Topology: What survives stretching: polyhedra, Möbius strips, and invariants

Take a coffee mug and suppose it's made of soft clay that never cracks. Your job is to reshape it into a ring doughnut. You may push, pull, stretch, and squash as much as you like. You may not tear the clay, poke a new hole through it, or press two separate parts together so they stick. Can it be done? Try it in your head before you read on, or with a lump of real clay if you have one.

Here is one way. Press the hollow of the mug flat, so the bowl becomes a thick disk with the handle still attached. Let the disk shrink and flow into the handle, thickening it as it goes. What's left is a fat ring of clay with one hole through it. That hole was there from the start: it's the gap between the handle and the side of the mug. Smooth the ring and you have a doughnut.

Look at what the reshaping destroyed. The mug's height, the curve of its rim, its flat base, and the volume it could hold are all gone. One thing survived: the hole through the handle. You could stretch that hole, shrink it, or move it around. Without tearing, you couldn't get rid of it. Without gluing, you couldn't add a second one.

Now try a cup with no handle. Its hollow can be pressed flat and the clay rolled into a ball, because nothing goes all the way through it. A handled mug can't become a ball. However you push the clay, the handle's hole stays, and a ball has none. The two objects differ in exactly the one way that stretching can't change.

The branch of mathematics that studies such properties is topology: the study of the features of a shape that survive stretching and bending without tearing or gluing. The word first appeared in print in 1847, in a short study by the German mathematician Johann Benedict Listing, though the questions behind it are older (O'Connor and Robertson 1996c).

What topology offers is a way to tell shapes apart when measurement is beside the point. It asks questions like these:

• Does the shape come in one piece or several?
• How many holes run through it?
• Does its surface have an edge, and how many sides does it have?

Two shapes that give different answers can't be reshaped into each other, however much you stretch them. Mathematicians have turned these questions into precise counting methods. A single count of corners, edges, and faces, for example, gives a number that no amount of stretching can alter. Methods like these now show up in cell biology, in the analysis of large data sets, and in the physics of new materials.

References

• O'Connor, J. J., and E. F. Robertson. 1996c. "A History of Topology." MacTutor History of Mathematics Archive, University of St Andrews. Last updated May 1996.
  • Free: MacTutor: A history of Topology

Introduction

Topology is the study of the features of a shape that survive stretching and bending without tearing or gluing. It has no founding moment. Its key observations piled up over about 150 years, made by people who often didn't know they were starting a new branch of mathematics.

This reading traces the main steps: Euler's count for solids and its first correction, Listing's new word, the discovery of a surface with one side, and Poincaré's work that gave the subject a firm footing.

Euler's Count for Solids

A polyhedron is a solid bounded by flat faces. Each face is a polygon, the faces meet along straight edges, and the edges meet at corners called vertices. A cube is a polyhedron with 6 square faces, 12 edges, and 8 vertices.

In 1750 the Swiss mathematician Leonhard Euler wrote to his colleague Christian Goldbach about a pattern he had noticed. For every polyhedron he tried, the number of vertices minus the number of edges plus the number of faces came to 2. Euler published the result in 1752 in two papers. In the first he admitted that he couldn't prove it. In the second he offered a proof that worked by slicing pieces off the solid one at a time (O'Connor and Robertson 1996c).

What made the observation unusual was what it ignored. Geometry had always been about lengths, angles, and areas. Euler's count used none of them. You could make the cube taller or lean it over, and the count stayed the same.

The rule had limits that Euler didn't state. In 1813 the Swiss mathematician Simon Lhuilier found that solids with holes through them break it, and break it in a regular way: each hole lowers the total by 2. MacTutor describes this as the first known result on a topological invariant, a number that doesn't change when a shape is distorted (O'Connor and Robertson 1996c).

A Name and a Band with One Side

Johann Benedict Listing, a German mathematician, had been using the word "topology" in letters for about ten years when he published Vorstudien zur Topologie (Preliminary Studies on Topology) in 1847. It was the first time the word appeared in print (O'Connor and Robertson 1996c).

