Returning to the timing study, the line A superposed on Figure 2 marks the start of the increase in . Many of the ground magnetometers also show the onset of perturbations. The perturbations are gradual until the time of line B, and thus they are consistent with the loading phase of the substorm. The difference field, , at POLAR suggests that the flux pileup continued during this interval. We note there was a minimum around 0330-0337 UT in the data from FCHU and PBQ. (The Z component at FCHU also showed a general minimum, with twin (negative) peaks around 0330 and 0337 UT, (not shown).) However ESKI and GILL, adjacent stations of FCHU in the same meridional chain, observed little peak-like perturbation. This feature suggests that this minimum occurred only within a limited range of latitudes including FCHU and PBQ, possibly on a narrow auroral oval during the loading phase. We note the Z component of the magnetometer data showed a small positive perturbation at ESKI, a large negative perturbation at FCHU, and a small negative perturbation at GILL (not shown). This also suggests that the current was narrow, and was located between ESKI and GILL, and a little north of FCHU. We further notice that at the POLAR spacecraft there is a decrease in and around 0339 UT. This might be related with the above-mentioned small ground activity, but particle data obtained by POLAR suggests a different explanation, which will be discussed later.
Around 0408 UT (line B), FCHU and GILL recorded a sudden decrease in the X component. Lines C and D mark the following sudden decreases in the X component at several stations (around 0435 and 0500 UT), indicative of intensification of the westward electrojet. Whether they all correspond to substorm onsets is an important question, because they are not necessarily reflected in the magnetic field at POLAR. The difference field, , at POLAR reached a maximum around 0430 UT, and the interval 0312-0430 UT (78 min) is comparable in length to the interval 0301-0413 UT (72 min) when the IMF was largely southward (see Figure 1). It is therefore likely that the interval 0312-0430 UT corresponds to the substorm loading phase. (We note that we estimated above the time lag from INTERBALL-1 to the Earth to have been 3 min, while the observed lag appears to be 11-17 min. This difference may come in part from the response time of the magnetopause to a southward turning of the IMF.) However, the electrojet intensification at 0408 UT (line B) preceded the 0430 UT field maximum at POLAR. We interpret this intensification at 0408 UT as the first onset signature of multiple expansion onsets (lines B, C, and D); the interval 0408-0430 UT was in the initial unloading phase, but the pileup of the magnetic flux onto the polar magnetosphere continued, because the IMF was still southward. It is possible that the time variation of BT in the polar magnetosphere reflects the time variation of the energy input from the solar wind more directly than in the magnetotail proper where the effects of substorm onsets are also apparent. Another possible reason for the delay from the initial onset (line B) to the maximum at POLAR would be the propagation time delay of the signal of the onset from the source region (in the near-Earth tail) to the polar magnetosphere. However, as stated below, there is some difficulty with this explanation.