The highly siderophile elements (HSE's: Ru, Rh, Pd, Re, Os, Ir, Pt and Au) are those elements with distribution coefficients between Fe-rich metal and silicate phases which exceed 10(4). The large magnitude of these distribution coefficients makes them exceedingly difficult to measure experimentally. We describe a new experimental campaign aimed at obtaining reliable values Of D-M(mets/sil melt) for selected HSE's indirectly, by measuring the solubilities of the pure metals (or simple HSE alloys) in M haplobasaltic melts as a function of oxygen fugacity. Preliminary results for Pd, Au, Ir and Re indicate that the HSE's may dissolve in silicate melts in unusually low valence states, e.g., 2 + for Ir and 1 + for the others. These unusual valence states may be important in understanding the geochemical properties of the HSE's. Inferred values of D-M(met)/(sil melt) from the solubility data at 1400 degrees C and IW - 1 are similar to 10(7) for Pd and Au, and 10(9)- 10(12) for Ir. Metal/silicate partition coefficients are thus confirmed to be very large, and are also different for the different HSE's. A review of the abundance of the HSE's in the Earth's upper mantle shows that they are all present at similar to 0.8% of chondritic, i.e. they have the same relative abundance, and the ratios of their concentrations are chondritic (e.g., Re/Os). Both the low degree of depletion (compared to the high values of D-M(met/sil met)) and the chondritic relative abundances support the idea that the M mantle's HSE's were added in a ''late veneer'' after the cessation of core formation. Sulfur is even more depleted in the mantle relative to CI chondrites than the HSE's: this implies a late veneer which was depleted in volatile elements, and which was added to a mantle stripped of S. Since considerable S dissolves in silicate melt, this further implies that core formation in the Earth either occurred under P-T conditions below the silicate solidus, or, if the process occurred over a range of temperatures in a cooling Earth, then the process continued down to conditions below the silicate solidus. The chondritic relative abundances of the HSE's in the upper mantle argue for a chemically unstratified primitive mantle, unless the late veneer was mixed only into the upper mantle.