Super- sleuthing the Hum, by Dr Chris Barnes, Bangor Scientific  and Educational Consultants, Gwynedd, Wales LL57 2TW.    Draft version July 2020.   


Some existing protocols for low frequency noise ( LFN) measurement are discussed and shown to be inadequate.  A new method of taking LFN measurements not only at the external façade but also centrally  in affected rooms and on the inner surfaces of their windows  and load bearing walls is described. The method allows quick identification of airborne and ground borne components. Acoustic signals appear to be generated by house windows and walls, some of which are at power system frequencies and sub-harmonics and are present even when the mains electricity is turned off. Similar  and simultaneous  frequencies are confirmed by electric potential measurements in the fabric of the walls and by magnetic field measurements, which based on recently discovered properties of architectural materials such as brick and concrete suggest that in our homes we are all effectively living within a vast loudspeaker.  The discovery is shown to provide a new and highly feasible explanation of the hitherto enigmatic LFN phenomenon referred to as the Hum.   Further an explanation of  the time evolution of the history of the Hum first appearing in the UK and not until some 20 years later in the USA is offered in terms of the effects of PME electrical earthing technology on electrical ground currents.  Ground current frequency spectra are shown to change dramatically when the Hum is present and to match those measured at electricity sub-stations and for power grid systems in general.  It seems likely that only when power systems are behaving unstably and with additional frequency components as to create ‘comb’ spectra that strong Hum is observed, at least by the two subjects in this present study.   Most likely instability at the source or sink of a power system could create the Hum.   Also, the possibility exists that present-day Hums are accentuated by the growth in renewable energy systems.   Some have suggested  that ground vibrations from other utilities such as gas and water may have a part to play in the Hum.  At least at the author’s premises  this does not seem to be the case, but some further work remains to be done.      




In this paper I will show that we all are in fact effectively living within a loudspeaker!  Is it time therefore to re-assess how we define and deal with low frequency noise?  I will approach the topic through a partial examination of the literature and some personal experimentation.   Moreover, I will explain how airborne sound, architectural acoustics, seismic vibration and electromagnetics and our nation’s choice of electrical earth system may have come together in a unique interface to give us the enigmatic low frequency noise phenomenon commonly referred to as the Hum.    The paper should be of interest not only to acoustics and environmental health professionals but also to amateur radio, electronics and computer hobbyists who can easily make their own measurements according to the procedures I will describe.     


Noise in general has emerged over the past few decades as a leading environmental nuisance in the WHO European Region, and the public complains about excessive noise more and more often.  Low frequency noise is defined as noise with most of its components at frequencies below 200 Hz.  Below 20 Hz the low frequency noise is strictly defined as infrasound.  It is, however, a fallacy to believe that humans cannot perceive infrasound they merely require a higher sound pressure level to do so [1].   Although the cochlear may have less involvement with infrasound, other sensory cells or structures in the inner ear, such as the outer hair cells, are more sensitive to infrasound than the inner hair cells and can be stimulated by low frequency sounds at levels below those that are heard. The concept that an infrasonic sound that cannot be heard or can have no influence on inner ear physiology is incorrect.


In fact, under some clinical conditions, such as Meniere’s disease, superior canal dehiscence, or even asymptomatic cases of endolymphatic hydrops, individuals may be hypersensitive to infrasound [2].


According to Leventhall [3],  low frequency noise, the frequency range from about 10Hz to 200Hz, needs to  be recognized as a special environmental noise problem, particularly to sensitive people in their homes. Conventional methods of assessing annoyance, typically based on A-weighted equivalent level, are inadequate for low frequency noise and lead to incorrect decisions by regulatory authorities. There have been many laboratory measurements considering    annoyance to subjects  by low frequency noise, each with different spectra and levels, making comparisons difficult, but the main conclusions are that annoyance of low frequencies increases rapidly with level. Additionally, the A-weighted level underestimates the effects of low frequency noises. There is a possibility of learned aversion to low frequency noise, leading to annoyance and stress which may receive unsympathetic treatment from regulatory authorities.   


Noise is an underestimated threat that can cause several short- and long-term health problems, such as for example sleep disturbance, cardiovascular effects, poorer work and school performance, hearing impairment, etc.      The vibrations of low frequency noise and infrasound can cause all the foregoing together with damage to lung tissue as well, so called vibro-acoustic   disease [4].


Most people typically describe low frequency noise as something like a low rumble, a buzzing, a pulsation, a vibration or a pressure  or a popping within one or both ears and any  combination  of these.     Only a small proportion of people are affected, but those who are can be severely distressed, can lose sleep and suffer other symptoms like depression. Sometimes the complaint can be traced to a single sound source, typically an industrial site, or an adjoining property. If the source can be found, then there is a chance the noise can be controlled.   


In most cases where a sound source can be traced, it is ‘airborne’, meaning that it travels through the air as a sound wave and enters the dwelling through windows, roof, etc.  Sound can also be ‘ground-borne’, meaning that it travels through the ground and is converted into sound in the dwelling, or is sensed as vibration.    One of the characteristics of low frequency sound is that it can travel relatively long distances without much attenuation (reduction in level). It is not uncommon that a source is traced to a site several kilometres away from the complainant’s property.


However, in some cases no single sound can be traced that could account for the disturbance. 

This can be distressing, especially if to say a single sufferer the sound could appear to be loud, even deafening whereas others cannot hear it. A partial explanation for some cases is as follows: disturbance by LFN is known to occur at levels only slightly higher than hearing threshold, which varies from one individual to the next. Also, loudness varies rapidly with level at low frequencies so a sound only slightly above one person’s threshold could appear loud to them yet inaudible to someone else.  So, for example if the investigator cannot even hear the noise in the first place, they  might not be very sympathetic to the cause.  Although there is some logic in this argument it does not explain anything like all cases.


