There a few neat mechanisms that allows only one way motion without the use of a pawl or ratchet. This is one of them:


What I'm looking for is a mechanism that can't be undone once it has gone from one position to another (without taking it apart; i.e., removing a pin) after it is rotated. This could be useful in certain applications, but my search turns up nothing like I remember.

I can think there might be a four or five-bar mechanism that has a critical point where the mechanism can't undone, but can't remember (or where to search for it)!

I do remember seeing one many years ago at one of my first engineering jobs, but can't recall the details other than "that was neat!" The mechanism I saw had I believe 3 to 4 moving parts.

There are some other mechanisms that don't use springs or latches; a search for this turned up something also nifty, but not quite what I was looking for:


Any thoughts on this?

No magnets, no springs; this was a strictly mechanical thing (nothing hidden; I could take it apart, and the mechanical lock was fascinating). The closest analogy I can think of is to those Sokoban games (and perhaps this is a way to think about it) where if you perform a certain order of operations, you lock up the puzzle.

This is one solution I came up with this morning, but I'm wondering if there is a list of them, a search term I'm missing, or other examples.

In this example, a plunger enters from the left; I'm not too sure of the geometry, but I think this is an example of how it could work. When the plunger brings the two locking elements behind it, it can't go back, because of a stop at the end of the plunger, prevents the linkage from reversing.

enter image description here

Even cooler - I believe the mechanism I saw had no linkages - just a series of Sokoban-like moves where the system went from one state to another, but couldn't be reversed without taking a part out of the 2D plane it was in.

  • $\begingroup$ Would something that involves a magnet meet your requirements? There are diagonal dovetail locking puzzles that have a bearing to block movement, held in locked position by a magnet. A sharp knock moves the ball to free the dovetail. I suspect one could modify such a puzzle to make the lock permanent, apart from disassembly. $\endgroup$
    – fred_dot_u
    Commented Mar 30, 2019 at 9:44
  • $\begingroup$ Nikola Tesla invented a one way valve for water/air that has no moving parts, but I don't think this helps you :-( The existence of one way functions is also a MAJOR open problem in computer science/mathematics. $\endgroup$ Commented Mar 30, 2019 at 13:57
  • $\begingroup$ I belive that what you ask is possible. for example you can replace the linkage with gravity in your example. Or by having a prong and let the pushing part puch down the plunger, itsel or via a 3rd part $\endgroup$
    – joojaa
    Commented Aug 27, 2019 at 19:40

3 Answers 3


You need something to introduce an asymmetry. First try with friction.

Let's start based on a (disc detainer lock](http://www.lockwiki.com/index.php/Disc_detainer) idea and modify the mechanism. We don't care about locking some outer device, we want to latch the key in permanently, so the sidebar goes away and so does the only spring, attached to it. We don't care about the outer gates, just about the keyway area.

We leave three discs, and configure them like this:

  • bottom-mostdisc (3) is fixed to the body, keyway unmodified.

  • middle disc (2) has keyway modified, its sides cut to such angles the key with zero-cut has 30 degrees of play in it. Due to it being bound, this is all the range of rotary motion the key can make, but initially turned 30 degrees back, so the key inserted engages it immediately, turning it turns the disc.

  • top disc (1) is bound to the middle disc, rotating with it. Its keyway is unmodified from normal disc detainer (fits zero-cut snugly, the key can rotate in it freely only as much as the matching cut in the key allows)

  • The key's cut matching disc 3 is 30 degrees, meaning this is the range of motion the fixed disc allows the key to rotate. Additionally, it's narrower than others, less play on the sides of the disc (spacer area) meaning pulling/pushing on the key will apply tension to that one, fixed disc first without applying force to other discs.

  • key position 2 is zero-cut, which would normally mean the disc has no play and always turns with the key. But our modified keyway will only engage with the key in the 'locking' direction. Turnng the key back won't reach it.

  • key position 1 is another 30 degree cut. That means the key in reality never engages disc 1. It can turn the range allowed by disc 1, and touch disc 3 at each extreme of its motion, but not move it.

Now after inserting the key and turning it, pulling it will engage disc 3 against the 'blank' zero-cut area past key's cut 3. If you turn the key back, so the blank tip can be pulled through the disc 3's keyway, you'll press tke key's position 2's zero-cutagainst disc 1... and since we have the tip already within keyway of disc 3 (narrower cut!) it won't budge sideways so you can't move disc 1.

Approach 2: use randomness.

Again, let's start with something like a disc detainer core, but more modified than before. The key is just a cylinder with a big number of notches cut by a lathe around its sides. The notches have rectangular profile, no sloped walls. The discs of the core are smaller than the core by the same amount as depth of the notches. Their central holes (keyway) match the diameter of the 'key'. And there are spacers in between, matching the uncut areas of the key (plus tolerance).

Upon inserting the 'key', just give it a shake or a whirl. The discs will slide around randomly, engaging the notches in the 'key'. Only aligning them all centrally would release the key, but even if one or two align that way naturally, there's still ten or so that won't. The chance random motion would align them all into a perfectly centered stack releasing the key is astronomically small.


In order for a mechanical movement to be "one way", something has to change state as the elements in the mechanism move. The expansion and contraction of a spring and the appearance or disappearance of a magnetic field constitute such a change of state. I don't think what you are looking for exists , I would be amazed if there was a device that was "one way" without having any springs or electro-magnetic phenomena. You claim you saw such a device early in your career, but did you actually inspect it to see if it had any hidden springs or magnets?

  • $\begingroup$ What about a sprag clutch? $\endgroup$ Commented Jun 23, 2020 at 14:07

A simple weighted ratchet fulfills your specification. Gravity is the "symmetry breaker" in that case. So what else do you implicitly assume that excludes a ratchet?

And: Your plunger thing does not work, I think: Unless the "connection wires" are elastic, you either cannot shove through the plunger; or it can move back - unless you assume that gravity keeps one of the two pallets in the slots - but then, you would not need the "wires".


  • $\begingroup$ Not sure why the plunger wouldn't work - as long as those are strings, and not rigid wires, and there's at least a little friction on the locking pins, it should be fine. Inserting the plunger pulls the pins into position from tension on the string, but pulling the plunger back does not push the pins back out (you can't push a rope). $\endgroup$ Commented Jan 24, 2020 at 21:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.