Mechanical Puzzles

Quick answer: A mechanical puzzle is a physical object with a clear objective and no obvious steps to reach it. There are seven main types, a puzzle can belong to more than one, and knowing which type you enjoy saves you more money than any other thing you can learn about them.

Best for: anyone about to buy their first one, and anyone who has bought several and noticed they only really liked some of them.

Most people meet mechanical puzzles one at a time. Someone gives you a wooden cross that falls apart in your hands, or you pick up a metal tangle in a gift shop, and the whole category gets judged on that one object.

That is an expensive way to learn. The seven types below ask genuinely different things of you, and liking one tells you almost nothing about whether you will like another. This guide exists so you can find your type first and buy second.

What Is a Mechanical Puzzle?

The definition worth using is the one from our sister site: a physical object with a certain objective, without obvious right steps to achieve it, where the solver has to use wits and logic to discover the steps.

Three things in that definition do real work. It is physical, so it is not a riddle or a crossword. It has one objective, usually obvious, often printed on the box. And the steps are hidden, which is the entire point: you are never confused about what you are trying to do, only about how.

That last part is why they are satisfying and why they are frustrating in a way that feels fair. Nobody is withholding information from you. The object is sitting in your hand, complete, and it will show you the answer the moment you think of it.

How Is a Mechanical Puzzle Different From a Jigsaw?

A jigsaw is technically a put-together puzzle, so the honest answer is that a jigsaw is one, just an unusual one.

The difference is in method. A jigsaw rewards sorting, patience and visual matching, and the piece in your hand either fits or it does not. There is no insight, just progress. A mechanical puzzle usually has few pieces, often fewer than ten, and no amount of sorting helps. You will sit with five pieces for an hour and then see it all at once.

One is a pleasant process with a guaranteed end. The other is a problem with a solution you might not find.

Where Did Mechanical Puzzles Come From?

Older than you would guess, and from two directions at once: mathematics and security.

The oldest known is the Ostomachion, also called the Stomachion, attributed to Archimedes in the third century BC and preserved in the Archimedes Palimpsest. It is a square cut into fourteen pieces, and the question Archimedes seems to have been asking is how many ways those pieces can be rearranged back into the square. It is a dissection puzzle, which makes the oldest mechanical puzzle we know of an assembly puzzle.

Running alongside that is a less romantic lineage. Romano-Celtic puzzle padlocks from the second century BC were ring-sized locks that opened only through a non-obvious sequence, and they were not toys. They were security. The idea that a mechanism can be its own password is older than the idea that it can be entertainment.

In China the Lu Ban lock comes from the same impulse as a building technique rather than a game. Lu Ban was a structural engineer during the Spring and Autumn period, and the interlocking mortise-and-tenon joinery that holds those puzzles together is the same joinery that held up buildings. An interlocking puzzle is, quite literally, carpentry with the usefulness removed.

The written record for the European burr starts in 1698, in a John Sturt frontispiece engraved for Chambers's Cyclopaedia, which contains the earliest known picture of one. The word burr itself does not appear in print until Edwin Wyatt's Puzzles in Wood in 1928, and it is used there as if readers already knew it, which tells you it had been spoken for some time before anyone wrote it down.

The modern field begins in 1986 with Puzzles Old and New by Jerry Slocum and Jack Botermans, the first serious book devoted to mechanical puzzles. Slocum's collection eventually passed 30,000 items and now sits at the Lilly Library at Indiana University, catalogued under his own classification system. More on that below, because it is the reason the categories in this guide look the way they do.

What Are the Main Types of Mechanical Puzzles?

Seven, and every puzzle gets at least one of them. Plenty get two.

1. Assembly (put-together) - build the shape
2. Take-apart - open it
3. Interlocking - pieces lock each other in
4. Combination (sequential movement) - right moves, right order
5. Maze - find the path
6. Impossible objects - work out how it was made
7. Dexterity - your hands, not your head

This guide is about what they are. If you already know and just want recommendations, Greg's list of the best mechanical puzzles is the companion piece, organised by these same categories.

What follows is what each one actually asks of you, who tends to love it, and what we stock. We do not stock all seven, and where we do not, this says so and points you somewhere useful instead.

What Is an Assembly (Put-Together) Puzzle?

You get loose pieces and a target shape. Build it. Nothing interlocks, so once solved it comes apart easily, and that asymmetry is the form: taking apart is trivial, putting together is everything.