As seen above, there is a strong case to treat LFN  separately from  other types of environmental noise. Unwanted sound in the environment is commonly referred to as environmental noise. The Environmental Health section of your local authority is responsible for ensuring that environmental noise does not cause a nuisance. There are established procedures for assessing environmental noise from industrial sites which are based on measurements of the so-called dB(A). The term dB is an abbreviation for decibels which are the units for measuring sound. The (A) means that the sound is filtered to mimic the varying sensitivity of the human ear at different frequencies. Whilst this procedure works well for most industrial noise, the ‘A’ filtering operation strongly attenuates (reduces) the low frequency content of  sounds.  Thus if the  dB(A) measurement scale were applied to a LFN it most likely will not show up any problem. Therefore, several countries have adopted separate guidelines for assessment of low frequency noise.




LFN criteria in some countries compared to ISO threshold | Download  Scientific Diagram


Figure 1:  Comparison of thresholds for  action against LFN in various countries.


For example, it can be seen that at 10Hz Poland is more stringent than Germany by almost 20 dB. 


Many problems of LFN remain unresolved.   Throughout the world since the 1970’s in Britain and the 1990’s in the USA unresolved LFN cases have often loosely been described as ‘the Hum’.   An approximate estimate is that about 2.5% of the population may have a low frequency threshold which is at least 12dB more sensitive than the average threshold, corresponding to over 1,000,000 persons in the 50-59-year-old age group in the EU countries. This is the group which generates most complaints of both LFN and the Hum.  I will attempt to address the timing discrepancy between the start of high-profile Hum cases in the UK and the USA later in this paper.


The characteristics of the LFN known as the Hum are commonly somewhat narrower than those for LFN as a whole, often being described as sounding rather like a very large bee trapped and muffled in a bottle or a distant but irregularly idling engine.  They seem worse in the quiet of night and very difficult to screen out, sometimes even with earplugs.   A comprehensive search suggests these too to be very similar to the properties of high-level infrasound.   Some have suggested that the Hum many be a single worldwide phenomenon, see for example Deming (2004) [5]   whereas others have suggested the involvement of otoacoustic emissions [6] and still others have suggested it is due to mixed source  and even ground borne infrasound components [7-9] .   A modern potential cause of some cases of the Hum may be wind turbines [10]  but since the Hum as a general phenomenon predates these then they are unlikely to be the sole or the main cause.   Wind turbines do however produce fluctuating noise and their pressure waves can modulate other noises in a phenomenon known as OAM.     Low frequency noise specific criteria have been introduced in some countries, but do not deal adequately with fluctuations. Validation of the criteria has been for a limited range of noises and subjects.  Earlier causes of the Hum have been suggested to be motorways and gas mains [11]  .  More recently Leventhall has suggested that the Hum could have its causes seated in things like Ventilation Fans, Diesel Engines, Buses, Diesel Generators, Air and Water Pumps.   Whereas MacPherson [12] has re-iterated the idea of traffic and has also suggested air traffic, mining, blast furnaces, large air ducts and electrical sub-stations as possible causes.   Given the history of the Hum and the fact that it was first observed in the UK and more latterly in the USA and both countries have all these forms of transport and pieces of infrastructure it is at first sight very difficult  to see how any of them could  account   for the Hum as a widespread and ‘general’ phenomenon  although I will revisit this angle later in this work, especially from the perspective of electrical sub-stations.           


When a local authority attempts to investigate a noise, they will often obtain noise levels at the façade of the property.    A sound level meter with appropriate weighting is employed. When more detailed information about a complex sound is needed, the frequency range can be split into sections or bands. This is done electronically within a sound level meter.


These bands usually have a bandwidth of one octave or one third octave. More advanced instruments may be able to give a narrow band analysis of the noise data. This may be an FFT (Fast Fourier Transform) or information in 1/12 octaves.


An octave band is a frequency band where the highest frequency is twice the lowest frequency.


For example, an octave filter with a centre frequency of 1kHz has a lower frequency of 707Hz and an upper frequency of 1.414kHz. Any frequencies below and above these limits are rejected. A third octave has a width of 1/3 of that of an octave band.   A 1/12 octave band has a frequency width of a quarter of that of a 1/3 octave band.  An FFT system can be adjusted to any frequency window and any frequency step but is far more expensive.   Basic sound meters retail for as little as £70  but octave band and FFT based systems may be  circa £5000.


Different countries all apply slightly different methods and weightings to the assessment of LFN [13].     Likewise, different countries estimate the attenuation due to the façade differently.   What is interesting is that at frequencies   as low as 8 Hz is that facades provide zero attenuation and there is some evidence to suggest that even lower frequencies may actually be amplified by up 7dB by the façade, see Figure 2.    For example, it  has been stated that the levels of infrasound from wind farms are too low at property facades to be perceived within but that amplification by window vibration may be a mechanism for actual perception and similarly with other weak airborne noises. 

Figure 2   Typical Façade attenuation near a wind farm for wind speed =1.1m/s.



Some have discussed the need to completely re-assess the way we deal with and assess LFN in that no one acoustic weighting is ideal and in that in psychological acoustics pure tones are perceived more than broad band noise   and those intermittent or pulsating pure tones can be   even more annoying.     Thus, only by discrete frequency analysis can one really know what is going on.      Councils and Local Authorities have tended to oppose such suggestions due to equipment cost limitations.          


Air borne noise at the façade is not the only way noise can enter  premises.  The other possible route is by ground borne vibration, involving either bulk, shear, or surface ( Rayleigh) waves.   Perhaps somewhat unexpectedly, Leventhall (2003) [14]  claims that LFN cases due to ground vibration are in minority or extremely rare, this is perhaps somewhat surprising since he discussed the it as a major cause not three years earlier! [8].   Even more  surprising given that   ground borne vibration is virtually impossible to stop and can travel very large distances relatively un-attenuated, see  Srbulov, Ground Vibration Engineering: Simplified Analyses with Case Studies and Examples [15].


