The Onion Box: Description and Build Notes
The Onion Box
I call it the Onion Box because there are nested layers of puzzles in the box. In order to open it, you have to solve each of these problems in sequence, sort of like peeling the layers of an onion. It was motivated by a box I saw on the internet, but that box used electronics, and I wanted one that was purely mechanical. So my box rather diverges from the one I saw.
With the box, there is a note:
Welcome to the Onion Box! In order to open the box, you need to solve a series of puzzles, sort of like peeling away the layers of an onion.
I can’t tell you how to open the box, but I can tell you how NOT to open the box.
No excessive force is needed. No extraneous objects are needed; only that which is containing in the box. You do not need to hit or tap the box, shake it, or otherwise violently jostle it.
I am going to write this first portion from the point of view of someone trying to open the box.
The box has an obvious lid, hinged at the back. It moves slightly, but some sort of catch holds it closed at the front. Presumably the trick is find some way to open the latch.
There are three “dials” on the front face. They turn, but they don’t seem to do anything.
Each side face is divided in half, but the significance of this is not obvious. The bottom of the left side moves slightly, so apparently it is a door or drawer or something like that. But there is of course no apparently way to release it.
The first break comess when one notices something about the legs. The two left legs, i.e. the legs near the “door”, do not move and are glued in place. But the front right leg can unscrew! However, unscrewing it seems to achieve nothing.
Removing the leg doesn’t seem to have gotten us any closer to solving the box, but the second break comes when we notice that the leg has a magnet inside! When one passes the bottom of the leg around the surface of the “door”, at one point one hears a small “click”. Holding the magnet there, one can pull the drawer out of the box! It contains a stack of three gears, where the gear teeth are numbered. One would guess that these gears allow you to adjust the knobs in some way, but we don’t know what that way is yet.
The next break comes when we try the remaining fourth leg. It also unscrews. Great! That drawer now slides out. About a quarter inch… Something is keeping the drawer from opening further We will need to work harder to find out what is inside the drawer.
We can try playing with the magnet all we want, but it doesn’t do anything further. Finally, we stumble upon the secret: if the front right leg is screwed back into the box, the screw of the leg releases a catch, and the drawer slides out about an inch. Then if we remove the leg (yet again…), we can finally remove the right drawer.
It contains a diagram containing two rows of three gears each. The top three gears each have a number next to a dot. Only the bottom left gear has a number.
Eventually, one can figure out that the gears should be arranged as shown in the top row, and then they are turned to match the configuration of the bottom row. But then a new problem arises: which gear is which?
After thinking about this for a while, one will realize that in order for the puzzle to work, the indicated numbers must be unique on its gear. Two of the gears have more than one “3” on them, but one gear only has a single “3”. It must be the left gear. Likewise of the two remaining gears, one has two 1’s appearing, but the other only has a single 1. It must be the second gear. The remaining gear would be the third.
The numbers are next to a small dot, which represents a brass circle on the front. So the next step is to arrange the gears as indicated in the top row.
We then carefully rotate the gears together until the left gear matches the bottom row.
We’ve solved it!
Well…not yet. The top still doesn’t open.
A careful examination of the inside of the mechanism finds a screw head sticking down in the middle for no obvious reason. Perhaps this is the key to opening the box.
But something needs to press against the screw. The only thing that fits is the right-hand drawer, but it needs to be slid in “upside down”. As it slides in, at a certain point it will push on the screw, the latch will unlatch, and door can finally be opened!
Now at last the box has been solved. Congratulations!
Notes on the mechanism and building of it.
The key part of the mechanism is shown below:
The three “knobs” turn disks that have a notch cut into them. Then there is something I’ll call the “slider” that has a protrusion pointing at the notches. When the disks are rotated the proper way, the slider can slide such that the protrusions enter the notches, and the latch opens.
If any of the knobs are turned the wrong way, the slider will not slide.
The latch pin is not attached firmly to the slider. Otherwise, if the knobs were not correctly positioned, you could not close the lid. That is, the latch pin can slide independent of the slider so as to enable the latch to close regardless of the knob positioning.
There is a screw head sticking down from the slider. It is such that if the drawer slides in normally, it clears the screw. But if you push the drawer in upside down, it hits the screw and pushes the slider.
Just to be perverse, I cut some small semi-notches into the disks. The point of these was that if you could not figure out the correct positioning, then you could not push against the slider, slowly rotate the discs, and “pick the lock”.
This almost came back to bite me. After I had gotten everything assembled and was testing it out, I found that the right hand disc would not turn! It would turn a little, but then it hit a hard stop. I nearly had a heart attack! If I could not fix this, then the whole effort would be wasted. Obviously, the right-hand protrusion was engaged in one of the semi-notches, and it wasn’t retracting enough to allow the disk to turn. At first, I could not figure out why it was happening now (when the mechanism was totally inaccessible) and not previously when I had been testing it.
Then I figured out what I think the problem must be. If I reached in from the left and pushed on the slider screw, I could then turn the dial. For some reason, the spring wasn’t enough to push the slider all the way to the right. But why now and not before?
My best guess is that previously, the latch pin was all the way to the right and was also pushing on the slider, but that with the latch closed, it was holding the latch pin mechanism just off of the slider, and the single spring wasn’t strong enough. At first, I wasn’t sure how to fix it. Then I got an idea. I took the latch off of the lid and filed the slot slightly deeper. I only had to remove a few hundredths of an inch. But after that, it seemed to work OK. Disaster averted.
Although this just gave me a thought: I could have made the puzzler that much harder by designing the slider to rest slightly to the left of its current position, so that it keeps the gear dials from turning. Then require one to put the right-hand drawer into the left side to release the gears, and then once they are set properly, put it in the right side to actually open the box.
You can also see the spring latch on the bottom right of the main mechanism. This is the mechanism that allows you to slide the drawer in, but when you try to open the drawer, you need to put the leg screw up through the slot to push the spring out of the way, so that it will clear the pin, and you can open the drawer the rest of the way.
The magnetic latch on the gear drawer is a commercial child-proof lock, although for opening it, I put some rare-earth magnets inside the leg.
One of the main challenges was making some of the odd metal parts. The spring for the spring-latch is a piece of hose clamp that was cut up, bent, and repurposed. Likewise, I couldn’t find an appropriate latch for the lid, so I had to get an appropriate size piece of steel, bend it into an L-shape, and then grind and file the latch mechanism into the end.
I looked online for ways to make gears. I couldn’t find anything that really seemed to work for me. In the end, I wrote some custom code that plotted out a gear shape, printed it out, glued it to some plywood, and then cut it out with a scroll saw. Then there was a little filing to work out a few tight spots, and I had my gears. I used rub-on numbers to label the gears, and then a few coats of polyurethane to keep the numbers from rubbing off. As a bonus, I used my gear shape to make the combination diagram.