The Hum · what the data shows

What the Hum data shows

We ran the Hum reports through the same engine that runs the rest of this record: where they concentrate, and whether they line up with an ordinary physical cause, on the ground or in the sky. Nothing here decides what the Hum is. It says what the numbers do and do not support, and it says so plainly.

Where it concentrates

The heaviest clusters, by report. These follow where the survey has reach, mostly the English-speaking world, so read them as where it is most reported, not where it is necessarily worst.

We checked every report against the real records

For all 2,933 reports that carry a date and a place, we asked the actual public records whether anything ordinary was happening at that spot on that day: the USGS earthquake catalogue, NASA's fireball survey, and the GFZ Potsdam geomagnetic index. The coincidences:

That is under two percent of reports, all three causes put together. The other ninety-eight percent line up with nothing in any of the records. One honest caveat: a report's date is the day the person filled in the survey, and the Hum is something people hear for years, not on one night. So this is a conservative test, and the near-total absence of a coincidence is exactly what you would expect from a sound that is always there rather than one that arrives with an event.

The frequency itself is a clue

Where people gave a pitch, 1,099 of them did, the number is telling. The median is 60 Hz, most between 40 and 100 Hz. That matters two ways. It is almost all above 20 Hz, so this is a sound people genuinely hear, not true infrasound (only 71 of 1,099 sit below 20 Hz). And it is far above the 7.8 Hz Schumann resonance that the electromagnetic theories lean on: 99 percent of reports are above it. A hum at 40 to 100 Hz is the range of electrical mains and machinery, not of a global electromagnetic resonance.

The hotspots are seismically quiet

Looked at by place rather than by day, it holds. For each hotspot we counted the earthquakes M2.5 and larger within about forty kilometres, across 2010 to 2024. 7 of the 8 top hotspots (88%) sit in seismically quiet ground, three quakes or fewer in fifteen years. And the reports cluster at mid-latitude (median 43.7 degrees, only 7% above 55), the wrong shape for an auroral source. Neither the ground nor the sky is where these point.

Not nearby industry either

The most ordinary explanation left is a local machine: a wind farm, a factory, a compressor station. So we counted wind turbines and industrial sites within about eight kilometres of each hotspot, from OpenStreetMap, and set them against ordinary cities that are not hum hotspots. The hotspots did not stand out. The median hotspot had 57 such sites nearby; the control cities had 279, more, not fewer. If anything the Hum concentrates where there is less heavy industry, not more. Mapping coverage differs by country, so read this as suggestive rather than settled, but it does not point at industry.

The one hint: weather that carries sound

A temperature inversion, warm air sitting over cold, bends distant low-frequency sound back down to the ground, and it is the one ordinary cause with any support here. Using ERA5 reanalysis we measured how often each place has an inversion in winter, when they peak. The hotspots came out marginally higher, inverted 8 percent of winter hours against 7 percent in ordinary cities. It is a hint in the right direction, distant sound ducted to a listener, but it is far too small a gap to carry a phenomenon this widespread. Worth chasing with finer data, not an answer.

Even the tides

There is an idea that tidal stress in the crust could drive a hum, so we computed the tidal force for every report and set it against random days. No difference: 29 percent of reports fall on strong-tide days, against 27 percent expected by chance. We say plainly this is not yet a fair test, because the date we hold is when someone filled in the survey, not the night the Hum was worst. The instrument is built and waiting for timing fine enough to ask it properly.

What people report with it

From the survey, the symptoms that come up most often. They cluster tightly, which is its own kind of signal.

insomnia (1,875) · sense of vibration (1,872) · anxiety (1,439) · ear discomfort (999) · headache (741) · nausea (318) · annoyance (36) · insomnia;sense of vibration;anxiety (32)

They reach for the same words

Thousands of people, filling in a survey on their own, describe it with the same handful of words: an idling engine, a constant low hum, a vibration in the house at night. Nobody copied anybody. Independent people converging on the same description is the strongest sign there is that they are hearing a real, specific thing.

frequency · engine · sounds · night · house · humming · idling · constant · outside · diesel

What this is and is not

This is a set of cross-references, not a verdict, and the hotspots reflect where the survey reaches rather than a clean global sample. But taken together the pattern is consistent and worth saying plainly: the explanations people reach for first do not fit. Not earthquakes, not meteors, not geomagnetic storms or the aurora, not the Schumann resonance that the frequency rules out, and not simple nearness to industry. What is left is harder to see from a survey alone. Two physical ideas are still open and testable with finer data, temperature inversions that duct distant sound and tidal stress, and both need to know when the Hum is worst, not only where. The leading non-physical possibility, that much of it is generated inside the ear, is not something this data can reach. So the honest end is an open question, with several of the loudest answers taken off the table.