Step 1 of 8 · The machine

A machine that draws in 3D

A box on a desk is quietly building a little tower. No mould, no carving, no glue: just a nozzle drawing with hot plastic.

A 3D printer builds solid things by drawing them in melted plastic, one paper-thin layer at a time, from the bottom up.

1. Layer by layer. Look closely at the tower on the bed. It isn't carved or moulded. It's built from flat layers stacked on top of each other, each about 0.2 mm thick, about twice the thickness of a sheet of paper. The printer draws one layer, moves up a tiny bit, and draws the next. A 6 cm tall object takes about 300 layers. Like building a tower out of very thin pancakes, stacked one by one.

   Engineers call this 'additive manufacturing', because material is added only where it's needed. Older ways of making things, like carving or drilling, start with a big block and take material away.

   The layers are thin enough that from arm's length they blend together into a smooth surface. Up close, or with a fingernail, you can feel the tiny ridges.

   Thinner layers look smoother but take longer. Halve the layer height and the printer has to draw twice as many layers.

2. A string of plastic. The printer's 'ink' is a long string of plastic called filament, about 1.75 mm thick, wound on a spool on top. Plastic goes soft when it's hot and hard again when it cools; the usual kind here is PLA, made from plants. The printer pulls the string in, melts it, and lays it down where it's needed. When it cools, it's solid again, in its new shape. Like a hot glue gun: a solid stick goes in, a runny line comes out, and it sets.

   The most common filament is PLA, a plastic made from plants like corn or sugar cane. It melts easily and doesn't smell much, which makes it good for home printers.

   Other filaments include PETG (the plastic of drink bottles), tough ABS (the plastic of LEGO bricks), bendy TPU, and even plastics mixed with wood dust or metal powder.

   A 1 kg spool holds about 330 metres of string, enough for dozens of small objects.

3. A robot that follows orders. The printer doesn't know what it's making. It's a simple robot that follows a long list of instructions: move here, squeeze this much plastic, move there. The print head (the part holding the nozzle) moves left and right, the bed (the flat plate the object sits on) slides forwards and back, and the head steps up after each layer. Three directions are enough to reach any point inside the box. Like drawing a picture by following 'two steps right, one step back' instructions.

   Engineers call the three directions X (left–right), Y (forwards–back) and Z (up–down). Any point in a box can be described by three numbers like these.

   A typical object needs tens of thousands to millions of instructions. The printer's small computer reads them one by one from a memory card or over Wi-Fi.

   Different printers split the work differently. On this one the bed moves for Y. Others keep the bed still and move the head in both X and Y, which allows faster printing.

Layer height

≈ 0.2 mm

Filament

1.75 mm

Did you know? In 2014, astronauts printed the first object ever made in space on a 3D printer aboard the International Space Station: a part for the printer itself.

Read more

Desktop 3D printers like this one use a method called fused filament fabrication (FFF), also known by its trademarked name, FDM. 'Fused' because the hot plastic fuses to what's below; 'filament' because it starts as a string.

The idea was patented in 1989 by Scott Crump, who wanted a way to make a toy frog for his daughter using a glue gun on a moving table. His company, Stratasys, still makes industrial printers today.

In 2005, Adrian Bowyer at the University of Bath started RepRap, a free printer design that could print many of its own plastic parts. When the original patent ran out in 2009, cheap home printers based on RepRap spread around the world.

Today 3D printers make toys, phone cases, spare parts, dental braces, prosthetic hands, rocket engine parts and even houses, using plastics, metals, resins and concrete.

The biggest 3D printers squeeze out concrete instead of plastic and can print the walls of a small house in about a day.

Step 2 of 8 · The machine

Slicing the model

So the printer follows a list of moves. But who writes that list? It starts on the laptop next to it, as a 3D model.

A program called a slicer cuts the 3D model into hundreds of flat layers and plans the nozzle's path around each one.

1. A shape made of triangles. Every print starts as a 3D model: a shape drawn on a computer, or scanned from a real object. Computers store the outside of a shape as thousands of tiny flat triangles joined edge to edge, like a skin. A curved surface is just lots of very small triangles. The file only describes the surface; it says nothing about how to print it yet. Like a paper globe folded from lots of small flat triangles.

   The most common 3D printing file is called STL. It's simply a list of triangles: three corner points for each, plus which way the triangle faces.

   More triangles give smoother curves but bigger files. A detailed model can have millions of triangles.

