Visually estimating the collapse time of the WTC is difficult, as dust clouds obscure the end of the fall, but perhaps there’s another way. The Naudets 9/11 film includes a segment where the cameraman is very close to the North Tower, we see the reaction of firefighters as the collapse begins, and hear the sound of the North Tower falling apart as they turn and run. It’s a chilling moment, but does present one interesting possibility: can we begin to estimate how long the collapse took, by combining these visual and audio clues? 



Click the image above to display the video, or right-click to save it to your PC (it’s 2.4 MB). Review it and see if you agree with the following.

• The clip begins with footage spliced in of the top of the tower collapsing (Naudet didn’t film this himself). Still, the audio sounds like it may come from the Naudet recording, and based on that there’s an arguable case for saying the collapse time starts at the beginning of this video.
   
• On the other hand, this edit makes the start point questionable. Looking at the Naudets footage alone shows a fireman apparently reacting to something around frame 91, with the cameraman looking up almost immediately afterwards. If these were demolition charges then the collapse time could be later, but then the time it takes sound to reach the firefighters, plus the time to react, would push it earlier, so on balance this seems a reasonable compromise. Frame rate is 29.97 frames per second, so we’ll put the latest start time at 3 seconds.
   
•  Naudet turns and runs, while the sound of destruction gets increasingly louder. There’s a noticeable reduction at around 18 seconds, though; it’s not stopped, but there’s a clear difference. We’d hesitate to say for sure this means the collapse is basically over (listen to the audio and see what you think), but it can’t be completely ruled out. So looking for the shortest collapse time here gives us 15 seconds.
   
• The sound now tails off up to the 22 second point. By this point Naudet is sheltering behind a vehicle so it may be possible to argue that the last debris was still falling. However we could also be hearing something else by now, echoes of the original collapse, and something like 19 seconds seems more defensible as an example of the maximum end of the collapse.

One possible objection to this view could be that the firefighters were initially reacting to demolition explosions, not the sound of the collapse itself, which followed afterwards. There’s no audio evidence to these explosions in the clip, though, and testimony from Chief Joseph Pfiefer (who was running immediately behind Naudet) only records a rumble:

I was told to set up a command post up a couple of blocks. But before anything could be done, we heard a rumbling sound again and then people yelling to run. We ran maybe 50 yards the most. It was less than 10 seconds. The camera guy went behind the truck and I came up after him. He was a little faster than me, he didn’t have all that gear on, and I wound up laying across, protecting him, because he had no helmet or anything.

And then the white dust cloud came, engulfed us, immediately followed by the street now becoming black. At that point, I thought it was going to be it. We could hear things crashing and steel smashing down all around us and everything’s black and I thought we were just waiting for the squish at that point.
http://www.firehouse.com/terrorist/911/magazine/gz/pfiefer.html

Still, there’s no harm in reviewing the clip in a different way. Here’s the audio waveforms of the first 31 seconds. You can reproduce it (or do something similar with another clip) by opening an AVI file in VirtualDub, saving the audio track (File > Save WAV), then reviewing it in Audacity.



There’s arguably a rumble at the beginning of the track, however don’t read too much into the spikes at 2 to 5 seconds (Naudet is speaking). An increase in volume is noticeable, though, with a consistent roar stretching from around 5.5 to 18 seconds. This tails off up to around the 21, 22 second point, then picks up a few seconds later as the dust cloud hits.

The audio file suggests a core collapse time of 12.5 seconds, then. To that we must add the time it took from the fall initiating, to produce that high volume roar. And there may also be time to add at the end, if the collapse didn’t end at the moment that volume begins to diminish.

Of course there are issues that should make us cautious, and we’re not saying this is a method for achieving a precise collapse time, only to try and decide if it’s nearer the 8.4 or 15 second end of the usual spectrum. Our view is it strongly supports the longer time, and is quite possibly more than 15 seconds, but don’t take our word for it: download the video and check it out for yourself.