Kimberlites are the surface manifestation of deeply-derived (>150 km) and rapidly ascended magmas. Fresh kimberlite rocks are exceptionally rare, as most of them are invariably modified by pervasive deuteric and/or post-magmatic fluids that overprint the original mineralogy. In this study, we examined fresh archetypal kimberlite from the Mark pipe (Lac de Gras, Canada), which is characterised by well-preserved olivine and groundmass minerals. The sequence of crystallisation of the parental melt and its major compositional features, including oxygen fugacity, were reconstructed using textural relationships between magmatic minerals, their zoning patterns and crystal/melt/fluid inclusions. Crystal and multiphase primary, pseudosecondary and secondary melt/fluid inclusions in olivine, Cr-diopside, spinel, perovskite, phlogopite/kinoshitalite, apatite and calcite preserve a record of different stages of kimberlite melt evolution. Melt/fluid inclusions are generally more depleted in silica and more enriched in alkalis (K, Na), alkali-earth (Ba, Sr) and halogens (Cl, F) relative to the whole-rock composition of the Mark kimberlite. These melt/fluid inclusion compositions, in combination with presence of elevated CaO (up to 1.73 wt%), in Mg-rich olivine rinds, crystallisation of groundmass kinoshitalite, carbonates (calcite, Sr-Ba-bearing) and alkali-enriched rims around apatite suggest that there was progressive enrichment in CO₂, alkalis and halogens in the evolving parental melt.

The Mark kimberlite groundmass is characterised by the following stages of in-situ crystallisation: (1) olivine rims around xenocrystic cores + Cr-spinel/TIMAC. (2) Mg-rich olivine rinds around olivine rims/cores + MUM-spinel (followed by pleonaste and Mg-magnetite) + monticellite (+ partial resorption of olivine, along with the formation of ferropericlase and CO₂ as a result of decarbonation reactions) + perovskite + apatite. (3) Olivine outmost rinds, which are coeval with phlogopite/kinoshitalite + apatite + sulphides + carbonate (calcite, Ba-Sr-Na-bearing varieties). In addition, oxygen fugacity of the Mark kimberlite was constrained by olivine-chromite, perovskite and monticellite oxygen barometry and showed that the parental melt became progressively more oxidised in response to fractional crystallisation. (4) Deuteric (i.e. late-stage magmatic) and/or post-magmatic (i.e. external fluids) alteration of magmatic minerals (e.g., olivine, monticellite, ferropericlase) and crystallisation of mesostasis serpentine, K-bearing chlorite and brucite (i.e. replacement of ferropericlase). The absence of any alkali (Na, K) and halogen (F, Cl) rich groundmass minerals in the Mark kimberlite may be attributed to these elements becoming concentrated in the late-stage melt where they potentially formed unstable, water-soluble carbonates (such as those observed in melt inclusions). Consequently, these minerals were most likely removed from the groundmass by deuteric and/or post-magmatic alteration.