   You can design your own models in free programs like Tinkercad, or download millions of ready-made ones from online libraries.

2. Cut into slices. The slicer program cuts the model into flat slices, exactly as thick as one layer, usually 0.2 mm. Each slice is the outline of the object at that height. Near the bottom of our tower, the outline is a wide circle; near the top, a narrower one. Stack all the outlines back up and you get the shape again, in steps. Like slicing a loaf of bread, then looking at the shape of each slice.

   The slicer finds each outline by checking which triangles cross that height, and joining the crossing points into closed loops.

   Slopes that lean out too far, like the underside of an arch, have nothing below them to sit on. The slicer can add thin 'supports' underneath that you snap off afterwards.

   Layer height is a choice: 0.1 mm for fine detail, 0.3 mm for quick, rough parts.

3. A path for the nozzle. For every slice, the slicer plans a route for the nozzle: first round the outline to make the walls, then back and forth inside to fill it. It writes the route as G-code, a list of simple lines like 'move to here while squeezing out this much plastic'. That list, often over a million lines long, is what the printer follows. Like a dot-to-dot puzzle, where the numbers tell you which dot to draw to next.

   A typical G-code line looks like 'G1 X50.2 Y30.0 E2.4'. G1 means 'move in a straight line', X and Y are the target position in millimetres, and E is how much filament to push.

   The slicer also sets the temperatures, the speed of every move and when the cooling fan turns on.

   Because the printer just follows the list, the same file prints the same object on any printer of that model, anywhere in the world.

Slices

≈ 300

Instructions

> 100,000

Did you know? G-code was invented in the 1950s at MIT to control milling machines. 3D printers still speak almost the same language.

Read more

Slicing turns a shape into instructions. The 3D model only says where the surface is. The printer needs to know where to move, how fast, and how much plastic to push at every moment.

To find each slice, the program imagines a flat plane cutting through the model at that height. Wherever a triangle crosses the plane, it marks a short line. Join all those little lines end to end and you get the outline of the slice.

Then it plans the route. It shrinks the outline inwards by one line width to place the next wall, and fills what's left with a pattern. It also decides where each layer starts, so the little seam where the nozzle begins and ends is hidden.

Finally, it adds up how far the nozzle travels and how much plastic it needs, so it can tell you how long the print will take and how many grams of filament it'll use, before you start.

A slicer can turn a model into millions of moves in a few seconds, then the printer takes hours to carry them all out.

Step 3 of 8 · The machine

Moving to the exact spot

Now the printer has its route. To follow it, the nozzle must land on exactly the right spot, again and again, for hours.

Motors that turn in tiny exact steps pull belts and turn screws, moving the head and bed to within a fraction of a millimetre.

1. Motors that count. Ordinary motors just spin. A stepper motor has rings of electromagnets, coils of wire that turn into magnets while electricity flows. Switching them on one after another pulls the shaft round in tiny, equal clicks, 200 per turn. Each short burst of electricity, a pulse, moves it one click, so the printer knows where everything is just by counting pulses. No camera, no sensor. Like the second hand of a clock, which ticks forward one exact step at a time.

   Inside a stepper motor, a ring of electromagnets surrounds a magnetic shaft with teeth. Switching the magnets on in turn pulls the teeth round, one step at a time.

   If something blocks the head, the motor can skip steps without the printer knowing. The rest of the print then comes out shifted sideways.

   Most desktop printers have 4 or 5 of these motors: one for each direction, sometimes two for up-down, and one to push the filament.

2. Belts and pulleys. A toothed rubber belt runs from a small pulley on the motor shaft to the print head. When the motor turns, the belt pulls the head left or right; another belt underneath slides the bed forwards and back. One full turn of the pulley moves the belt 40 mm, so one click moves it 0.2 mm. The electronics can split each click into 16 smaller ones, so the head moves just 0.0125 mm at a time, far thinner than a hair. Like a bike chain: the pedals turn a gear, and the chain pulls the wheel.

   The belt's teeth fit into grooves on the pulley so it can't slip. The usual pulley has 20 teeth spaced 2 mm apart: 20 × 2 = 40 mm per turn.

   40 mm ÷ (200 steps × 16 micro-steps) = 0.0125 mm per step, or 80 steps for every millimetre.

   Belts are light and fast. A desktop head can rush around at hundreds of millimetres per second while still landing on its exact spot.

