with velocity dispersion $2.4^{+0.9}_{-0.5}$,km/s, and a minority `hot stellar component, with velocity dispersion $sim 9$,km/s, although we can not completely rule out a single component distribution with velocity dispersion $4.6^{0.8}_{-0.6}$,km/s. We speculate this complex velocity distribution actually reflects the distribution of velocity anisotropy in Bootes-1, which is a measure of its formation processes.
Our knowledge of the Universe remains discovery-led: in the absence of adequate physics-based theory, interpretation of new results requires a scientific methodology. Commonly, scientific progress in astrophysics is motivated by the empirical success
of the Copernican Principle, that the simplest and most objective analysis of observation leads to progress. A complementary approach tests the prediction of models against observation. In practise, astrophysics has few real theories, and has little control over what we can observe. Compromise is unavoidable. Advances in understanding complex non-linear situations, such as galaxy formation, require that models attempt to isolate key physical properties, rather than trying to reproduce complexity. A specific example is discussed, where substantial progress in fundamental physics could be made with an ambitious approach to modelling: simulating the spectrum of perturbations on small scales.
There has been a vast recent improvement in photometric and kinematic data for star clusters, Ultra Compact dwarfs, galactic nuclei, and local dSph galaxies, with Subaru contributing substantially to the photometric studies in particular. These data
show that there is a bimodal distribution in half-light radii, with stable star clusters always being smaller than 35pc, while stable galaxies are always larger than 120pc. We extend the previously known observational relationships and interpret them in terms of a more fundamental pair of intrinsic properties of dark matter itself: dark matter forms cored mass distributions, with a core scale length of greater than about 100pc, and always has a maximum central mass density with a narrow range. The dark matter in dSph galaxies appears to be clustered such that there is a mean volume mass density within the stellar distribution which has the very low value of about 0.1$M_{odot}$ pc$^{-3}$. None of the dSphs displays kinematics which require the presence of an inner cusp, while in two dSphs there is evidence that the density profile is shallow (cored) in the inner regions. The maximum central dark matter density derived is model dependent, but is likely to have a mean value (averaged over a volume of radius 10pc) of about 0.1$M_{odot}$ pc$^{-3}$, which is 5GeV/c$^2$cm$^{-3}$). Galaxies are embedded in dark matter halos with these properties; smaller systems containing dark matter are not observed.
