The physics of the solid 3He magnetic problem is reviewed. The data base includes C(T, 0), P(T, 0), P(T, H), X, S(T, 0), S(T, H), and the location of the magnetic phase transition on the melting curve. This data base permits determination of only two moments of the magnetic Hamiltonian, Jx~z and jx2x.Contrary to conventional wisdom, the moment J3xx cannot be reliably determined from the data. As there is no model-independent theory for the analysis of magnetic data near Tc, the data base is limited to giving three numbers (two moments and To) and many qualitative properties. Three theories of the solid 3He magnet are reviewed: (a ) theories based on quadratic Hamiltonians, (b ) theories that employ some form of four-spin Hamiltonian, and (c ) the theory of vacancy-induced ferromagnetism. In agreement with Johnson and Cohen, we find quadratic Hamiltonians to be hard-pressed to provide an adequate description of the data. A number of four-spin Hamiltonians have been treated in a variety of approximations and are found to yield reasonable agreement with certain features of experiment. These theories are reviewed. The number of adjustable parameters in these models, three, is comparable to the number of independent pieces of data. For at least one of the four-spin theories the parameters that are required to describe the data are implausible. This implausibility is deduced from a comparison of the numbers with the microscopic theory of the exchange Hamiltonian. Although the microscopic theory is not able to give a quantitatively useful Hamiltonian, it is capable of giving the correct sign and relative magnitude of the exchange rates. Thus the microscopic Hamiltonian constitutes a constraint on phenomenological theories that must be taken seriously. Sokoloff and Widom have suggested that the solid 3He magnet undergoes a ferromagnetic transition due to the presence of ground-state vacancies. A calculation of the location of the effective band edge and a mean field treatment of vacancy-induced ferromagnetism is given. We conclude that the present data base is much too sparse to provide a guide to theory or to provide a serious test for the variety of existing theories.