I only know JG as a commentator on this blog, but I asked their permission to publish some comment they made on Dr Barrie Colvin’s paper from the JSPR which was the subject of my last piece. They cheerfully granted me permission, so here is JG’s research. Please note I am not the author; JG is – so questions should be directed to he or she, by commenting here on the blog!
JG writes —
“I decided to test the theory that the delay in reaching peak sound intensity (the ‘attack’), seen in the poltergeist acoustics paper, might be primarily caused by sound reflections from room walls.
First I tried making successive recordings of a knocking sound at increasing distances from the recorder. As the distance increased, the proportion of the sound arriving from reflections off walls should increase relative to that arriving directly from the source. This showed a significant increased attack but not enough.
Then I tried having the sound source around a corner, so there was a wall blocking the most direct path of the sound. This consistently produced an attack around an order of magnitude longer compared to a direct line of sight recording. The actual attack length was highly sensitive to the precise position of the sound source. Trying a large piece of furniture between the sound source and the recorder also produced a significant lengthening of the attack, though not as big.
The most likely explanation for the large lengthening of the attack time, when the source and recorder are separated around a corner, or some other large object, is the difference in different path lengths taken by the sound. With a heavy object or wall absorbing most of the sound taking the direct path, the remaining sound must either be diffracted or reflected to reach the recorder. The diffracted path closely follows the outside of the obstructing objects and always arrives quite quickly at the recorder. However, a significant proportion of the reflected sound must go via multiple wall surfaces in order to circumvent the obstruction. It is this big difference between the relatively direct diffracted path and the much longer reflected route that causes the longer attack.
So, unless the sound source and recorder are in direct line of sight of each other, there is likely to be a significant lengthening of the attack, of the type seen in the paper. The magnitude of that lengthening will depend on the precise layout of the room and be highly sensitive to relatively small changes in the relative position of sound source and recorder. This is because of the many different reflection paths that are possible.
Typical attack times when there is line of sight between source and recorder is in milliseconds. When there is a bulky obstruction between them it goes up to hundredths of a second. This makes sense since sound travels at about 340 m/s in air and the extra path length due to reflection (often multiple) will be in the order of metres up to low tens of metres. This is due to the diagonal paths taken by reflections to cross rooms that have wall lengths typically of a few metres.
Thus, if sound reflection is the cause of the slow attack in the poltergeist cases in the paper, you would expect the delay to all be in the region of hundredths of a second. Much longer or shorter would suggest that there must be some alternative explanation.
I decided to see if the figures from the paper itself agreed with this prediction. It is just possible to make out the time scales in the figures in the paper. Thus, it was possible to quantify, approximately, the attack times in the cases in the paper.
So here are the results. The figures are the approximate attack time in seconds, followed by a length in metres. By multiplying the time by 340 m/s you can see how much longer the bulk to the reflected waves travelled compared to diffracted ones (assuming the hypothesis is valid).
- Andover peak intensity after 0.03s (extra path length implied 10.2m)
- Euston square (1) 0.025s (8.5m)
- Euston Square (2) 0.007s (2.4m)
- Sauchie 0.04s (13.6m)
- Thun 0.02s (6.8m)
- Schleswig 0.015s (5.1m)
- Pursruck 0.02s (6.8m)
- Ipiranga 0.05s (17m)
- La machine 0.04s (13.6m)
- Enfield 0.01s (3.4m)
The figures are reasonably consistent with each other and all fall within the expected range, so supporting the hypothesis that sound reflections round an obstruction are likely to be the primary reason for the slow attack.
The delay in the arrival of sound from reflection would also tend to extend the overall length of a rap compared to one seen in line of sight from the recorder. Also, if the sound source is relatively remote from the recorder, as the figures above suggest, the higher frequencies will be missing as these are preferentially absorbed by the intervening air.”