The IAGA-endorsed Polar Cap indices for the northern and southern hemispheres, PCN and PCS, are compared for 1998-2021. Potential effects of the slightly-different, and changing, magnetic coordinates of the two magnetic stations employed, Thule (Qaanaaq) in Greenland and Vostok in Antarctica, are investigated. It is shown that the agreement in overall behaviour of the two indices is very close indeed but that PCS consistently correlates slightly better with solar wind parameters than PCN. Optimum lags for these correlations are 19 min. for 1-min. data and 37 min. for hourly averages. The correlations are significantly higher with the predicted magnetopause reconnection voltage, which is a linear predictor of PCN and PCS for all 1-hour data and for all but the largest 0.1% of 1-min. values. The indices show lower correlation and marked non-linearity (tending to saturation) at all levels with the estimated magnetopause reconnection electric field or the estimated power input into the magnetosphere. The PCN index is shown to correlate closely with the transpolar voltage measured by the northern-hemisphere SuperDARN radar network and both PCN and PCS clearly show the Russell-McPherron effect of dipole tilt and the Y-component of the interplanetary magnetic field. However the patterns in time-of-year and Universal Time (UT) are complicated by lobe reconnection during northward-IMF, the effect of which on the indices is shown to be predominantly a summer hemisphere phenomenon and gives UT dependence on the IMF Y-component that is predicted theoretically.