Ground borne vibration can of course couple into house walls and floors from whence it may be re-radiated.    But what if more complex effects can also occur? What if houses could act as acoustic filters and amplifiers or sound sources in their own right?  What if walls has ‘mouth as well as ears?   I will discuss ideas later.  


The picture we are building up is fast suggesting that the true picture of LFN must be a function of both source and sink; i.e. the architectural nature of the premises the sound finishes up is as critical to its perception as the nature of the source itself.    Hitherto   this is simply not and never has been considered by local authorities in LFN assessment.      Moreover, such authorities are usually obliged to seek a single offender when it comes to noise problems.       And even considering Occam’s razor, things are rarely that simple.      For example, one of America’s most well-known LFN problems from the 1990’s, the so-called Kokomo Hum was shown to be the result of two separate acoustic sources beating together.   These were a 10Hz tone with measurable harmonics up to 60Hz and a 36 Hz tone.  A 360-degree rotating dual-microphone boom was used outside at three locations to localize the 10 Hz tone to air  compressors  in  an  industrial  facility  near  the  center  of  the  City.  The  36  Hz  tone  was  localized  to  a  cooling  tower  on  the  roof  of  another  industrial  plant  in  a  different  section of town. In each case, the tones were clearly detectable at more than a 1 km, see Cowan (2008) [16].     When the tones were alleviated most but not all residents got relief.          One possible reason is sensitization.  Once an individual is sensitized to an environmental sound, they will sub-consciously seek it or a similar sound  or frequency components out,  possibly even hearing it or perceiving  at much lower levels, rather akin to a radio signal almost buried in noise.     


 The only other facet of the Kokomo investigation noted by Cowan  found higher than average   60 Hz A.C.  magnetic fields in some of the affected resident’s houses of values between 3-50 mG.   Based on these and the values  of magnetic fields found in the homes of dwellers in the USA’s  other Hum hotspot, namely Taos, New Mexico, it was suggested that possibly electro or magneto- phonic hearing might be a mechanism for some to perceive the Hum.  Very recently indeed evidence has been presented to suggest that just like birds, humans  may actually be able to perceive the earth’s magnetic field and subtle changes therein, see Wang (2019) [17].  Could this be an additional mechanism for perception of and/or entrainment to the Hum?     As I mentioned earlier there is often learned aversion to low frequency noise.   It certainly seems with some individuals that once one is sensitized to one source one may also become receptive to similar sources be they acoustic or electro-magnetic or a combination of both and this may, to some extent, have accounted for the incomplete resolution of the Kokomo Hum case      and even ongoing reports to this day.  


Hypothesis:   a new way of assessing LFN

 I propose a new way of dealing with LFN where we examine not only the noise levels at the façade but also the precise frequency components as well.     Moreover, frequency components should also be investigated within affected premises in multiple places such as the center of rooms, the windows, the walls and floors.   This will allow separation of sources and differentiation routes of entry i.e. airborne or ground vibration. It will also test to see if any architectural features are generating sounds for other reasons.  I  see this  as essential,  as undoubtedly some more complex cases of LFN or Hum will be found to be due to beating of multiple acoustic components which are  either arriving from the same source via multiple paths  or from different sources, as was the classic example of the Kokomo Hum.    Non-linear acoustics within the human cochlear can allow beats to form, thereby increasing perceived annoyance.   Because professional octave band and FFT analyzers are prohibitively   expensive for home use,   I decided to use a piece of Freeware designed primarily for the Ham Radio market and known as Spectrum Lab [18] .      Whilst it is possible to use Spectrum Lab on a Laptop computer in conjunction with the built-in electret microphone, I would caution against this especially with a Laptop with a conventional hard drive  and cooling fan both of which will of course generate internal chatter noise and vibration usually in the region of 90Hz.  


A technique I used instead was to create an external low frequency microphone using a bass loudspeaker ‘in reverse’.   Such a system as a ‘big ear’ and it was first used in the Taos Hum investigation.  For those not familiar with the idea, it is the same as a homemade sub-kick microphone [19].  I used a 12 inch woofer speaker.       Such a microphone can, if necessary, be used on the end of a long  and screened cable to get the laptop into a completely different room.   Moreover, the low frequency response can be increased by sticking additional mass onto the centre of the speaker cone.   I found that loading the cone with a 1kg mass could give a response down to a few Hz.


Personal experiments 

Doing science with a phenomenon like the Hum is notoriously difficult.  This is because most reports are anecdotal unless one is privileged enough to conduct interviews with afflicted subjects.   Even then, when asked for example, to ‘tone match’ the Hum many come up with different frequencies.   I do not view this as entirely unreasonable given that often skewed sound spectra with multiple narrow band ‘carriers’ some of constant amplitude and some pulsating are often found in the homes of people afflicted by the Hum.  So yes one could label the Hum a ‘psycho-acoustic’ effect but non-the-less a very real effect rather than an imagined or internal effect such as tinnitus  or an effect perpetuated by conspiracy theorists  or the ‘tin hat brigade’!   


The author and his wife are the experimental subjects here. Both have first -hand experience of the Hum. We have lived at his present address for some 17 years now.  About two years after moving in, the author’s wife began to complain of a strange intermittent buzzing noise in the dead of night which was causing her significant distress and sleep disturbance.    Eventually after focusing intensely, and wishing I hadn’t,  the author too heard the sound.   Some research was conducted, and it seemed that this noise bore all the hallmarks of the acoustic phenomenon commonly referred to as the ‘Hum’.        In common with all Hum hearers we went through various stages including trying to locate the noise locally outside the house but to no avail.   We tried switching off all the electricity to the house again to no avail.    


I had read the work of Dawes way back close to start of the Bristol Hum.  Some at that time had claimed the Hum was associated was the then  Radio Navigation System known as Loran C but  Dawes eventually  claimed that the Hum was somehow connected with the electricity supply industry and was still doing so recently [20].   Dawes’ early work, published only on his then website,  referred to measurement of a varying AC  ‘wall’ voltage measurable by sinking meter probes into house plaster work  and moreover the finding that this voltage was not in phase with the house mains voltage, that it varied by time of day and that it was still present even with the house electricity supply turned off.    Unfortunately, these original references  no longer appear to be on-line. I decided to duplicate Dawes’ experiment.   Sure enough an AC wall voltage was found.   Moreover, placing a powerful magnet changed the induced voltage.  


