We attempt to solve the magnetic structure of the gadolinium analogue of `spin-ice, using a mixture of experimental and theoretical assumptions. The eventual predictions are essentially consistent with both the Mossbauer and neutron measurements but
are unrelated to previous proposals. We find two possible distinct states, one of which is coplanar and the other is fully three-dimensional. We predict that close to the initial transition the preferred state is coplanar but that at the lowest temperature the ground-state becomes fully three-dimensional. Unfortunately the energetics are consequently complicated. There is a dominant nearest-neighbour Heisenberg interaction but then a compromise solution for lifting the final degeneracy resulting from a competition between longer-range Heisenberg interactions and direct dipolar interactions on similar energy scales.
One dimensional systems sometimes show pathologically slow decay of currents. This robustness can be traced to the fact that an integrable model is nearby in parameter space. In integrable models some part of the current can be conserved, explaining
this slow decay. Unfortunately, although this conservation law is formally anticipated, in practice it has been difficult to find in concrete cases, such as the Heisenberg model. We investigate this issue both analytically and numerically and find that the appropriate conservation law can be a non-analytic combination of the known local conservation laws and hence is invisible to elementary assumptions.