3. Screws that lift. Up and down is slower but must be very precise, because it sets the layer height. Two motors at the sides turn long threaded rods called lead screws. A nut on each screw can't spin, so as the screw turns, the nut climbs it, lifting the whole head. After each layer, the screws turn a fraction of a turn and the head rises exactly 0.2 mm. Like twisting a bottle cap: turning it moves it along the thread.

   Each full turn of a typical lead screw lifts the nut 8 mm. So 0.2 mm is only 1/40 of a turn.

   Screws are slow but strong and steady: the head can't sag or bounce, even though it hangs on them all day.

   If the two side screws get out of step, the bar holding the head tilts, and every layer comes out slightly slanted.

Steps per turn

200 × 16

Per step

0.0125 mm

Did you know? A desktop printer can place its nozzle to about 0.0125 mm, roughly one-eighth the thickness of a human hair.

Read more

The printer's little computer reads one G-code line at a time and turns it into pulses for each motor. To move 10 mm to the right, it sends 800 pulses to the left-right motor, carefully spaced out so the head speeds up gently, cruises, and slows down gently.

Speeding up and slowing down matters. A heavy head that starts or stops too suddenly shakes the frame, and that wobble prints as ripples on the surface, a bit like ripples on a pond.

When you switch a printer on, it doesn't know where anything is. So it 'homes': it moves each part slowly until it touches a small switch at the end. That spot becomes zero, and from then on it just counts.

Many printers also probe the bed before printing, touching it at a grid of points. If the bed is slightly uneven, the printer nudges the height up and down as it goes to follow the bumps.

Over a long print, the motors may take tens of millions of steps without losing count.

Step 4 of 8 · The print head

Gears that push the plastic

The motors can put the nozzle anywhere. Now let's zoom into the print head itself, where the plastic string goes in.

A motor turns a toothed gear that grips the filament and pushes it down into the hot end, the heated tip of the head, exactly as much as each move needs.

1. In goes the string. The filament from the spool enters the top of the print head. At this point it's still a cold, stiff string of plastic, about as thick as a strand of spaghetti. Everything the printer makes is drawn from this single string, so it must arrive smoothly, never tangled and never pulled too hard, or the printer starves for plastic. Like a single thread feeding a sewing machine.

   Filament is made to a tight size: 1.75 mm, accurate to about ±0.02 mm. If it were thicker in places, more plastic would squeeze out there.

   Some plastics soak up water from the air, which turns to steam bubbles in the hot nozzle. That's why spools come sealed with little bags of drying beads.

2. Gears grip and push. Inside, the filament is squeezed between two wheels. One is a toothed gear turned by a stepper motor; its teeth bite into the plastic. The other is a smooth wheel on a spring that presses the filament against the gear. When the motor turns, the string is pushed down. Turn it backwards and the string is pulled back up, to stop drips. Like two rollers in a pasta machine gripping the dough and pushing it through.

   The part with the gear and motor is called the extruder. The printer's computer tells it exactly how many millimetres of filament to push for each move.

   Pulling back a little before each jump is called 'retraction'. It sucks the melt up so the nozzle doesn't leave thin hairy strings across gaps.

   If the gear bites too hard it chews the plastic into dust; too softly and it slips. The spring sets that balance.

3. The solid is the piston. The gear never touches the melted plastic. It only pushes the cold, solid part, and that solid string pushes the melt ahead of it, like a piston in a syringe. Its particles already touch, so it can't be squashed smaller: every millimetre pushed in at the top forces the same amount out of the bottom. That's how the printer controls exactly how much comes out. Like pushing toothpaste out by squeezing from the end of the tube.

   The filament is 1.75 mm wide, but the nozzle hole is only 0.4 mm. So 1 mm of filament pushed in comes out as about 27 mm of thin line.

   This only works if the top stays solid. If heat creeps up and softens the filament too early, it bends and jams instead of pushing. The next step shows how the printer stops that.

Filament width

1.75 mm

Line per mm in

≈ 27 mm

Did you know? The nozzle lays down a line about 27 times longer than the length of filament fed in, because it squeezes a thick string into a thin bead.

Read more

The whole print head has two jobs: push the filament (the extruder at the top) and melt it (the hot end at the bottom). In this one, the motor sits right on the head. Some printers keep the motor on the frame and push the filament down a long tube instead, making the head lighter.