Original wall voltage  15mV AC over 15cm       Wall voltage with magnet in situ

Thus, I could only assume that brick must be a piezoelectric material with magnetic properties to boot.  A little more research produced the evidence I had been looking for.  Namely that brick is indeed both a piezoelectric material as evidenced by acoustic emissions during bending  [21], an electret style supercapacitor because of its porosity [22] and together with cement is also a magnetic material [23].   The thought occurred to me that not only would house walls radiate ground borne seismic vibrations, but they would also act as electro-acoustic converters and magneto-acoustic converters to boot, directly converting changing electric and magnetic fields to sound.  I would thus most certainly expect a certain acoustic ‘richness’ if I were to place my big ear on the walls of the house, especially the load bearing walls most stressed so expected to produce most piezo-electric effect.       


The first experiment I conducted was simply to take a set of narrow band acoustic measurements in the centre of the master bedroom of the house, one affected badly by the Hum.   Two scientific subjects were used in the experiment, namely my wife and I, both readily perceive the Hum.    I took simple frequency only measurements  using the ‘big ear’ and Spectrum Lab  on  several occasions when the Hum was present,  one occasion when there  was an obvious LFN from a ‘boy racer’s’  sound system outside the house and on the only occasion I could record when the Hum was not perceived.    The results are presented in Table 1 below:


0-10 Hz

10..20 Hz

20..30 Hz

30..40 Hz

40--60 Hz

60..100 Hz

100..200 Hz

>200 Hz


3.3, 5.5, 9












50 P,58





















Y7 (LFN)








50 B/C




Table 1

The most striking commonality in the above table is the presence of infrasound below 20Hz on each occasion when the hum is perceived by both subjects.  The most intense Hum, condition Y6,   seemed to correspond the presence  of most narrow band infrasound below 20 Hz and with most narrow band sound below 60 Hz with acoustic silence at higher frequencies.    When the Hum was not present there was no infrasound,  no narrow band acoustic sound in the region of 30-33Hz  and broad band acoustic sound at and above 50 Hz.    


An example of the  spectrum lab plot  obtained is shown for Y1 is shown below. 

Spectrum Lab Plot Y1 

The original screenshot has been reduced in size and the frequency scale was not clear, so I have superimposed it in larger print. 

Note how the 50 Hz signal appears to be pulsating.   When the Hum was not present the 50 Hz signal amplitude appeared to be more constant.

In four out of the six recorded Hum instances I also noticed frequencies in the region  of 17 Hz.   I wondered if I could somehow synthesize a Hum like effect using acoustic synthesis.   I used freely available software by V.Burel [24]  ran on  additional laptops and speakers.   Following the above subjective effects and initial hypothesis that the HUM needs more than one external component for perception and that internal perception is a function of human audition and that at all the above HUM sites infrasound appears to be present at two or more frequencies below 20 Hz in addition often to low frequency sound in the region  of 30 Hz it was decided to see if conditions could be set up experimentally which would synthesise the HUM.    The V.Burel software was set up on two other computers and all three computer sound cards were fed into separate loudspeaker systems.  The subjects sat about 1 metre away from the loudspeaker systems and listened binaurally. It was found that HUM like effects could be perceived for any infrasound in the region of 9-17 Hz when outputted at similar amplitude to any audio tone in the region 29-75 Hz.   Particularly disturbing effects were obtained as follows:  

16.55 Hz

49.731 Hz

Hum-like effect

16.689 Hz

33.092 Hz

Hum-like effect

16.689 Hz

50.261 HZ

Hum-like effect


A third tone was not always necessary but equal amplitude tones of 4, 10 and 29 Hz gave a the most y disturbing effect and the subjects felt quite giddy for several hours afterwards.

HUM like effects with typical quasi-periodicity were also maximised for two tones one audio in the range 29-75 Hz one infrasonic wherein the infrasonic tone fell such that its third overtone was within 2Hz or so of the audio tone.  The results of these experiments give very strong support to the notion that at least one component of the HUM is infrasonic and that it might never be possible to measure a single HUM characteristic or single per se in that the quasi-periodicity which subjects experience may well be an internal effect related to the latency periods of linear and non-linear products in hearing. Non-linear acoustic products of the human cochlea are known to be of the form 2f1 – f at frequencies of the order of a few kilohertz  but odd harmonic generation n f1,  n=1,3,5 etc. is known at lower frequencies in line with the above result, see Gorga et al (2007) [25].        


In the next experiment, I decided on another occasion   when the Hum was being perceived and attempted to measure the frequencies present  outside the façade,  on the inner surface of the double glazing window  and the inner surface of the supporting wall.  The results are shown in Table 2.      .    























Table 2


What is particularly apparent is the lack of detail present at the façade.     The frequency of 4 Hz is interesting and may possibly be associated with wind turbines, see Jakobsen (2005) [26].      I would expect this frequency to vary somewhat with wind speed.  For example,   with three blades spinning at 16 RPM, a blade passes the mast 48 times per minute, or once every 0.8 second (i.e. 0.8 Hz). Typical measurements of wind turbine noise showed a peak in the spectrum at 0.8 Hz with additional peaks at 1.6 Hz, 2.4 Hz, 3.2 Hz, 4.0 Hz, 4.8 Hz, 5.6 Hz, 6.4 Hz, 7.2 Hz, and 8.0 Hz, see measurements by M.G. acoustics, Canada [27].  Although looking at the results airborne wind-turbine infrasound at such very low frequencies did not seem to be an absolute pre-requisite for perception of the Hum by either   subject, its presence certainly seems to cause an intensification of the effect.    The Hum in Britain of course pre-dates abundant wind farms.


