Significant interest has arisen, owing to the detection of slow magnetic relaxation in single-molecule magnets (SMMs), [1] where magnetic anisotropy arises from zero-field splitting of a large spin ground state. Slow relaxation in SMMs is rarely seen at temperatures above 1.8 K, and, even after two decades of study, the highest blocking temperatures (below which slow relaxation is observed) remain at around 4 K. [2] Recently, we observed slow relaxation at 17 K in a {Ni 10 } cage complex, [3] which led us to propose a quite different mechanism to account for slow relaxation in the absence of any significant energy barrier from either high anisotropic spins (as in SMMs) or intermolecular interactions.The {Ni 10 } clusters possess a supertetrahedral core, [4] incorporating Ni II ions on the vertices and edges of a tetrahedron (Figure 1), centred on a m 6 oxide, with trisalkoxide ligands [5] binding the {Ni 6 } faces. This topology leads to a magnetic structure consisting of dense, well-separated bands. There is strong antiferromagnetic coupling (2 j J j % 120 cm À1 ) between the pairs of edge-Ni II ions, which define an inner octahedron centred around the m 6 oxide, and very weak coupling between the four vertex Ni ions, which define the outer tetrahedron. The result is a lowestenergy "band" consisting of (2s + 1) 4 = 81 levels, separated by approximately 120 cm À1 from the first excited band. This band structure requires both the highly frustrated topology and the weak magnetic coupling between vertex sites. Magnetic relaxation requires that low-energy phonons are dissipated into the lattice. However, in the model proposed for {Ni 10 }, the existence of 81 levels within the lowest-energy band makes it much more probable that the phonons are reabsorbed by a neighboring molecule, and hence trapped, leading to slow relaxation at relatively high temperatures compared with SMMs. This mechanism was initially proposed by Anderson for relaxation in paramagnetic salts at low temperature, [6] but not at high temperatures, as in {Ni 10 }.To test this theory, we required new materials, perhaps with the same topology [7] but with different values for either the local spin s or the exchange J. Although such isomorphous high-nuclearity structures are very rare, a limited number of examples, incorporating nickel or cobalt, are known [8][9][10] and, therefore, this target seemed achievable. Herein, we show that the Co II -triol solvothermal reaction system is much more versatile than its Ni II equivalent, with minor changes in reaction conditions leading to structurally related species including a symmetrical {Co II 10 } supertetrahedron, an opened, unsymmetrical {Co II 10 } supertetrahedron, and a vertex-sharing {Co II 19 } bi-supertetrahedron. Crucially, the symmetrical {Co 10 } complex exhibits qualitatively similar magnetic behav-Figure 1. a) Structure of 1 in the crystal, with supertetrahedron highlighted; b) the {Co 10 O 13 } fused-heterocubane core (Co dark gray, O light gray, C skeleton only, H omitted for clarity).