The sub-types are worth knowing because they feel different:

• Classic dissection - a shape cut into pieces that go back one way. Tangrams are the famous version.
• Soma cubes and polycubes - pieces built from joined cubes, assembled into a larger cube or a named shape.
• 2D packing and framed - flat pieces into a tray, usually with an opening that restricts how they enter.
• 3D packing - the same idea with depth, which is much harder because you cannot see the whole problem at once.

This is what we stock most. Cube Alchemy is the polycube entry point, seven pieces into a 3x3x3 cube with a booklet of 180 further shapes, and the form is explained in the Soma Cube guide. Three-Piece Block and Multi-Grain Burr are Stewart Coffin dissections. Shinobi is three flat pieces and a symmetry condition.

Packing is the biggest shelf we have, covered properly in packing puzzles explained, and the distinction between packing and pure assembly is in assembly puzzles explained, and the shelf itself is 3D assembly puzzles. There is a deeper treatment of the whole category on explorepuzzles.

You will like these if you think creatively, find beauty in simplicity, and want to practise spatial thinking.

What Is a Take-Apart Puzzle?

It is closed and you have to open it. No instructions, no tools, and usually no visible mechanism.

These look simple and are built to deceive. The moves tend to be unusual: spinning, tapping, holding at an angle, pressing something that does not look like it presses. And they are two puzzles for the price of one, because once it is open you have to close it, and an hour later you will have forgotten how.

Sub-types include puzzle boxes (a hidden compartment and a sequence to reach it), locks (which look like padlocks and do not behave like them), and disentanglement (two pieces tangled together, usually metal, that have exactly one way to separate).

We stock the boxes: Secret Box has a compartment big enough for a small gift, and Impossible Dovetail Box is a secret-mechanism box that looks permanently sealed. Our one disentanglement puzzle is Hedgehog in a Cage, which is a genuine classic, and there is a full walkthrough.

We do not stock metal disentanglement puzzles of the Hanayama kind. If that is what you are after, the take-apart guide on explorepuzzles covers them in detail and will serve you better than we can.

You will like these if you think outside the box, enjoy unusual mechanisms, or collect.

How Do You Tell Assembly and Take-Apart Apart?

This confuses almost everyone, because you assemble both.

The test is one question: what is the challenge? If the difficulty is in building it, it is an assembly puzzle. If the difficulty is in opening it, it is a take-apart puzzle. A Soma cube comes apart by lifting pieces off, so the challenge is assembly. A puzzle box is trivial to close once you know the trick, so the challenge is opening it.

When the answer is genuinely both, you are usually holding an interlocking puzzle.

What Is an Interlocking Puzzle?

Pieces that hold each other in place through shape alone, with no glue and no hardware. Nothing moves at first. Then one piece shifts, then another, and the whole thing loosens until the first piece comes free.

Taking it apart is the easy half. Reassembly is where these earn their reputation, and some of them are, in plain terms, beyond most people without software help.

Three minutes on the six-piece burr, which is the single most useful thing to watch if you have never seen one solved. Note how little of it is dexterity and how much is deciding the order before anything moves.

The sub-types:

• Burrs - notched sticks passing through one another. The six-piece cross is the archetype, explained in burr puzzles explained and solved step by step in this guide.
• Caged burrs - a burr built inside a frame, which removes most of the room you were relying on.
• Coordinated motion - several pieces that must all move at once, in different directions, or nothing happens at all.
• Turning interlocking cubes - at least one piece rotates rather than slides. This is our speciality and there are over forty of them, covered in the TIC guide.

Start with 6-Piece Cross or, for rotation, EcstaTIC, which is two pieces and one turn. The broader form is in the complete guide to interlocking puzzles, difficulty notation is explained in level notation, and explorepuzzles has a deeper dive including BurrTools, the program serious solvers use when they are genuinely stuck.

You will like these if you have high spatial imagination, deduction and patience.

What Is a Combination (Sequential Movement) Puzzle?

The moves are obvious. The order is not.

Nothing is hidden in a combination puzzle: you can see every move available. The difficulty is that only one sequence makes progress, and a wrong move can undo a great deal of work. Solvers often learn algorithms, fixed sequences that make one predictable change regardless of the current state.

Sub-types include classic twisty puzzles (the Rubik's Cube and its relatives), modern twisty puzzles in shapes that barely resemble cubes, n-ary puzzles built on counting systems where the solution can run to thousands of moves, and sliding puzzles of the Rush Hour kind.

We do not stock these. No twisty puzzles, no sliding puzzles. If this is your type, the combination guide on explorepuzzles is the right place to start and we would rather send you there than sell you something adjacent.

Worth knowing: our puzzles come apart completely, which is the opposite of a twisty puzzle, where the pieces are captive on an internal mechanism and never separate. People who love one often dislike the other.

