The long-term variation of the interplanetary electric field is inferred back to 1926 from a correlation analysis with the magnetograms recorded at Godhavn and Thule, two polar cap geomagnetic observatories. The method is reliable because of the large dependence of the magnetic perturbation on the cross-polar cap electric field, i.e., the penetration and mapping of the interplanetary electric field into the magnetosphere-ionosphere system. This dependence is isolated by minimizing Sq and the Svalgaard-Mansurov effect. Both appear when an observatory moves closer to the polar cap boundary and are found to be a minimum in a direction almost perpendicular to the magnetic north. Strictly speaking, no secular trend in the solar wind-magnetosphere large-scale coupling is indicated for the past 77 years. This suggests that there is no secular trend in the interplanetary electric field and by inference in the Sun's open magnetic flux and in the solar wind speed. The method is independent of the aa geomagnetic index and the sunspot cycle characteristics.