The weakening of the overall field is largely due to patches of reversed magnetic field that have appeared in the Southern Hemisphere. Gubbins said his computer models indicated that these patches did not exist before 1840 and that the magnetic field was probably steady before the recent decline.

John A. Tarduno, a professor of geophysics at the University of Rochester, disagreed with that conclusion. “Maybe that’s only part of the story,” he said. ‘‘Maybe there’s a much longer trend, and these flux patches are part of this larger trend.’’

Tarduno said Gubbins’s data showed that the decline sped up in the mid-1800s, but ‘‘that’s as far as I would go’’. Other data suggest that the magnetic field was even stronger 2,500 years ago and has declined 40 percent since then.

The magnetic field is the only direct clue that scientists have for understanding the currents of molten iron that flow in the earth’s outer core and generate the field. Its collapse would let greater amounts of radiation from space reach the surface. But the fate of the field is unknown. Gubbins’s analysis could indicate that the recent drop is a temporary blip, or it could suggest that the decay is accelerating.

The earth’s magnetic field sporadically flip-flops, about once every 300,000 years, with the North Pole becoming the South Pole and vice versa. Some scientists think the recent decay is the start of the next reversal. The last switch occurred 780,000 years ago.

‘‘I think it’s going to go all the way,’’ Gubbins said. But a definitive answer is centuries away. ‘‘There’s no point in putting money on it,’’ he said, ‘‘because we’re not going to find out.’’