In 2010, the Journal for the Society of Psychical Research published ‘The Acoustic Properties of
Unexplained Rapping Sounds’ by Barrie Colvin, which advanced the theory that rapping sounds produced by (alleged) poltergeists have acoustic properties different from those produced by regular means. In brief, Colvin suggested that whereas normal knocks start at their maximum volume and decrease, poltergeist knocks build up to their crescendo.
The unexplained rapping effects described in this study […] appear to involve the relatively slow build-up of a stress within a material, culminating in an audible sound when the level of stress reaches a specific magnitude.
Here are some examples:
Colvin addresses in his paper the inevitable objections concerning the reliability of his ‘poltergeist’ samples, and acknowledges that further work remains in order to support his hypothesis.
Recently, Chairman of the Ghost Club, Alan Murdie, highlighted Colvin’s work as possibly the first instrumental (rather than anecdotal) evidence for the existence of poltergeists .
As audio analysis software  is freely available these days, I decided to investigate Colvin’s hypothesis for myself.
The first thing I discovered is that good quality recordings of alleged poltergeists are extremely hard to find. I had imagined the main difficulty would lie in distinguishing ‘poltergeists’ from sounds that are hoaxed or have other natural causes. Recordings offered on YouTube and on the websites of paranormal groups were either obvious fakes, or else too faint or poor-quality to allow detailed analysis of their sound wave properties.
My search turned up only one usable recording that wasn’t included already in Colvin’s study. This was a YouTube clip featuring unexplained knocking sounds recorded by paranormal investigators at Borley church . I took a sample from the clip, and was intrigued to discover that its soundwave properties agreed with Colvin’s hypothesis.
But the second thing I discovered was that although Colvin was unable to replicate the ‘poltergeist’ wave form by natural means, other researches have, in the meantime, enjoyed far more success. One of these is Maurice Townsend, who accomplished the ‘poltergeist’ waveform in a variety of ways – for example, by pushing an object rather than by striking it; or by recording noises that sound like knocks, but are actually sharp creaks .
C.J. Romer  has also worked out how to replicate ‘poltergeist’ knocks by natural means. He suggests it requires nothing more complicated than positioning the microphone at a distance from the source of the sound, or by placing the microphone inside a room and then knocking on the wall from outside. In the first case, the effect might be explained by the way sound waves move within a confined space, bouncing from surrounding walls and arriving at the microphone at slightly different times, thus creating the effect of a ‘gradual build’. The closer the microphone to its sound source, the less apparent this effect.
In the second case, sound is passing through two mediums before it reaches the microphone: firstly, the solid wall; and then the air inside the room. This, too, seems to have an attenuating effect on the waveform produced by the recording.
So – case closed? Not quite! Romer presented 12 waveform samples on his website, four of which (obtained from Colvin himself) were allegedly ‘poltergeists’. Romer professed himself unable to take the difference. However, when I took the test myself, I was surprised to have identified three out of four, the fourth being a sample I’d narrowly rejected.
I wasn’t selecting at random. I’d noticed that the ‘poltergeist’ samples had a simplicity in their waveform that the non-spooky samples lacked. The amplitude of the non-spooky samples seemed to oscillate in a more complex fashion than the ‘poltergeist’ samples.
To show what I mean, I’ve coloured the lower and upper part of the sound waves differently. In the ‘poltergeist’ sample the transitions between red and green are more regular and equal (‘tit for tat’), whereas in the non-spooky samples, one of the colours dominates for a while, with the other in a minor role, before the roles reverse (‘yin and yang’).
Was I really onto something? Or was I projecting patterns onto something insignificant? One of the difficulties with the hypothesis is the lack of examples to work with. And even where examples can be found, we cannot be absolutely certain they aren’t natural or hoaxed.
Colvin had taken some of his samples from a commercially available CD of paranormal recordings . It was only after I’d purchased my own copy that I was in a position to conclude the feature I thought I’d identified probably isn’t significant. The samples from the CD simply didn’t display the symmetry that had been apparent in Romer’s samples.
See if you can work out which of the following three sound waves are ‘poltergeist’ and which are natural. Answers are in the footnotes .
I’ve written to Romer and have asked him for more details concerning his samples. I haven’t heard from him yet, but for now I would guess that all four samples might come from the same case. The ‘symmetry’ I thought I’d found would (if that’s true) be simply a feature of the acoustical environment where the recordings were made.
There is nothing paranormal about sound waves – although that doesn’t mean there cannot be a paranormal cause for sound waves. However, this remains to be demonstrated. And having undertaken the exercise I’ve described here, I’m far from confident that Colvin’s hypothesis will lead to the ‘instrumental evidence’ that Murdie suggested. To claim that we could identify a poltergeist from how it sounds is surely as problematic as the claim that we might recognise a ‘ghost’ from how it looks. Few people, these days, would be willing to hold up an alleged photograph of a ghost as ‘instrumental evidence’ for their existence. If the paranormal can be said to lie anywhere, then it must be in the cause of these supposed manifestations, rather than in those manifestations on their own account.
To take Colvin’s hypothesis forward, we would need to establish a causal link between the poltergeist and its knocking sounds. Then we could introduce variables into the recording procedure – such as different positionings of the microphone – to explore the effect on the sound waves produced. Two possibilities come to mind: the first is the establishment of long-term sitting groups, with the explicit aim of producing poltergeist phenomena under controlled conditions (like that described in Owen & Sparrow’s Conjuring Up Philip). The second is that future investigators of spontaneous cases should simply go in prepared and forearmed with knowledge of Colvin’s work, and – please – put the microphone as close to the sound source as possible!
 C.J. Romer, ‘Knocking away the poltergeist evidence? A follow up on Dr. Colvin’s JSPR paper’ (September 6th, 2010).
 Okkulte Stimmen – Mediale Musik: Recordings of Unseen Intelligences 1905-2007, 3-CD box set (Berlin: supposé).
 Sample 1 is allegedly paranormal, from a case in Pusruck, Germany (1971). Sample 2 is me, knocking on the outside wall of a room. Sample 3 is allegedly paranormal, from a case in Thun, Switzerland (1967).