It is uncertain what the frequency of 122 Hz was.    At this point, the origin of the frequencies at 9Hz and those between 31 -88 Hz was uncertain.  They might simply have been  ground borne vibrations transmitted by the walls and windows.      Vibration of the ground or of a building beneath one's feet is generally disliked and can be annoying. Man is sensitive to mechanical oscillations ranging in frequency from well below 1 Hz up to at least 100 kHz. This range of sensitivity to vibration is much broader than the range of human hearing.   Generally, vibration of relatively low intensity of the kind induced in buildings and other structures excited by traffic or industrial operations outside.  The human vestibular system is sensitive to vibration i.e. acceleration in the frequency range .1-4 Hz.     The skin, muscles, and tendons sense vibrations up to several KHz.    The visual system can also perceive vibrations because of differences in the acceleration between the eye and its socket.   Human thresholds for vibration as a function of frequency are shown in Figure 3 below: 


Human perception to vibration proposed by Goldman (1948 apud Goldman... |  Download Scientific Diagram

Figure 3


For frequencies above about 20 Hz  ground attenuation is such that even any vibrations from local sources such as industrial sites, road  or rail traffic  would be below threshold at distances more than about 100m.    Since the Hum is heard at nighttime in the author’s residence and when no cars are passing and when no trains are running, the only other possible source of local ground borne vibrations might be electricity sub-stations. Power transformers and their associated core vibrations are known to be able to supply complex acoustic spectra [28].  

 The author’s property has some four separate substations with 400m.  One of these is associated with a local college of technology and the other three with a University Halls of Residence site.  Either ground borne vibrations were coming from them or  alternatively, it could be concluded that the acoustic sound emanating from the walls and windows might be in some way be internally generated.   


The question is posed what if the walls themselves could act in some way like a loudspeaker?   We have heard the expression ‘walls have ears’, but could they have ‘mouths’ as well?    This may then, in some way, marry up with the ‘wall voltage’ first observed by Dawes in connection with the famous ‘Bristol Hum’.  


I have shown  that brick is a piezo-electric material [21].  Hence if it has an alternating electric field across it then sound radiation at that field frequency would be expected.   Similarly,   brick and cement are known to also have magnetic properties.     Thus, if they are magnetically polarized in the  earth’s  magnetic field and further subjected to a varying magnetic field I would expect the potential  for sound radiation in this manner also.    The converse could of course also be true. That is if a sound field were to be directed at the walls from within, electric and magnetic fields could be generated.       There was of course no explicit sound source within the house, but this does not preclude  the possibility of externally driven room modes and resonances.  The house also has a chimney  in the front room, and I calculated in a  possible Helmoltz resonance  in the region of 9 Hz for that.       It is vaguely possible that fluctuations in outside air pressure could excite chimney resonance more strongly sometimes than others which could in turn  intermittently drive room resonance.  If this were the case, I would expect a reciprocal piezoelectric effect to generate frequencies  of the region of 28 Hz within the wall but this was not observed. 


To fully test the above  hypothesis, I would need to measure the electric and magnetic fields at the walls as a function of frequency.   My electric field sensor was simply a pair of  nails driven into the plaster work separated by 1m and connected directly to the laptop running Spectrum Lab.   For a magnetic field sensor, I utilized one winding of a torroidal mains transformer. Much to my amazement the magnetic and electric fields were still present with the mains electricity supply to the house switched off at the master switch.           The frequencies present are highlighted in Table 3.









60,67 Broad







Table 3

I concluded that the magnetic field is providing the 9 Hz found in the wall audio signal and moreover that between them, the electric and the magnetic fields provide every other component in the 0-10 Hz range for ALL Hum events  other than the frequency of 4Hz which I had already established as being airborne.          In the band 10-20 Hz I find that the electric and magnetic fields between them can provide components heard on Hum events Y4 and Y5.    I conclude that the frequency of 28 Hz is probably not relevant.     The magnetic frequency of 30 Hz is very close to the 31Hz of Hum events  Y1, Y2 and Y7.      The narrow band frequency of 50 Hz is common to ALL Hum events.  The frequency of 60 Hz is common to Hum  events Y5 and Y6.   It is possible that there is non-linearity in the system.   Certainly, by considering sum and difference frequencies of the observed electric and magnetic frequencies together with the 4 Hz  and 122 Hz airborne  frequencies, one can arrive at virtually  every other frequency observed within the room during any of Hum events Y1-Y7.     


In addition to the home-made magnetic field probe I also experimented with  a tri-field  or  Cornet electrosmog meter (Cornet Microsystems Inc. Type ED88T), typical of the sort that might be used to measure field strengths in an amateur radio station.   I was intrigued to see that the low frequency magnetic field strength present on the house outside wall almost doubled  from .3mG to .5mG  and also pulsated when the Hum was present.   Below are two photographs of the meter the first with no Hum present and the second a few minutes later in this case  at 2122 GMT on August 9th  2020 when the Hum had first  ‘switched on’,  see Figure 4.


Figure 4.


Interestingly, observations of high AC background magnetic fields   have been made for both the Taos and the Kokomo Hums.   The question arises as to how such fields can be present even when the power to the house is turned off.     The answer may be in the PME earthing system which are now more or less ubiquitous  in Britain and are described by SP Energy Networks [ 29].    Essentially even when the electricity supply is switched off, the earth conductor in the house is effectively connected all the way back to the substation.      It seems to the author that any phase imbalanced currents may have the potential to flow through the earth under the house.    Hence there is a potential coupling loop albeit high impedance involving the earth as sockets are grounded into the brick and plaster work and the earth under the house itself.   


If I have a varying magnetic field, I expect an induced electric field by the very nature of the architectural materials I have described.  Thus, I   also tried connecting the wall voltage probes which were identical masonry nails some 14 cm apart directly to an auto-RMS reading multi-meter.     Present were both DC and AC voltages.   The DC voltage varied between -17mV and -23mV with respect   to the upper probe.    The AC voltage was of the order of 30 mV   and was constant for periods of several minutes when no Hum was present but was varying continuously and randomly from 24-39 mV when the Hum was present, in this particular   case on the evening of 18th August 2020 starting at around 2210 BST.        Similar voltages could be measured between the brickwork on the outside of the cavity walls of the house too. 