Pushing plastic out of a tiny hole takes force, because melted plastic is thick and sticky, a bit like honey. The gear may need to push with the weight of a few kilograms.

The push and the move must match. If the head speeds up but the gear doesn't push faster, the line gets thin and weak. The slicer works out both together, so the line stays the same width everywhere.

Some extruders use two gears that both grip the filament from opposite sides, so it can be pushed hard without slipping.

Step 5 of 8 · The print head

Melting it, at the right spot

The cold string is being pushed down. At the bottom of the head, it meets a small metal block that is hotter than an oven.

A heater melts the plastic just above the nozzle, while a heatsink and fan keep the plastic above it cold and solid.

1. A tiny oven. Temperature tells you how fast the tiny particles in a material are jiggling. The heater block is an aluminium cube with a heating rod inside, kept at about 210 °C, hot enough to soften PLA plastic into a thick goo. A small sensor next to it measures the temperature many times a second, and the printer switches the heater on and off to hold it steady. Like an oven with a thermostat that clicks on and off to stay at one temperature.

   The heating rod is a 'cartridge heater': a wire inside that gets hot when current flows through it, like a toaster.

   The sensor is a 'thermistor', a tiny bead whose electrical resistance changes with temperature. The printer reads that resistance and works out the temperature.

   If the sensor ever falls out, the printer would think the block is cold and heat it forever. Modern printers spot this and shut the heater off, a safety feature called thermal runaway protection.

2. Cold on top. Above the hot block, the plastic must stay cold and stiff to work as the piston. Heat travels through metal as jiggling passed from particle to particle, so the hot block hangs from a very thin tube, the heat break: little metal, little path for heat. Above it, metal fins with a fan blowing over them carry away what sneaks through. If the fan stops, the filament softens too high up and jams. Like holding a hot pan by a long thin handle that stays cool.

   The fins are a 'heatsink'. Lots of thin fins give lots of surface for air to touch, so heat leaves quickly, the same trick a computer chip's cooler uses.

   The heat break is thin on purpose. Heat flows through metal like water through a pipe; a narrow pipe lets only a trickle through.

   The zone where solid turns to goo is only a few millimetres tall. Keeping it short and in the right place is what makes a print head reliable.

3. The nozzle. At the very bottom is the nozzle, a small brass cone screwed into the hot block. Brass passes heat well, so the plastic stays melted right to the tip. The hole at the end is usually 0.4 mm wide, about four hairs side by side. Bigger holes print faster but rougher; smaller holes print finer details, but take longer.

   Nozzles come in many sizes, from 0.1 mm for tiny models to over 1 mm for big, fast prints.

   Some filaments contain hard bits like carbon fibre or glow-in-the-dark powder, which grind brass nozzles down. Those need hardened steel or even ruby-tipped nozzles.

   The flat end of the nozzle around the hole also matters: it presses on the fresh line and smooths its top, as we'll see next.

Nozzle temp

≈ 210 °C

Hole size

0.4 mm

Did you know? The heater block sits at about 210 °C, while the fins just 2 cm above it stay barely warm to the touch.

Read more

Heating plastic doesn't make it pour like water. Plastic is made of extremely long molecules, so even when hot it stays thick, like honey or warm toffee. That's handy: it holds its shape for a moment after it comes out.

Different plastics need different temperatures. PLA prints at about 190–220 °C, PETG at about 230–250 °C, and tough ABS at about 240–260 °C. Too cold and it won't flow; too hot and it burns or gets runny.

The printer keeps the temperature steady with a control loop called PID. It doesn't just switch fully on or off; it predicts how fast the block is heating or cooling and adjusts the power gently, so the temperature barely moves even when the fan blows on it.

Most jams happen when heat creeps up into the cold zone. The plastic softens there, swells, and sticks to the walls. That's why the heatsink fan on a printer runs whenever the hot end is warm, even if nothing is printing.

If a heater block got stuck on with no sensor, it could pass 400 °C. Printer firmware watches for that and cuts the power within seconds.

Step 6 of 8 · Drawing with plastic

Drawing a line of plastic

Now we zoom past the nozzle's tip, cut open. The goo is pushed down through the hole and out underneath, onto the layer below.

The nozzle squeezes out melted plastic and presses it flat onto the layer below, and a fan sets it hard within about a second.

