The atomic rates for nickel are less accurate, and less well debugged, than those for other elements. It is recommended that any model be run first with the nickel abundance set to zero. Substantial differences between models with zero and non-zero nickel, in terms of temperature, opacity, etc., should be treated with great caution.