You will like these if you are organised, logical, and enjoy very long solutions.

What Is a Maze Puzzle?

Find the path. Usually you cannot see it.

The mechanics are consistent: gravity does much of the work, a ball bearing or a pin navigates, and harder mazes include traps, dead ends that are hard to back out of. In an open maze you can see the route. In a hidden maze you cannot, and have to build the map by feel, which is a strange and genuinely absorbing experience.

We do not stock true maze puzzles. Alpacka 1 is sometimes described as behaving like a maze, because pieces have to be routed past obstructions, but it is a packing puzzle and it would be wrong to sell it as anything else. The maze guide on explorepuzzles covers the real thing.

You will like these if you can concentrate for long stretches and enjoy working blind.

What Is an Impossible Object?

Not a puzzle you solve. A puzzle you explain.

An impossible object is made in a way that appears to break physics: an arrow through a bottleneck narrower than the arrowhead, a dovetail joint that could not have been slid together, a bolt inside a sealed glass. Some can be taken apart and some genuinely cannot. The challenge is working out how it was made.

These are the ones that live on desks and start conversations, which is most of their appeal.

We have several. Impossible Dovetail Box is the classic joint that should not exist, explained in the dovetail article without spoiling it. Newton's Gravity Defying Puzzle balances a rocket on its base and asks you to remove it without touching the base. Infinity Chocolate is a geometric dissection illusion where rearranging the pieces appears to create a square from nothing.

There is more on the category on explorepuzzles. Objects of this kind end up in the collectible puzzles shelf, because they get kept rather than solved and put away.

You will like these if you enjoy wondering, like art, or work somewhere with a desk and visitors.

What Is a Dexterity Puzzle?

Your hands, not your head.

Knowing the solution does not solve a dexterity puzzle. You have to execute it, usually repeatedly, usually badly at first. Gravity and ball bearings do most of the work, and the skill involved is real but physical: hand-eye coordination, a steady hand, and a calm mind, because tension makes you worse at these in a way it does not with the other six types.

Sub-types are balance, trick motion, obstacle courses and the bottle puzzles that sit on the border with impossible objects.

There is a fuller account in dexterity puzzles explained, including why the people who bounce off them hardest are often the strongest puzzlers.

Ours is thin here deliberately, because dexterity is not what our catalogue is for. Tippe Top is the honest example: spin it hard enough and it inverts to spin on its stem, which is a physics demonstration as much as a puzzle. The dexterity guide on explorepuzzles covers the category properly.

You will like these if you would rather do than sit and think.

Where Do Sequential Discovery Puzzles Fit?

They do not fit, which is why people argue about them.

A sequential discovery puzzle gives you an object and, as you solve it, tools: a pin, a rod, a key that falls out at step three and opens something at step nine. It borrows from take-apart, combination and sometimes maze, and the defining feature is that the puzzle supplies its own equipment.

They are widely considered the most satisfying form there is, and also the most expensive, because the mechanisms are complex and the runs are small. We have a full explanation of why solvers love them.

Which Puzzles Should Everyone Know?

A short canon. Not the best puzzles, necessarily, but the ones that changed what came after them. Knowing these gives you the whole field in ten objects.

• The Ostomachion, third century BC. Fourteen pieces, one square, and the oldest mechanical puzzle we have. It makes dissection the founding form.
• Chinese rings, also called the baguenaudier. Rings on a loop that come off only in a strict order, and the order is counting in binary. The first puzzle whose solution is an algorithm rather than an insight.
• The Lu Ban lock. Interlocking joinery borrowed from architecture, and the ancestor of every burr.
• The Tower of Hanoi, 1883. Three pegs, a stack of discs, and a solution of exactly 2 to the power n minus 1 moves. The puzzle that taught recursion to generations of programmers, and we have a full guide to it.
• The 15 puzzle, 1880, and the best story in the field. It was invented by Noyes Palmer Chapman, a postmaster in Canastota, New York, who applied for a patent in March 1880. Sam Loyd then spent twenty years from 1891 until his death falsely claiming he had invented it, and largely succeeded. Half the internet still credits Loyd. It is also the puzzle that introduced the idea of an unsolvable starting position: exactly half of all arrangements cannot be solved at all, which is why Loyd could safely offer a prize for one of them.
• The tangram. Seven pieces, thousands of target silhouettes, and proof that a puzzle does not need one answer to be good.
• The six-piece burr. The most-produced mechanical puzzle in history and the one most people have actually held. Covered in burr puzzles explained.
• The Soma Cube, 1933. Piet Hein reportedly conceived it during a lecture on quantum mechanics. Seven pieces, 240 distinct solutions, and still hard. Full guide here.
• The Rubik's Cube, 1974. Not one of ours, and the most successful puzzle ever made. It created the idea that a puzzle could have a competitive solving community.
• The Hanayama Cast series. Mass-produced metal puzzles rated one to six, which did more than anything else to give ordinary buyers a difficulty scale they could trust.