1. Squeezing out. Inside the nozzle, the melted plastic is pushed down a wide channel that narrows to the 0.4 mm hole. The same amount must get through every second, so as the passage narrows the goo speeds up, like a river through a gap. It comes out as a soft, glowing thread, and the nozzle moves along at the same time, so the thread is laid down as a line. Like icing squeezed out of a piping bag as you move it across a cake.

   The amount squeezed out each second has to match how fast the nozzle moves. A typical line uses only about 1 to 15 cubic millimetres of plastic per second.

   That flow limit is often what sets a printer's top speed: the heater can only melt plastic so fast.

2. The squish. The nozzle doesn't drop the plastic from high up. Its tip glides just 0.2 mm above the layer below, lower than the line is wide. So the soft plastic is pressed down and spread out into a flat line about 0.45 mm wide. The squish pushes it firmly against the layer below, and the flat tip smooths its top so the next layer has a level surface to sit on. Like pressing a sausage of play dough flat with your thumb as you lay it down.

   The gap between the nozzle and the layer below is the layer height. That's why the first layer is so important: if the nozzle is too high, the line just sits on the bed in a round string and doesn't stick.

   If the nozzle is too low, there's no room for the plastic. It gets scraped thin and may block the nozzle.

3. Set by the fan. A second fan points a jet of air at the fresh line just behind the nozzle. Heat flows from hot things into cooler air; the fan keeps swapping warmed air for fresh cool air, so the line turns hard in about a second. It must stiffen fast enough to hold its shape, but stay hot long enough to bond with the layer below (the last step shows how). Like blowing on hot soup so it cools before your next spoonful.

   Different plastics need different amounts of cooling. PLA likes lots of air. ABS shrinks a lot when it cools, so it's printed with little or no fan, often inside a warm box.

   Fans also let printers draw short 'bridges' across empty space: the line is pulled tight between two points and cooled before it can sag.

Line width

≈ 0.45 mm

Sets in

≈ 1 s

Did you know? A typical print lays down a line of plastic hundreds of metres long, all made from one continuous string.

Read more

Melted plastic is 'viscous': thick and slow to flow, like honey. To push it through a tiny hole takes pressure, built up by the gear pushing the cold filament from above.

When the line comes out, two things happen at once. The fresh line is pressed onto the layer below, and heat flows from the hot line into the cooler plastic underneath, softening its top for a moment. That moment is when the layers fuse together.

Then the fan takes over. The fresh line cools from about 210 °C to below 60 °C, where PLA goes from soft to stiff. Above about 60 °C, PLA bends like warm chocolate; below it, it's hard.

Get the timing wrong and you see it in the print. Too little cooling and overhangs droop and sag. Too much and the layers don't bond well and can split apart.

The fresh line cools from about 210 °C to under 60 °C in roughly a second, a temperature drop of 150 degrees in the blink of an eye.

Step 7 of 8 · Drawing with plastic

Walls, infill and the next layer

One line is laid. But a layer is made of many lines, in a careful order, and the inside of the object is mostly empty.

Each layer starts with solid outer walls, then a sparse pattern fills the inside; then the nozzle steps up 0.2 mm and starts again.

1. Walls first. In each layer, the nozzle first traces the outline of the object, usually two or three lines side by side. These are the walls: the skin you see and touch. Stacked layer on layer, they make the smooth outside of the print. Doubling them up makes the skin thicker and stronger, and hides any gaps behind it. Like drawing the outline of a shape before you colour it in.

   Walls are often printed slower than the rest, because they're what you see. A slower, steadier nozzle leaves a neater surface.

   The very top and bottom layers are filled in solid too, so the object is a closed box with no holes.

2. Mostly air inside. Filling the inside solid would waste plastic and hours. So the nozzle draws a sparse grid inside instead, called infill, often just 15–20% plastic and the rest air. The grid lines run one way on one layer and across on the next, like a criss-cross lattice. It's enough to hold up the top and stiffen the walls, while staying light. Like the honeycomb inside a bee's nest: mostly air, but still surprisingly strong.

   Common infill patterns include grids, triangles, honeycombs and wavy 'gyroids'. Each trades strength, speed and plastic differently.

   For a decoration, 10% infill is plenty. For a part that must take a beating, like a bracket or a tool, people use 40% or more, or even solid.

3. Up one layer. When the layer is finished, the lead screws turn a fraction of a turn and the whole head rises exactly 0.2 mm. Then it starts the next layer, sitting on top of the one it just drew. Walls, infill, step up; walls, infill, step up. A 6 cm object repeats this about 300 times, which is why prints often take several hours.