The pulsating magnetic field at 50 Hz and the randomly varying AC wall voltage seems very consistent with the presence of pulsating 50 Hz audio in the room as evidenced by the spectrum lab plot Y1.    At least in the author’s    house we would truly seem to be living inside a loudspeaker. 


Since this is a very radical hypothesis, I set up yet another experiment using an earth loop between two outside ground connections, one an old earth near the gas main prior to PME  and one an old earth connection on a disused telecom pole some 50 meters away and  again     using Spectrum Lab to test it in more detail and provide frequency spectra.    I would expect considerable changes in ground currents flowing under Hum conditions, if ground currents rather than just ground borne vibrations alone are a feature of or are proving the driving signal for my newly discovered ‘speaker’.    I would also expect to some extent frequency spectra in the ground currents which might match and mirror the audio and internal electromagnetic signals I have been observing.


Signals were recorded by a standard laptop sound card and laptop running 'Spectrum Lab software' providing an  extremely effective near real time FFT program with optional waterfall output.



The system was tested for sensitivity by seeking to see if it could record natural radio emissions like whistlers. Using the full scan range of DC to 5.6 KHz, a whistler was captured early into the first night’s experimentation. Whistlers are due to lightening interactions in the earth ionosphere cavity.


Indeed there is so much energy injected from whistlers worldwide with up to 2000 separate lightning storms going on at any one moment that it is thought they dominate ULF production in the magnetosphere.


Spectra were recorded at various times on the 10th and 11th September 2011 when the Hum could be either be heard constantly, at other times when it was fading in and out and finally at times when it had apparently ceased. These spectra were recorded across various frequency ranges of interest to home in on specific detail. Typical ranges of interest might be 0-700 Hz, 200-1700Hz and 0-5.6 kHz.

Results and Discussion of the Ground Loop Experiments

As predicted these results are in striking support of the hypothesis.  There are dramatic changes in the ground current spectra when the Hum is present. 


The results are simply image files of the frequency-amplitude-time waterfalls obtained in each experiment. They yield up striking differences across significant frequency ranges between the occasions when the Hum was present and the occasions when it was not. After showing the recording of a 'whistler' which validates the receiving system, it is then better to continue with presenting the displays obtained when the Hum was not present and discuss the features seen on these. This will be followed by the same for cases when the Hum was present. In all cases there is a significant wealth of information on each waterfall display.

System Validation 2048 GMT   0-5.6 KHz


A strong whistler can be seen commencing at about 3.2 kHz and decaying in frequency down to about 2 kHz. This is followed by a sequence of weaker, shorter lived natural radio events between 5 and 2.7 kHz. UK mains frequency harmonics can also be seen.


1047 GMT 0-1700Hz No Hum perceived

0-500 Hz  2156 GMT Strong Hum being perceived


Even as well as odd harmonics of the UK mains can be seen and there are pulses of other frequencies at about 5, 10, 20 and 25 Hz also at 60,75 95,120,125,130,160,165,170,190Hz and various higher frequencies, with repeated similar frequency spacing. Note that some frequency separations of 20 Hz are present a known characteristic of power line harmonic radiation.



 Hum perceived expanded frequency range to 1500 Hz.


Weaker Hum being perceived

In addition to UK Mains fundamental and harmonics, there are weak signal bursts at 10, 20, 40, 60,120,140,160,175 and 275 Hz.


When the Hum is perceived most strongly at the author’s residence frequencies of 5, 40,50 and 60 Hz are found in the audio spectrum.  These frequencies have also been identified in the above two ground current spectra.  Interestingly acoustic frequencies of 40 and 50 Hz have been detected by National Grid in relation to a Hum in the Wrexham area.     Moreover, frequencies of 5,10,30,40,50 and 60 Hz were found in the magnetic spectra in the author’s master bedroom.     


General Discussion


i)                   Could the Hum  as a simple sound ever be recorded?


The brief answer is no.   A more complex answer is that individual frequency components of the Hum or even whole frequency spectra can easily be recorded as seen above but that the precise sound any individual perceives cannot for it will depend on their precise cochlear mechanics according to the equations quoted earlier.    However, it may be possible to make either electronic circuitry or a computer algorithm to synthesize a noise like some perceive the Hum.   Since the Hum is so low in frequency it may be necessary to Octave shift the frequencies to give others any idea of what is perceived. This is something I may well explore in future.   


ii)                  Can the time history of the Hum be accounted for?

There are accounts of a Humming phenomenon as far back in history as the dawn of the industrial revolution which was called the ‘Hummadruz’.   This was a distant sound like a swarm of bees heard outside at dusk and dawn a few kilometers from Northern Mills and the like and thus not thought to be the same phenomenon as today’s ‘indoor only’ Hum.       


What is really intriguing is that contemporary Hum reports in the UK began in the 1970’s, some 20 years prior to those in the USA.  Both are developed countries with very similar infrastructure.  UK buildings are often somewhat different from US buildings especially in rural areas where UK buildings contain more bricks and mortar and US buildings more timber.  Double glazing was also starting to be used much more prevalently at that time.  The author noticed that the Hum intensified at his premises after installation of double glazing which basically acts as a low pass acoustic filter.   Fox suggested that new gas mains and motorways may be the cause of the 70’s Hum  but the US has similar infrastructure, so I doubt this.  Britain built its first pumped storage power stations in the 1970’s which may feature.  Certainly there is a  known case where the then CEGB actually paid compensation for disturbances [30].  Yes, the US too has hydropower but being a much larger land mass may make ground propagation effects less noticeable.    I asked myself what further changes in infrastructure began in the UK in the 1970’S not paralleled in the USA.   Britain had TT earthing  until the 1970’s [31] ,  and then progressively changed to PME.         Presently in the United States  according to the National Electrical Code and Canadian Electrical Code since the mid 1990’s the feed from the distribution transformer uses a combined neutral and grounding conductor, but within the structure separate neutral and protective earth conductors are used (TN-C-S). The neutral must be connected to earth only on the supply side of the customer's disconnecting switch.  With the advent of this grounding system, the Hum began.  I also wondered if by the 1970’s there was sufficient power density in the US power grid to couple with the UK power grid via ionosphere modes and harmonic radiation events.    


