Cubic spline fits to the DGRF/IGRF series indicate agreement with other analyses showing the 1969-1970 magnetic jerk in the h ??12 and g ??02 secular change coefficients, and agreement that the h ??11 term showed no sharp change. The variation of the g ??01 term is out of phase with other analyses indicating a likely error in its representation in the 1965-1975 interval. We recommend that future derivations of the 'definitive' geomagnetic reference models take into consideration the times of impulses or jerks so as to not be bound to a standard 5 year interval, and otherwise to make more considered analyses before adopting sets of coefficients. ?? 1987.