This experiment was started at Trinity College Dublin way back in 1944. Its purpose is to prove that tar flows, and indeed it does let go of a drop about every ten years. The thing is that nobody has ever seen that happen, bringing up the “if a tree falls in the forest” scenario. The Nature article on this event even mentions another experiment whose last drop was missed because the camera monitoring it was offline. This time around they did get some footage of the (un)momentous event which you can see below.
So here’s the challenge for clever hackers: What’s the easiest rig you can think of that won’t just continuously film the experiment but can also ensure that you get the goods on tape when a drop does fall? We see all kinds of high-speed shutter triggers — here’s one of the latest. But we don’t remember seeing an extremely slow version of the same. Let us know your idea by leaving a comment.
[Thanks Ferdinand via Endandit]
43 thoughts on “Hackit: Researchers Wait 69-years To See Tar Move”
This one is only a copy… They are watching tar in Queensland since 1927: http://smp.uq.edu.au/content/pitch-drop-experiment
Reading around on this original experiment, a story goes that the 7th drop in 1988 occurred earlier than expected due to crowds bumping its display at World Expo 88, and that the 8th drop in 2000 was missed due to the Physics IT person not verifying his scripts for capturing images. Ouch.
You could always use a circular buffer (of some fairly long length for this), and then stop recording after a photo-interrupter set right at the bottom of the funnel becomes unblocked. That way you have a single small file to look at, without having to worry about an insanely large video file.
I think this is the best way to do it.
Have it record at something small as well, like 10 fps.
Have it use compression algorithms to discard any data that hasn’t changed as well.
Should be able to have the buffer be a whole 24 hours for about 1 gb this way.
It should really be at least normal-framerate video, just because there’s no good reason for it not to be any more. Store to redundant machines with redundant disks on separate sites.
Hope this wasn’t someone’s final project, that’s a long time to wait for a degree.
They didn’t make the tube long enough to see an actual drop. The blob landed before it broke. 69 years is a long time for such a mistake.
Exactly what I thought, then some jackass thought they’d raise the funnel after-the-fact in an attempt to allow it to drop further, but ended up just clean shearing off the droplet at the funnel’s base.
Re: capture video.
A Canon camera that will run chdk running a script to start video when there is movement. And, a simple circuit that will cause something to move once the tar reaches a certain point. I figure, if the tar takes 10 years to drip, the motion sensing of the camera won’t trigger when you need it to. So, give it a bit of a heads up.
problem is, the only movement recognizable by the camera is the drop actually falling.
So the camera triggers and starts recording… a drop on the floor.
Proper movement detection isn’t between consecutive frames, but between the last movement-detected frame and the current frame.
This story actually goes MUCH much farther back.
This isn’t just “tar” like a road tar, it’s pitch, which is actually a compound used in polishing optical components. Pitch is heated and poured onto a lapping table. The lapping table is used (http://en.wikipedia.org/wiki/Lapping) in conjunction with an abrasive to polish the optical components.
As my optics professor had told me, all pitch was supplied by one person to many of the major optics companies back around the turn of the century. Then the lone supplier died, taking all the knowledge of pitch and it’s processing with him. He was one of those people who knew their process and never wrote anything down. The result was that many of the companies using the pitch bought up all the remaining pitch and began a very long and arduous journey to characterize the remaining samples. To reverse engineer the material.
The pitch drop test is one of the experiments devised to characterize the pitch. Companies have been able to make synthetic pitch, but in a lot of cases, the older material will out perform it.
There’s your useless knowledge for the day.
Its comments like this that really brings these articles to life. Thank you.
. Pitch is made from wood distillation or petroleum distillation, and has been used for thousands of years.
While provoking, I’ve heard the same story before with various items substituted for pitch, and all have proved false in the end.
It does make for an interesting, although perhaps ultimately untrue story.
It did provide credence to some things my professor taught, like how most pitch for optics is generally clear now, as opposed to the old traditional dark brown colors.
Monitor the weight of the system that is suspending the drop. When the weight changes, the drop is detaching.
That’s too late to start though.
use a series of vertical laser or led beam to shine through the glass to a sensitive photocell or light sensitive device to trigger a video camera. You can also with a more elaborate setup also using light as mentioned previously to measure the width of the
long strand of pitch getting thinner. It can be done.
