Long-lived isomers in 212 Bi have been studied following 238 U projectile fragmentation at 670 MeV per nucleon. The fragmentation products were injected as highly charged ions into the GSI storage ring, giving access to masses and half-lives. While the excitation energy of the first isomer of 212 Bi was confirmed, the second isomer was observed at 1478(30) keV, in contrast to the previously accepted value of >1910 keV. It was also found to have an extended Lorentz-corrected in-ring halflife >30 min, compared to 7.0(3) min for the neutral atom. Both the energy and half-life differences can be understood as being due a substantial, though previously unrecognised, internal decay branch for neutral atoms. Earlier shell-model calculations are now found to give good agreement with the isomer excitation energy. Furthermore, these and new calculations predict the existence of states at slightly higher energy that could facilitate isomer de-excitation studies. Isomers are long-lived excited states of atomic nuclei [1]. Their inhibited decays arise from nuclear shape changes and angular momentum (spin) selection rules, leading to a special role in nuclear physics and astrophysics research [2] and the the possibility of novel applications such as energy-storage devices, if appropriate conditions can be realised [3]. The understanding of the structure and properties of extreme isomers, combining a long half-life with high spin and/or excitation energy, is a key part of these investigations.The nuclide 212 Bi has an excited state with a unique combination of properties for a spherical nucleus: I ≥ 16, and t 1/2 = 7.0(3) min [4]. Nevertheless, it remains poorly characterised, with, for example, unmeasured excitation energy. This quantity is needed foremost to test the predictive power of nuclear shell-model calculations [5], which themselves are required for modeling elemental synthesis in explosive rapid-neutron-capture (r-process) astrophysical environments [6,7]. With only four nucleons (one proton and three neutrons) outside the doubly magic core of 208 Pb, shell-model calculations should be reliable for 212 Bi. Surprisingly, however, the estimate of E * >1910 from its β-decay rate [4] is substantially different from the calculated energy of 1496 keV for the best isomer candidate, with I π = 18 − [5]. Notwithstanding these contrary indications, if the differences could be properly understood, then the possibility of exploiting the isomer for energy-release studies could be addressed.In the quest for the manipulation of nuclear isomers with low-energy electromagnetic probes, isomer targets had seemed the most promising, for example 180m Ta, with t 1/2 > 7×10 15 yr [8], and 178m2 Hf, with t 1/2 = 31 yr [9]. However, the former requires >1 MeV photons [10], and initial claims for the latter [11] have been refuted [12,13]. Nevertheless, with new radioactive-beam developments, the half-life requirement is now less stringent. This has been demonstrated through the induced depopulation of 68m Cu, with t 1/2 = 3.8 min [14], explo...