I asked myself the question could PME be exciting wall voltages and hence wall noises from our wall ‘loudspeaker’. Together with the very strong experimental evidence presented above, there also exists an online Hum complaints mapping system.  I superimposed onto this map the type of earthing systems present around the world, see Figure 5.


Figure 5

It is clear that in parts of the world without PME earthing there are virtually no Hum reports.  Conversely the parts of the world with PME and renewable energy seem to show the most Hum reports.  For those interested I have expanded on this topic on my personal website.  


Are there any other situations where we would expect the Hum?

As I explained earlier, there are several references to and reasons for   LFN sensitization.  One hypothesis is that it is a phenomenon connected with the fight/flight response [32].   Both the author and his wife have lived through two moderate earthquakes.   A mini-survey I conducted suggested that a considerable percentage of Hum hearers are earthquake survivors.   It may be once one has been exposed to the low rumble of an earthquake one generally becomes more sensitive to LFN and vibration in general.   


The Hum and Cars


From the above results it is apparent that when the Hum in the house is present, we are effectively living in what is almost tantamount to a pair of   acoustic and magnetic comb spectra.  One possibly associated with wind turbine infrasound and the other associated with the electricity grid.    It is perhaps not that unremarkable then, that after such sensitization, any other source which could provide a ‘comb’ might also provide Hum like effects?   


On a number of occasions both subjects have noticed hearing Hum like noises when parking up our cars upon engine off and one alights while the other sits in the vehicle. I asked myself could there be some feature of the vehicle’s electrical system that provides the Hum and a feature of the body that acts as a loudspeaker.     I noted that when the vehicle   stops, and someone alights the interior light comes on and then slowly dims over a period of time.  The process is the same for my Vauxhall car and my wife’s Landrover car.  The process is essentially controlled by a PWM system.   I was able to detect both magnetic and acoustic pulse in both cars as the lights dimmed as a ‘comb’ spectrum with a fundamental frequency in the region of 15 Hz.  I speculated that the acoustic generation was due to out of phase earth return currents flowing from the interior lamp down both frontal roof supports towards the bonnet and front wings and over the roof to the rear of the car.  Placing the Cornet meter in these positions confirmed my suspicions and it could be seen pulsing. 

Once the car interior light has gone out, the pulsing stops and so does the Hum like effect.  






Parked cars under high voltage power lines.

The other place where both subjects have perceived a Hum similar to the one we perceive at home is in a parked car underneath the 400 kV power grids.  I speculate that eddy currents are induced in the various parts of the car body and chassis which vibrate rather like an EMAT in ‘speaker’ like manner accordingly.   Using a laptop with spectrum lab one is able to capture spectra under power lines very similar to the one of Y1 above.  For clarity,   I should point out that the noise perceived is not corona noise.   I should also point out that it is not always possible to perceive the Hum in this manner.  Just as it comes and goes at the author’s location, it also comes and goes under power lines.  It seems to depend on precise ratios of sub-harmonic and harmonic content present and whether or not the amplitudes are fluctuating.  Modern power systems with Smart Grid Control, solid state DC links and the like are on occasion known to exhibit parametric instability [33].    


This links back to the idea of sub-stations. I ask, could the Hum in people’s houses be a combination of ground borne acoustic signals (vibrations) from such sub-stations accentuated by coherent arrivals of strong ground currents exciting the loudspeaker effect in walls?   Some time ago I was asked to assist a resident in Charlbury  who was eventually  drive out of her home by the Hum. Nestling in the Evenlode Valley, sequestered down winding B roads but with its own railway station, Charlbury is a vibrant yet picturesque Oxfordshire town. The local council supplied acoustic spectra recorded outside two nearby substations and in her home.   Most of the frequencies present at the sub –stations were also present at the home.  What they could never account for was the lack of attenuation.    The notion of ‘walls have mouths’ together with the newly discovered architectural speaker discovered here goes a considerable way to clarifying the matter and very interestingly indeed many of the frequencies observed for the Charlbury Hum have also been observed in the author’s house.  


Is the Hum getting more intense or is something sensitizing more people to the Hum?

Certainly, as far as the UK is concerned there do seem to be more Hum reports of late. Most of Britain now uses the PME earth system and most houses in Britain have double glazing.  The only infra-structure that has continued to grow almost unchecked of late has been renewable energy infrastructure especially wind-farms.     For the Hum at the author’s house it seems that airborne infrasound provides components found both at the façade and on the inside of the window and house whereas the wall ‘loudspeaker’ provides the rest of the frequencies.   The Hum levels at the author’s house depend strongly on wind-speed and direction a pointer that the local wind-farms may be involved.         Wind energy also causes frequency instability [34]  and flicker [35]  on the mains electricity supply and additionally phase imbalances which need SVC compensation [36].   Potentially thus it is feasible that wind energy alone is, although not the sole cause of the Hum noise, a highly contentious catalyst for its increase in both prevalence and intensity especially as there will be a quasi- coherence between the airborne acoustic signals from the turbines and the sounds arising from  the newly discovered ‘speaker’ in our house walls .    


There are a few anecdotal reports of deaf and partially deaf people perceiving the Hum.  There are reports that people can get relief from the Hum by going into deep underground limestone caves but not copper bearing caves.  In those cases, one must perhaps retain the possibly of low frequency electro- magnetic perception an area where the science is lacking and more work needs to be done.   