   The spot where each layer starts and ends leaves a tiny bump called the seam. Slicers try to hide it in a corner or at the back.

   Some printers raise the nozzle smoothly as it goes round, in one continuous spiral, for things like vases. That's called 'vase mode'.

Infill

15–20 %

Layer step

0.2 mm

Did you know? A print with 15% infill can weigh less than half as much as the same object printed solid, and still be strong enough for most uses.

Read more

Inside a print, strength comes mostly from the walls and the top and bottom skins, just like a hollow box or a bone. Infill holds the skins apart and stops them buckling, but you don't need much of it.

Each infill line has to rest on something. That's why the grid alternates direction each layer: every new line crosses the lines below it, so it always has something to sit on.

The object's weight, the plastic used and the printing time all grow with infill. Going from 20% to 100% can more than double the weight and time, while the strength goes up by much less.

Engineers sometimes go further and vary the infill inside one part: dense where the forces are big, sparse where they're small. Nature does the same with the spongy bone inside your skeleton.

Gyroid infill was first described by a NASA scientist, Alan Schoen, in 1970, long before 3D printers used it.

Step 8 of 8 · Holding together

Why the layers stick

Layer after layer, the print grows. But what actually holds each new line to the one under it? Shrink down a thousand times to the seam.

Plastic is made of very long, tangled molecules. While hot, they wriggle across the seam into the layer below, and when they cool, they lock in place.

1. Long chains. Everything is made of atoms, and atoms join into groups called molecules. In plastic, the molecules are enormous: thousands of small units linked end to end in a long chain, like beads on a necklace. A plastic is a huge tangle of these chains. Cold, they're frozen in place; hot, they slither and wriggle past each other, which is why the plastic flows. Like a bowl of cooked spaghetti: long strands, all tangled up.

   These long-chain molecules are called polymers, from Greek words meaning 'many parts'. PLA's chains are made of thousands of the same small unit, repeated.

   Real chains are far too thin to see, a few millionths of a millimetre across. We've drawn them as beads so you can follow them.

2. Tangling across. When the hot new line is pressed onto the cooler layer below, it warms that layer's top for a moment. For that instant, chains on both sides can move. They wriggle across the boundary and tangle with each other, just as they're tangled inside each line. When enough chains reach across, the seam disappears and the two layers become one piece. Like pressing two pieces of Velcro together: thousands of tiny hooks catch.

   This joining is a kind of welding: no glue, just the same material melting together. Hotter plastic and slower cooling give the chains more time to cross, so the bond gets stronger.

   That's why printing too cold makes weak parts that snap along the layers.

3. Locked, and a bit weaker. As the fan cools the line, the chains slow down and freeze where they are, locking the tangle in place. But they only had about a second to cross, so the seam ends up less tangled than the inside of a line. That's why 3D prints are usually weakest between layers. Cooling also shrinks the plastic slightly, which can bend tall or wide prints. Like a zip that only got done up part of the way.

   Tests show prints pulled apart across the layers are often only about half to three-quarters as strong as when pulled along them.

   Engineers turn parts on the bed so that the big forces run along the layers, not across them.

   When a print shrinks unevenly, its corners can curl up off the bed. That's called warping, and it's why the bed is heated: it keeps the bottom warm so it shrinks gently.

Weld time

≈ 1 s

Across layers

50–80 % as strong

Did you know? Some engineers bake finished prints in an oven at a gentle heat. The chains get a little more time to tangle across the seams, and the part gets stronger.

Read more

Atoms are the tiny building blocks of everything. When atoms hold on to each other they form molecules. Water's molecules have just three atoms. A single PLA chain can have tens of thousands.

The chains in a solid plastic aren't stuck to each other by glue. They hold together because they're so long and tangled that pulling one out means dragging it past thousands of others, a bit like pulling one strand out of a ball of wool.

When the new line lands on the old one, the boundary between them is sharp. Heat lets the chains near it move, and they spread across, a process called diffusion. The more they spread before cooling, the stronger the bond.

This is why printing settings matter so much for strength: nozzle temperature, how fast the line cools, the fan, and even the temperature of the room all change how long the chains can wriggle.

The PLA in many 3D printers is made from corn starch or sugar cane: the plastic starts life as a plant.