The analytical estimates for the ferric iron to total iron ratio of mid-ocean ridge basalts have been creeping up over time, from 0.08 in 1986 to 0.16 today. The implications for the oxidation state of the mantle and the recycling of oxidized material are obvious, but there is a different set of implications that have not been worked out. Our models that use the composition and thickness of oceanic crust to constrain the temperature and melting behavior of the oceanic mantle are very strongly dependent on Fe/Mg partitioning between melt and olivine, in which only ferrous iron counts. Previously these models seemed to work and offer a consistent solution wherein the mantle temperature that yields the right composition of primary basalt also yields the right thickness of oceanic crust, and fractionation of that primary basalt yields a complementary relationship between the cumulate lower crust and the volcanic upper crust. But increasing the ferric iron fraction breaks that consistency. It lowers the temperature and MgO content of the estimated primary magma, leaving very little room to form cumulates by fractionation and also requiring colder mantle that does not provide enough melt to explain the thickness of oceanic crust. I will explore these issues quantitatively using the PRIMELT3 and MELTS models and we will seek for a new coherent understanding, which will require changing either our analytical, numerical, or conceptual framework for interpreting the oceanic crust.