How Does This Compare to the Official Classification?

There is a standard, and it is worth knowing because serious collectors and auction listings use it.

Jerry Slocum's system has ten categories, and its great insight is the one this guide is built on: he classified puzzles by what the solver has to do, not by material, shape, age or country. Solutions are the common denominator. That is why asking "what is the challenge?" sorts almost any puzzle correctly.

His ten are put-together, take-apart, interlocking, disentanglement, sequential movement, dexterity, puzzle vessels, vanish puzzles, folding puzzles and impossible objects.

The seven in this guide map onto those directly, with three deliberate differences:

• Disentanglement is its own category for Slocum. We treat it as a sub-type of take-apart, because the solving experience is the same: a closed thing you have to open.
• Mazes are not a Slocum top-level category at all. We separate them because they genuinely feel unlike anything else, and because hidden mazes in particular have no equivalent elsewhere.
• Puzzle vessels, vanish puzzles and folding puzzles we leave out. Vessels are trick drinking cups, vanish puzzles are printed optical tricks where a figure disappears when parts are rearranged, and folding puzzles are flat things folded to a target. They are real categories with real history. They are simply nowhere near the kind of object this shop sells, and pretending otherwise would pad the list without helping you.

If you want the full academic version, Slocum's is the one to read. If you want to work out what you will enjoy, seven is enough.

Does a Puzzle With More Pieces Mean It Is Harder?

No, and believing it will lead you to buy the wrong things.

Two of the hardest puzzles we sell have two pieces. EcstaTIC is two interlocking pieces and one rotation, and people who have solved far bigger burrs get stuck on it, because the difficulty is seeing the move rather than tracking many parts.

Piece count measures bookkeeping. Difficulty lives in the move sequence, and interlocking puzzles have a proper measure for it called level notation, which counts the moves needed to free the first piece. A puzzle described as level 28 needs twenty-eight moves before anything comes off. That is explained in what makes a puzzle hard.

What Is the Mathematics Behind Them?

More than you would expect, and in several cases the puzzle came first and the mathematics followed.

Counting the answers is its own problem. The Soma Cube has 240 distinct solutions, and that number was not obtained by a computer. John Conway and Michael Guy enumerated all of them by hand in 1961. The useful lesson is counter-intuitive: 240 solutions does not make it easy, because no arrangement feels wrong until the last two pieces will not go.

Packing puzzles are genuinely hard in the formal sense. The general problem of fitting a set of shapes into a bounded space belongs to the packing family, which is NP-hard. There is no known efficient general method, and the practical consequence is that your brute-force instinct is not a character flaw, it is the only method anyone has.

Sliding-block puzzles are harder still. Robert Hearn and Erik Demaine proved that sliding-block puzzles are PSPACE-complete, and remain PSPACE-hard even when every piece is a simple domino and all you want is to move one particular block. That is a stronger statement than NP-hard, and it is the formal reason these puzzles can need enormously long solutions for a tiny number of pieces.

N-ary puzzles are literally counting. A puzzle described as ternary is working in base 3, and solving it is incrementing a number one unit at a time. This is why n-ary solutions run to thousands of moves and why the move count grows exponentially with the number of elements. Chinese rings is the binary case, which makes it the oldest of them.

Software does exhaustive search. BurrTools does not reason about a puzzle the way you do. It enumerates assemblies and tests which ones can actually be taken apart, which is why it can report that a design has, for instance, thousands of valid assemblies of which only a handful disassemble. That ratio is how designers find the hardest puzzles.

How Hard Do Mechanical Puzzles Actually Get?

Much harder than the shelf in front of you suggests, and the numbers are worth seeing once so the rest of the scale makes sense.

A typical good puzzle needs somewhere between three and fifteen moves to free its first piece. That is enough to take most people an evening.

Then there is the other end. Extreme Torture, a board burr by Frans de Vreugd, needs 28 moves to release the first board and 70 in total. Its design came out of a search over 8,008 piece sets and 1,364,932 possible assemblies, of which only 2,394 could be taken apart at all. That is 0.18 per cent.

Goh Pit Khiam's Ternary Burr goes further: twenty-two pieces, more than ninety notches and grooves, and seventy-five moves before the first piece comes free. It works by folding n-ary counting logic into an interlocking burr, which is two hard ideas in one object.

