White dwarfs are dense, cooling stellar embers consisting mostly of carbon and oxygen 1 , or oxygen and neon (with a few percent carbon) at higher initial stellar masses 2 . These stellar cores are enveloped by a shell of helium which in turn is usually surrounded by a layer of hydrogen, generally prohibiting direct observation of the interior composition. However, 1 arXiv:2003.00028v1 [astro-ph.SR] 28 Feb 2020 carbon is observed at the surface of a sizeable fraction of white dwarfs 3, 4 , sometimes with traces of oxygen, and it is thought to be dredged-up from the core by a deep helium convection zone 5, 6 . In these objects only traces of hydrogen are found 7, 8 as large masses of hydrogen are predicted to inhibit hydrogen/helium convective mixing within the envelope 9 . We report the identification of WD J055134.612+413531.09, an ultra-massive (1.14 M ) white dwarf with a unique hydrogen/carbon mixed atmosphere (C/H = 0.15 in number ratio). Our analysis of the envelope and interior indicates that the total hydrogen and helium mass fractions must be several orders of magnitude lower than predictions of single star evolution 10 : less than 10 −9.5 and 10 −7.0 , respectively. Due to the fast kinematics (129 ± 5 km s −1 relative to the local standard of rest), large mass, and peculiar envelope composition, we argue that WD J0551+4135 is consistent with formation from the merger of two white dwarfs in a tight binary system [11][12][13][14] .WD J0551+4135 was identified as a candidate high-mass white dwarf 15 from its location in the Gaia Hertzsprung-Russell diagram 16 (Fig. 1). The Gaia parallax places WD J0551+4135 at 46.45 ± 0.15 pc away from the Sun. Compared to most other white dwarfs of the same G BP − G RP colour, WD J0551+4135 is fainter in absolute magnitude, signifying a relatively small radius, and thus a large mass given the mass-radius relation of white dwarfs whose structure is dominated by electron degeneracy 17 .− log(1 − m r /M wd ) = 4 (where m r is the mass enclosed within a sphere of radius r), the models are almost indistinguishable. Therefore the degree of oxygen dredge-up is not found to depend upon the core-composition, and so the atmospheric non-detection of oxygen cannot be used to test the evolutionary history. 6