In 1981, the Basaltic Volcanism Study Project (BVSP) listed five bulk mantle compositions for Venus that were derived from Solar System formation models and geophysical and cosmochemical constraints (BVSP, 1981). These models were informed by the limited information available for Venus at the time, however, approximately three years later the surface compositions obtained by the Venera 13, Venera 14 and Vega 2 Soviet landers were released. The landers' X-ray fluorescence analyses showed that Venus is comprised of alkaline and tholeiitic basalts, very similar to terrestrial basalts (Surkov et al. 1984, 1986; Treiman 2007). The mantle compositions from BVSP (1981) and Earth mantle compositions are regularly used when modelling Venus, however, it has not been tested if these mantle compositions can produce melts with compositions similar to the Venera and Vega lander data.

In this study, we calculated primary melt compositions using the five BVSP (1981) Venus mantle compositions and a terrestrial mantle composition (KLB-1, Davis et al. 2009). Primary melt compositions were calculated using Perple_X (Connolly 2005) and thermodynamic data sets from Holland, Green and Powell (2018) and Tomlinson and Holland (2021), and then compared to the Venera 14 and Vega 2 lander data. This comparison showed that none of the BVSP (1981) mantle compositions could produce primary melts with compositions similar to Venera 14 or Vega 2 lander data. The terrestrial mantle composition, KLB-1, also could not produce Venera 14 or Vega 2-like melts. Therefore, we propose a new bulk mantle composition for Venus that was determined through an iterative process whereby the terrestrial mantle composition, KLB-1, was modified until a new bulk mantle composition for Venus was found that is capable of producing Venera 14-like melts. However, no bulk mantle composition can produce primary melts that are close to the Vega 2 data.