Ghosts my shiny metal ass.

I’m following a formal debate over on Rationalskepticism.org titled:

Formal debate: “Existence of ghosts & apparitions”.

I don’t think I need to elaborate on my position.

The latest, and maybe last, entry by the supernaturalist in that debate, Jeremy, concentrates on waveforms of supposed rapping by ghosts.  He, and others, propose that these waveforms are unable to be duplicated in our world, so are evidence of the existence of ghosts.

Well, being the engineer I am, I had a good look at those waveforms and made a few of my own for comparison.  Here’s the results:

First, Jerome, the supernaturalist on the above debate, posts this image from a recording taken at Andover, Hampshire, UK.  This is supposed to be knuckles on a wall.  What the wall is made of or covered with is not stated.

Andover Hampshire, knuckles on wall

Note the attack time of roughly 5 milliseconds from zero to peak amplitude.

Here, Jerome makes this entry, which is purported to be a “poltergeist” knock, also from Andover:

Andover, Hampshire - "Poltergeist" knock

The time base is roughly the same, a little more than 120 milliseconds from the start to the decay of the knock.  But this waveform shows an attack time of roughly 10 milliseconds.  Must be because of those soft ghostie hands, is my guess.

Apparently, there was another horrifying poltergeist at a place called Euston, also in the UK.  Here is a normal rap from Euston:

Euston - normal rap

Note the similar time base.  What we do not know is what was rapping on what material.  But, the attack time of this waveform is roughly 3 milliseconds.

Ohhh, and the scary anomolous rap from Euston:

Euston - ghostie rapping

Well, well, well.  For visual effect to amplify their desired outcome, the author of this image expanded the time base.  Now, the entire waveform is displayed across 60 milliseconds.  To the untrained eye, clearly a much, much slower attack time.  To anyone who can read a fucking oscilloscope, the attack time is roughly 11 milliseconds.  Soft ghostie knuckles in Euston also, it seems.

But Jerome took some of his own recordings and here are their waveforms:

Jerome - knocking on wood

Jerome - knocking on a table five feet from the microphone

Jerome - sound of knocking on the other side of a wooden door

On all three of Jerome’s recordings, the time base is so slow, we can’t see the attack time.  It looks immediate.  His time base is ten times or more slower than the previous waveforms.  We can learn nothing from these images.

Now, to my simple experiments.  The only microphone I have for my Mac is attached to a USB boom mic/headset.  How I set it up was to pin it down to the top of my hardwood (beech) computer table with a closed-cell foam block.  I wanted good coupling to the table top and acoustical isolation from ambient noise.  I recorded these waveforms using Audacity for the Mac.  Wonderful program.  My sampling rate was double Jeremy’s, at 96000 Hz.

I adjusted all the waveforms to the same time base and amplitude settings.  There are two knocks in each recording.  The first is made on an upturned 12 ounce Corelle glass bowl on the table, and the other is made directly to the table top itself.  I used four materials:  a 345 gram steel bar, a baseball, a plastic bocci ball, and the middle knuckle of my right middle finger.

Here are the waveforms:

Steel bar on 28mm thick beechwood table

The steel bar gives a very fast attack time, as you would expect.  There will be little energy of motion during the impact lost to elasticity.  That energy of motion will be turned into sound waves in the wood table very quickly, as we see.  An attack time of one to two milliseconds.  Quite marked.

Baseball on 28mm beechwood table

Ahh, but a baseball isn’t nearly so hard.  It won’t be able to get that wood vibrating as rapidly as a steel bar, and these waveforms demonstrate it.  Note also that a baseball makes a lower pitched “thunk” on the table, which is displayed by the lower frequency of the waveform (longer time between cycles).  But. look at the slower attack time.  Looks like about five milliseconds.

Bocci ball on 28mm beechwood table

With a plastic bocci ball, the frequency is even lower.  That attack time is roughly five milliseconds again.

Right middle finger 2nd knuckle on 28mm beechwood table

Well, look at that!  Does this not look just about like the Euston ghostie?  Here it is again:

Euston ghostie

You may notice that the duration of the rap event is roughly the same for my recording and the Euston recording.  The attack time is the same.  The waveform envelope is the same.

So, waddya think?  Somebody’s knocking on wood with their hands.  If I can produce this in 10 minutes, this is not a hard thing to do.  I don’t doubt I can increase that attack time by using softer wood, and trying different hammer materials.  Something with a progressively harder surface, perhaps.  Like I said in my title:  ghosts, my shiny metal ass.