If that end of the scale appeals, it is gathered in extremely difficult puzzles, and the step below it is hard puzzles.

And difficulty is not the same as quality. A puzzle can be brutal and dull, or trivial and brilliant. The designers most admired in this field are usually not the ones producing the highest numbers, which is worth remembering while reading any spec.

What Do All the Words Mean?

The vocabulary is small but almost none of it is guessable, and it is the main thing that makes puzzle descriptions opaque to newcomers.

• Level - the number of moves needed before the first piece comes free. A level 28 puzzle needs twenty-eight separate movements before anything separates. Written as a series, like 28-22-9-8-3, it gives the moves for each successive piece. Explained fully in level notation.
• Burr - an interlocking puzzle of notched sticks. Not every puzzle with burr in its name is one.
• Key piece - in a burr, the piece with the fewest notches, often none. It is the last in and the first out, and it locks the others.
• Dissection - a shape cut into pieces that reassemble into it, or into something else.
• Polycube - a piece made of unit cubes joined face to face. The seven Soma pieces are polycubes.
• Apparent cube - a puzzle that reads as a solid cube from outside while being mostly hollow within. The goal is to cover every opening, not to fill the volume.
• Coordinated motion - pieces that must all move together, in different directions, or nothing moves at all.
• N-ary - a puzzle built on a counting system, where the solution is effectively counting in base 2 or 3 and can run to thousands of moves.
• Void - deliberate empty space inside a solved puzzle, which is usually what makes it hard.
• Restricted opening - a container whose entrance is partly blocked, so pieces cannot simply be lowered in.
• Tolerance - the slack between pieces. Too tight and a correct solution will not go together; too loose and the puzzle rattles. Good making is largely about this.
• BurrTools - free software by Andreas Röver that solves and analyses interlocking puzzles. Widely used, and not considered cheating by most people once you have genuinely given up.
• IPP - the International Puzzle Party, the annual gathering, and home of the Nob Yoshigahara Puzzle Design Competition.
• Exchange puzzle - a puzzle made in a small run specifically to be swapped at IPP. These become collectible immediately because the run size is the attendance.

Do Mechanical Puzzles Actually Do Anything for Your Brain?

This gets oversold constantly, so here is the careful version.

Spatial ability is real, measurable and predictive. Jonathan Wai, David Lubinski and Camilla Benbow tracked a stratified sample of 400,000 American students over eleven years and found that spatial skill measured at school predicted entry into science, technology, engineering and mathematics careers a decade later, even after controlling for mathematical and verbal ability. It was carrying its own weight rather than standing in for general intelligence.

It is trainable, and the gains last. David Uttal and colleagues meta-analysed 217 training studies and found an average improvement of 0.47 standard deviations, stable across delays of up to four months and transferring to spatial tasks that had never been trained.

And no study has tested these. Nobody has run a trial on mechanical puzzles specifically. Anyone telling you a packing puzzle is clinically proven to do anything is selling something.

The honest position is narrower and still worth having: these objects exercise a faculty that is measurable, trainable, durable and predictive of technical careers. Whether an evening with one moves that needle the way a structured course does is untested. The fuller treatment, with the research properly cited, is in 3D spatial reasoning puzzles, and the matching shelf is spatial puzzles.

Who Actually Makes Mechanical Puzzles?

Two different jobs, usually done by two different people, and understanding the split explains most of what confuses newcomers about price and availability.

The designer invents the puzzle. That is increasingly a computational job: a design is modelled, searched and verified in software before anything physical exists. Almost nobody does it full time. Alexander Magyarics is one of the very few who does, and he got there by producing designs at an extraordinary rate while holding down an unrelated job.

The craftsman makes it. Workshops like Pelikan in the Czech Republic, Cubic Dissection and Wood Wonders take designs from people they rate and cut them in hardwood, in runs that are often only a few dozen. That is why a puzzle can be famous and unbuyable at the same time: the design is public, the run sold out in an afternoon three years ago.

The two meet at the Nob Yoshigahara Puzzle Design Competition, held at IPP each year. Hundreds of designs are entered and most receive nothing. A jury awards most prizes, with one exception worth knowing about: the Puzzlers' Award is voted on by the attendees, which is to say by several hundred of the most experienced solvers alive after days of handling everything on the table. Designers we stock who have taken it include Yasuhiro Hashimoto and Goh Pit Khiam.