Further work: we know what but do we know where? 

OK, so we now know what the Hum is and we also know what makes it different from other LFN’s. 


In summary for Hum in the author’s house infrasound below 20 Hz, and some pulsating acoustic emissions at mains related frequencies are required together with subharmonics and possibly harmonics.  The Hum is strongest when there is an apparent ‘comb’ with a 10 Hz gap.   This appeared to be the same or very similar to the case of the Hum in Charlbury.  It seems that in certain locations the walls of houses can produce these frequencies by acting as piezo-electric or magnetic loudspeakers or both.      It seems this is not simply a ground vibration phenomenon because there are few if any Hum reports from parts of the world which do not have PME electrical earth systems.   Thus some of these frequencies appear to be ‘induced’ by strong electrical ground currents which usually appear at night.  


At least in the author’s house the Hum is heard to physically switch on, usually between 10 and 11 pm although very occasionally it has been heard in the daytime as well, especially at weekends.   The author lives in Gwynedd, North Wales.  Recently there have been Newspaper reports of a Hum  which ‘switches on’ every half hour or so throughout the night from Welsh locations as far apart as Cefn y Bedd, Llay, Deganwy and Llandudno.    National Grid have investigated this Hum and found frequency components of 40 and 50 Hz yet state that they cannot identify the source  [37].   These frequencies have also been found in Hum observation ‘Y6’ in the present work and in the ground current spectra of the present work.


The question that remains to be answered is  do the changes in the frequency spectrum that bring about the Hum happen at a local level or do they happen at a National level?   Certainly the observation of the   Hum in a car under power lines some considerable distance from the author’s residence is possibly suggestive of a wider area or even National phenomenon. Alternatively, does something happen to the electrical load on our local substations which brings about the Hum or does something happen to the ‘frequency’ quality of the voltage driving those substations?  


Both are feasible and both now warrant further and urgent investigation.     MacPherson has been investigating the famous Windsor   Hum.  In 2011, the Zug Island area was identified by Canadian scientists and Ontario's Ministry of Natural Resources as the source of mysterious rumblings and vibrations, known as "The Hum", that have plagued hundreds of area residents with cyclical vibrations reportedly being felt in the ground up to fifty miles.  He has suggested Electrical Blast Furnaces as a culprit.  These are known to cause enormous harmonic instability in electricity supplies.


At a local level it is feasible that any equipment operating on time switches could impact phase balance and earth currents, for example this could be loads such as;  economy 7 heating systems, overnight water and sewage pumping and the like.  


At a National level,   our pumped storage power stations usually start to pump water up to their upper reservoirs at about 11pm.    Moreover, more of the load on the power grid is now controlled in a digital manner via the Smart Grid.  


At the author’s house there are potential pointers for both above.   For example, he has shown that the Hum intensity varies considerably  depending which specific motor generator ‘pairs’ are pumping water up to the top reservoir lake at the Dinorwig Pumped Storage  Power Plant.


More locally, the times of the Hum seem to follow a sinusoidal pattern in line with the local coastal tides with the Hum maximizing at periods about halfway between high and low tides.  The author believes there are two    possible explanations for this.  Firstly, tides effect stresses within the earth itself and can change seismic  propagation.    Secondly,  Bangor has a  sewage discharge point which possibly would pump its outflow at such times.   Although according to the following reference, most sewage is discharged between 9-10 am and 6-7 pm. Smaller  discharges at other times are possible [38]. 


Tidal effects are also known to change ground water flow which could impact on ground electrical currents.   In the author’s home area the Menai Strait is a tidal strait which could potentially influence stray ground current flow between the mainland and the Island of Anglesey. 


Interestingly, all of the above might not only impact power frequency and quality but also associated ground borne vibrations as well.    Certainly, in the complex soundscape we live in, ground borne vibrations from any of the other utilities might not be ruled out.      The author has contacted the water utility company and has been advised that water in his area is gravity fed rather than pumped.   The author’s own study of other utilities suggests that at least in his residence they are not the cause of the Hum, although the gas grid could conceivably supply a signal at 36 Hz [39].       For completeness, the author now proposes to make a subsidiary study of the sewage system!    



The author has come a considerable way along the path to super-sleuthing the Hum but perhaps  a little more remains to be done.   It is hoped this article will act as a catalyst for others to make their own measurements and enquiries,   so that once and for all the world will have a solution to the enigmatic ‘far more than just a noise’ problem which is the Hum.     




1.      Moller and Pedersen (2004),;year=2004;volume=6;issue=23;spage=37;epage=57;aulast=Moller


2.       Salt and Hullar ( 2010)

3.      Leventhall (2004);year=2004;volume=6;issue=23;spage=59;epage=72;aulast=Leventhall


4.      Branco and Alves-Pereira (2004)


5.      Deming (2004)


6.       Frosch (2013)


7.      Rushford et al ( 1999)


8.      Leventhall (2000)


9.      Montago and  Gushika Uesu (2019)


10.  Salt and Lichtenhan (2104)  file:///C:/Users/CHRIST~1/AppData/Local/Temp/Salt-Lichtenhan-How-Does-Wind-Turbine-Noise-Affect-People.pdf


11.  Fox (1992)




13.  Broner (2010)


14.  Leventhall (2003)


15.  Ground Vibration Engineering: Simplified Analyses with Case Studies and Examples by Milutin Srbulov,  Google Books.


16.  Cowan (2008)


17.  C.X. Wang et al., “Transduction of the geomagnetic field as evidenced from alpha-band activity in the human brain,” eNeuro, doi:10.1523/ENEURO.0483-18.2019, 2019.




20.  John Dawes, Twitter Pages,

21.  Papargyris et al (2001)


22.  Bevan (2012)





25.  Gorga et al (2007),



28.  Power transformer acoustic spectra:





33.  Mohamed (2010)

34.  file:///C:/Users/docto/Downloads/Nguyen_grad.msu_0128D_15469.pdf

35.  Moreno et al (2002)