Magnetic signal before Loma Prieta
A Stanford magnetometer at Corralitos recorded unusual very-low-frequency magnetic noise in the weeks and hours before the 17 October 1989 Loma Prieta earthquake, raising hopes of earthquake prediction.
1989 CE · 37.04, -121.88 · status: disputed
The account
The result, published by Antony Fraser-Smith and colleagues in 1990, is often cited alongside reports of 'earthquake lights' and animal behaviour.
The sceptical view
Later studies suggested the signal may have been an instrument problem; similar signals were not seen before other quakes.
The researched reading
A famous earthquake precursor that looks, on full reanalysis, like a faulty instrument. Whatever it was, it is a question of geophysics and engineering, not of the unexplained.
| Explanation | Weight | Why |
|---|---|---|
| Something in the surroundings | 50% | A 2009 USGS reanalysis of all 21 months of Corralitos data found baseline jumps and timing problems throughout, concluding the "precursor" was an artefact of a malfunctioning sensor system rather than a signal from the fault. |
| A known natural phenomenon | 25% | If the signal was real, it may have been ordinary geomagnetic activity (Campbell, 2009), since October 1989 had strong solar storms; or, much less likely, a genuine electrokinetic precursor from stressed rock. |
| Undetermined | 10% | The share a generic explanation had beyond what has been shown for this case: it fits many cases like this one, but has not been demonstrated for this one. |
| Perception, memory or mind | 6% | Researchers searching after a disaster for a precursor can over-read noise; this is a known bias in prediction research. |
| A cultural story | 5% | The result was widely cited as proof that earthquakes can be predicted, which kept it alive after it was challenged. |
| Genuinely anomalous | 3% | Nothing in this case requires anything beyond physics; the question is between sensor error and an unusual but natural signal. |
| A hoax or fabrication | 1% | No one has suggested the data were faked. |
Official position: The original 1990 paper reported a precursor; a 2009 USGS reanalysis attributed it to sensor malfunction, and the original authors disputed this.