We sit in a third position: we 3D print designs from the originals, with the designer's involvement, which is why things stay in stock and affordable while hardwood editions do not. What that trade costs you and what it buys you is set out in 3D-printed vs wooden puzzles. Our designer profiles are collected across the Coffin, Yamamoto, Latussek and Crowell articles among others, and each has its own shelf: Coffin, Yamamoto, Latussek, Crowell, Pauwels and Goh Pit Khiam.

Who Writes About Mechanical Puzzles?

This field has almost no mainstream press and an unusually good amateur one. A handful of people have been reviewing puzzles carefully for fifteen years, and they are the reason anyone can find out whether a puzzle is worth owning before buying it.

The ones worth following, all still worth reading even where the archive has gone quiet:

• PuzzleMad, Kevin Sadler. The most prolific and the most useful on difficulty, because he is honest about what defeated him.
• Allard's Puzzling Times, Allard Walker. Strong on new releases and on the feel of a solve.
• Notes from Puzzle Palace and Puzzle Ramblings, both active.
• Jerry's Puzzle Collection, Brian's Damn Puzzle Blog, Gabriel Fernandes, Neil's Puzzle Building Blog and Puzzle Paradox. Dormant or slowed, with large archives that have not expired. A 2015 review of a puzzle still in production is still a review.

We cite these people by name throughout this blog rather than paraphrasing them anonymously, because in a field this small a named reviewer is the closest thing to a standard.

Who Makes Videos About Mechanical Puzzles?

The written blogs are where the detail lives. Video is where most people now meet the hobby at all, and the audiences are of a different order entirely: the big puzzle channels reach more people in a single upload than the written community does in a year.

Chris Ramsay is the one most people have seen. A German-born Canadian who came to puzzles from magic and cardistry, he has around seven million subscribers and solves on camera from the solver's point of view without the padding that makes most puzzle video hard to watch. More than any single person he is the reason mechanical puzzles reached a mainstream audience at all, and a feature from him can empty a workshop's stock in a day.

Puzzle Guy, at roughly two and a half million subscribers, covers the widest range of anyone, and is the channel to follow if you want to see a lot of different puzzles rather than a few in depth. He sells puzzles as well as filming them, which is worth knowing when you read a recommendation, and is exactly the thing you should hold against us too.

Cuby is the largest cubing channel anywhere, with over three million subscribers, and posts in Spanish. Mostly twisty puzzles and logic games rather than the wooden interlocking end, so he is the right follow if the Rubik's family is what pulled you in.

Mr. Puzzle, around a million strong, sits at the collector end, reviewing and solving wooden mechanical puzzles and dexterity games. Worth knowing that the name collides: Brian Young's Australian workshop has traded as Mr Puzzle for decades and is a maker, not a channel.

And Greg Puzzles, which is far smaller than any of the above and a much older part of this particular story. Greg was making puzzle videos for years before this shop existed, showing and solving whatever he had got hold of, and the shop grew out of that rather than the other way round. Most of the solution guides on this blog exist because there was already a video to write them from.

How Do You Solve One Without Being Told the Answer?

There is no single method, and anyone selling you one is overstating it. Puzzles differ too much, and a lot of solving is just time spent. But there are things worth trying before you resort to turning it over and over in your hands, and almost nobody is ever taught them.

Look before you touch. Most people start moving pieces within two seconds. Spend a minute first: count the pieces, find the ones that are different from the others, and look for the asymmetry. Puzzles are almost always more symmetric than their solution, and the odd piece is usually the whole story.

Work out the goal state, not the first move. Beginners search forward from where they are. Experienced solvers picture the finished object and ask what the last move into it must have been, then the one before. Working backwards from the end collapses the search enormously.

Put it on a table. Holding a partial assembly in mid-air means fighting every loose piece at once. Flat on a surface, gravity holds your progress for you.

Never force anything. This is the only real rule. Mechanical puzzles are made to tolerances where a correct move needs no strength at all. If it will not go, it is wrong, and pushing harder both fails and risks damage. The single most common way people break a puzzle is believing they have the right answer.

Put it down. Walking away for an hour is a genuine technique rather than an admission of defeat. Fresh eyes dissolve fixations that no amount of continued fiddling will.

Photograph it before you take it apart. Obvious, universally ignored, and the difference between an evening of pleasure and an evening of regret.

Which Individual Puzzles Are Worth Knowing?

The categories above describe shapes of problem. These are the specific objects, and most of them are public domain classics anybody can buy anywhere.

Dissections and assembly. The tangram, far younger than usually claimed; the Stomachion, the oldest known mechanical puzzle and the one Archimedes wrote about; pentominoes, where exactly twelve shapes exist; the T puzzle, four pieces that defeat nearly everyone; and the Soma Cube.