This one is closest to my solution, which is to take a snapshot (video frame) of the drop, then analyze the length vs width and use that calculation to decide when to take the next snapshot. When it has gotten to a certain point, the frames would be more or less continuous.
You could simply have a video camera simply videoing the last ten minutes (or other arbitrary time period) and combine this with a standard movement detection algorithm.
Thereby you capture the moments before the drop and the drop itself.
I’d also hook it up to send me an email / text message when it happened, simples!
The problem with any trigger, weight, or motion-based solution is that it will ALWAYS start recording too late.
The best way to do it is with pre-roll: Record constantly, and just discard the older video. Only keep the video when the event happens, but then you have video from before the event happened also.
Record constantly into a circular buffer. Check frames several minutes or hours apart for movement, simple optical flow should do it here. OpenCV is almost a turnkey solution for such things.
If there has been movement, flush the buffer and keep appending for some minutes/hours/days, depending on the speed of the process.
The size of the buffer and the time distance between compared frames are the parameters to tune for slow, slower, extremely slow and this kind of processes.
Of course you could also use, as Freddy said, weight as a trigger to flush the buffer and append a bit.
Camera is always on, saving last x minutes of video into ram memory (ring buffer), because ram has much more read/write cycles than flash.
Detection of a drop is done by either a light barrier or a scale, measuing the weight of the lower part, where the drop lands.
Upon detection, save ram content to non-volatile memory.
System like this could run for ages.
Use 2 systems for redundancy.
It would really suck if after 69 years you found out a software bug, like not accounting for wrap around (69 years > 2^31 seconds), that missed the big moment.
BTW, the drop literally looks like poop.
Sometimes it feels like 10 years for me too…
1. Tell the NSA it is linked to an average person and thus highly suspect.
2. Put it in the neutral zone between the koreas, then 2 parties will watch it closely all the time – for many decades.
#2 is a great idea, except the shock of the drop moving might set off a war.
The 2nd solution really made my day!!! :D
Buy a standard ip security camera and have it record on motion. Most cameras buffer a configurable number of seconds before and after an “event”
There’s a moving clock in the frame, changing light levels across the pitch, a number of things that will cause false positives. But it will work.
now that ive seen this my bucket list is empty
ive seen paint dry and ive seen obama tell the truth..whats to do next?
definition of DROP
a (1) : the quantity of fluid that falls in one spherical mass
I don’t know perhaps the viscosity of pitch is such it can’t form into what many would call a drop and free fall as such. Unless multiple reliable witnesses to the same event reports it did. Even at that I’d think high speed camera might reveal that the drop changes into blob shape as it falls.
Gases fit the definition of a fluid as could electrons. Anyone see a drop of the air we breath or a drop of electrons from their cellphone antenna? Anyway off to watch paint dry…
As far as gases go, I imagine you could do an experiment to see gases ‘drip’. Most of us have seen someone pour cold Carbon Dioxide vapors into a tank and float something on it like a boat.
Have a rolling ~30 minute recording. It constantly deletes any video older than 30 minutes.
Then sense the event with an ultrasonic (or motion) sensor which will trigger the machine to record a little longer and then dump the recording to external storage or whatever.
How about having a CNC machine drag a piece of metal through a jug of the stuff and measure the strain on the metal part with a strain gauge. Wouldn’t that give more accurate results? This waiting aeons for it to fall would be subject to many thermal cycles which would affect the viscosity. E.g. in winter it flows more slowly than in summer.
Cant believe no one said that
There someone said it, hahaha.
Can’t believe you said that.
Ten years constant loop recording = failed RAM, surely…?
if you gave it a shedload of ram you could spread the wear so each bit would only see a few hundred writes a day
Ram wears out from writes? I thought only flash did that? As I understand it DRAM is just a bunch of capacitors and the data is “written” thousands of times per second just to refresh the charges.
Use a high quality dash-cam and wire it’s crash detection into a scale/photo-eye at the bottom. The dash-cam will constantly loop it’s recording like normal, until an increase in weight or photo-eye blockage trips the dash-cams crash detection, causing it to retain and flag the last few minutes of video.
We have a winner.
For added scientific usefulness, attach both the upper and lower portions of the experiment to separate scales that have an ongoing record of the weight. Some postgrad will get a thesis out of correlating minute changes of weight over time to seasonal (moisture condensing on the pitch) and lunar variations.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)