Interlocking. The Lu Ban lock, which is carpentry with the building removed and the ancestor of every burr.

Sequence and sliding. The Chinese rings, which is counting in binary; the 15 puzzle, where half of all starting positions are impossible; sliding block puzzles from Huarong Dao to Rush Hour; peg solitaire, where one of the two standard boards cannot be solved at all; and the Tower of Hanoi.

Take-apart and boxes. Japanese puzzle boxes, where the decoration hides the seams; blacksmith and nail puzzles; Hanayama Cast puzzles, properly designed and sold for pocket money; and the cryptex, invented by a novelist in 2003.

Mazes and impossible objects. Revomaze, the maze you are never allowed to see; and impossible bottles, along with the man who took his methods to the grave.

Twisty. The Rubik's Cube, with forty-three quintillion positions all within twenty moves of solved, and the one category here that never comes apart.

And magic squares, which are not quite mechanical puzzles and belong to the same tradition of asking how many answers there are rather than what the answer is.

Which Type Should a Beginner Start With?

Start with assembly or a simple interlocking puzzle. Both are cheap, neither needs any prior knowledge, and both tell you quickly whether you enjoy the core experience of sitting with an object that will not cooperate.

Concretely: 2-Piece Pyramid to find out in ten minutes whether this is for you, then 6-Piece Cross for the classic experience, then Cube Alchemy if you want one that keeps going.

Avoid starting with anything rated Expert, anything with a four-hour solve time, and anything whose description uses the word brutal. There is no prize for starting at the top and a real risk of deciding the whole hobby is not for you based on one bad purchase. If you would rather be matched by taste than by rank, which mechanical puzzle to buy first does exactly that, and the easy puzzles shelf is deliberately small. If budget is the constraint rather than difficulty, puzzles under $25 maintains itself, and pocket puzzles covers anything that has to travel.

Where Should You Buy, and What Should You Watch For?

Three routes, and they suit different people.

Craft workshops such as Pelikan, Cubic Dissection and Wood Wonders cut designs in hardwood in runs of a few dozen. The results are beautiful and frequently sold out within hours of release. If you want the object as an object, this is where to look, and you will need to watch for drops.

Mass producers such as Hanayama make metal puzzles in quantity with a reliable difficulty scale. Easy to buy, always available, and a sensible way to find out whether you like disentanglement before spending more.

3D printed editions, which is what we do. Designs are printed from the originals with the designer involved, which keeps them in stock and affordable. The honest trade is set out in 3D-printed vs wooden puzzles: you are buying the design and the problem, not the woodwork.

Two things to watch for. Unlicensed copies are common, and a seller who cannot tell you who designed a puzzle usually has a reason. Designers in this field are mostly not paid by volume, so buying from someone who names and works with them is the difference between supporting the next design and not. And beware difficulty ratings across shops, which are not comparable: a Hanayama 6 and a wooden puzzle marked expert are not measuring the same thing. For interlocking puzzles, level notation is the only scale that travels, and it is explained here.

A broader comparison is in best online puzzle shops, which names competitors because a guide that only recommends itself is not a guide.

What Do Beginners Usually Get Wrong?

Four things, in order of how much they cost.

Buying by difficulty. Starting at Expert because easy sounds unsatisfying is the most expensive mistake available. Difficulty ratings here assume an adult who has solved several, and a four-hour puzzle met first is usually a four-hour puzzle unsolved.

Buying by piece count. Covered above and worth repeating, because the instinct is strong and wrong.

Taking it apart before looking. A puzzle that arrives assembled arrives with the answer visible. Study it first. You can only do this once.

Assuming one type represents all of them. Someone who disliked a metal tangle in a Christmas cracker has not met mechanical puzzles. They have met one sub-type of one category out of seven, and it is the one most people like least.

How Should You Look After Them?

Very little, but the little matters.

Wooden puzzles move with humidity. A puzzle cut in a dry workshop can swell in a damp room until a correct solution genuinely will not go together, which is maddening and not your fault. Keep them away from radiators, direct sun and bathrooms.

Ours are 3D printed, which sidesteps humidity entirely: no swelling, no seasonal tolerance change, no cracking. The trade is a different feel in the hand, not a different difficulty.

Printed plastic has one weakness of its own, and it is worth knowing because it is the only way people damage ours. Do not leave one in a hot car. A closed car in summer gets far hotter than the air outside, and held there long enough the plastic can soften and deform. A puzzle that comes back slightly out of true will never solve properly again, because these are cut to tolerances where a fraction of a millimetre decides whether a correct move goes. Glovebox in August is the one place not to keep it.

Store interlocking puzzles assembled where you can. It protects the pieces, it looks better on a shelf, and it means the puzzle is ready for the next person, which is most of the point of owning them.

What Makes a Mechanical Puzzle Good?

Three things, and difficulty is not one of them.

A clear objective. You should never be unsure what you are trying to do. Confusion about the goal is a design failure, not a challenge.

Fairness. Everything you need is in the object. No hidden tolerances, no move that only works if you squeeze hard, nothing that depends on luck. The best puzzles can be solved by thinking, and when you see the answer it feels inevitable rather than arbitrary.

An ending worth reaching. The moment it gives way should feel like something. Designers talk about this openly, and it is why a two-piece puzzle can be better than a twenty-piece one.

Material matters less than people expect, though it does change the feel, and the trade-offs are set out in 3D-printed vs wooden puzzles.

For specific recommendations rather than definitions, the best mechanical puzzles works through the same categories naming individual puzzles. The whole catalogue is in mechanical puzzles. The three biggest shelves are packing, interlocking and turning interlocking cubes. Narrower shelves worth knowing: cube puzzles, geometry puzzles and minimalist puzzles.

Frequently Asked Questions

What is a mechanical puzzle?

A physical object with a clear objective and no obvious steps to reach it, solved by working out the hidden sequence. Unlike a riddle it is a real object you hold, and unlike a jigsaw it usually has very few pieces and cannot be brute-forced.

What are the types of mechanical puzzles?

Seven: assembly (put-together), take-apart, interlocking, combination (sequential movement), maze, impossible objects and dexterity. A puzzle can belong to more than one, and sequential discovery puzzles deliberately combine several.

What are the best mechanical puzzles for adults?

It depends entirely on which of the seven types suits you, which is the point of this guide. For a first purchase, an assembly puzzle or a simple interlocking puzzle gives the clearest signal. Everything we sell is designed for adult solvers.

Are mechanical puzzles good for kids?

Some are, but our range is built for adults and the difficulty ratings assume an adult solver. An easy-rated puzzle here will frustrate most children. Dexterity and simple assembly puzzles travel best across ages.

What is the difference between an assembly puzzle and a take-apart puzzle?

Ask what the challenge is. If the hard part is building it, it is assembly. If the hard part is opening it, it is take-apart. If both are hard, it is almost certainly an interlocking puzzle.

Do more pieces make a puzzle harder?

No. Two of the hardest puzzles we sell have two pieces. Difficulty lives in the move sequence, which interlocking puzzles measure with level notation rather than piece count.

Is a brain teaser the same as a mechanical puzzle?

Brain teaser is the broader and looser word and covers riddles and logic problems with no physical object at all. Every mechanical puzzle is a brain teaser; most brain teasers are not mechanical puzzles.

What is the oldest mechanical puzzle?

The Ostomachion, a fourteen-piece square dissection attributed to Archimedes in the third century BC and preserved in the Archimedes Palimpsest. Romano-Celtic puzzle padlocks from the second century BC are older as objects, but they were made for security rather than amusement.

How are mechanical puzzles officially classified?

By Jerry Slocum's system, which has ten categories and sorts puzzles by what the solver has to do rather than by material or shape. The Lilly Library uses it to catalogue his collection of over 30,000 puzzles. The seven types in this guide map onto it, merging disentanglement into take-apart and separating mazes out.

What should you never do with a mechanical puzzle?

Force it. These are made to tolerances where the correct move needs no strength at all, so if a piece will not go, the move is wrong rather than stiff. Believing you have the right answer and pushing harder is the most common way people break one.

Who invented the 15 puzzle?

Noyes Palmer Chapman, a postmaster in Canastota, New York, who applied for a patent in March 1880. Sam Loyd spent twenty years from 1891 falsely claiming he had invented it, and the misattribution still appears widely today.

Are mechanical puzzles mathematically hard?

Formally, yes. The general packing problem is NP-hard, and sliding-block puzzles were proved PSPACE-complete by Robert Hearn and Erik Demaine, remaining PSPACE-hard even when every piece is a simple domino. That is the reason a handful of pieces can require an enormous solution.

Do mechanical puzzles improve your brain?

They exercise spatial reasoning, which is measurable, trainable by about 0.47 standard deviations across 217 studies, and predicts entry into STEM careers independently of maths and verbal ability. But no study has tested mechanical puzzles specifically, so claims that they are proven to make you smarter are not supportable.

One classic sits slightly outside this taxonomy because it is solved by colour rather than shape: Instant Insanity, four cubes with one correct stack.