In 1861 Listing published a more important paper. It described a strip of paper given a half-twist before its ends are joined. The result is a one-sided surface: a surface on which you can travel from any point to the spot directly behind it without crossing an edge. August Ferdinand Möbius published his own description of the same band in 1865, and it now carries his name. Möbius tried to capture its strangeness by showing that its surface couldn't be covered with small triangles all oriented the same way (O'Connor and Robertson 1996c). The two men appear to have found the band independently.

Poincaré's Analysis Situs

Through the 19th century, other mathematicians studied how surfaces are connected. Bernhard Riemann did so in 1851 and 1857, and Camille Jordan later showed that certain counts of loops on a surface don't depend on how the surface is drawn (O'Connor and Robertson 1996c). The results were scattered, and their proofs often leaned on pictures.

In 1895 the French mathematician Henri Poincaré published Analysis Situs, the first of a series of papers. The Latin title means "analysis of place." Analysis situs was the older name for the study of position and connection without measurement, and Poincaré's papers put the idea of connectivity on a fully rigorous basis. They also introduced tools that topologists still use for counting holes in any number of dimensions (O'Connor and Robertson 1996c).

Conclusion

By 1900 the scattered observations had a name, a set of standard examples, and a rigorous method. No single person founded topology. Euler's count, Lhuilier's correction, Listing's word and band, Möbius's description, and Poincaré's papers each supplied a piece, and the field took shape slowly from all of them.

Key Terms

• Topology: The study of the features of a shape that survive stretching and bending without tearing or gluing.
• Polyhedron: A solid bounded by flat polygon faces that meet along straight edges and at corners called vertices.
• One-sided surface: A surface on which you can travel from any point to the spot directly behind it without crossing an edge.
• Analysis situs: Latin for "analysis of place," an older name for the study of position and connection without measurement, and the title of Poincaré's 1895 work.

References

• O'Connor, J. J., and E. F. Robertson. 1996c. "A History of Topology." MacTutor History of Mathematics Archive, University of St Andrews. Last updated May 1996.
  • Free: MacTutor: A history of Topology

Introduction

A topologist will tell you that a square and a circle are the same shape. That sounds wrong until you see which question is being asked: not "do these look alike?" but "could one become the other without tearing?"

This reading sets out the rules for reshaping a figure, the properties those rules throw away, the properties they keep, and the technical name for "topologically the same."

The Rules of the Game

Topology treats every shape as if it were made of perfectly stretchy rubber. A continuous deformation is a change of shape made by stretching, squashing, or bending, with no cutting, tearing, or gluing. Points that start close together stay close together, and no new connections appear.

The last row of allowed moves matters. Corners don't count in topology. A square's four corners can be rounded off by stretching, and a circle can be pinched into four corners. So a square and a circle pass the test: either can be deformed into the other.

What Gets Thrown Away

Under these rules, most of the measurements that school geometry cares about stop mattering:

• Length. Any piece can be stretched or shrunk.
• Angle. Any corner can be opened, closed, or smoothed away.
• Area. A region can be expanded or squeezed.
• Straightness. Any straight segment can be bent into a curve.

This is a deliberate loss. Euler's count of vertices, edges, and faces on a solid used no lengths or angles, and mathematicians who studied it came to see that some facts about a shape don't depend on measurement at all (Richeson 2008). Topology is built from those facts.

What Survives

What's left is a short list of properties that no continuous deformation can change. Such a property is an invariant: a feature or number that stays the same when a shape is stretched or bent without tearing or gluing. Three invariants are easy to see:

• The number of pieces. One piece stays one piece, because separating it would need a cut.
• The number of holes. A hole here means a gap that runs all the way through a shape, like the opening in a ring or the space inside the letter O. Removing a hole would need gluing; adding one would need tearing.
• Whether the shape has edges. The surface of a ball has no edge. A flat disk has one, around its rim. No stretching turns one into the other.

Invariants are how topologists prove that two shapes differ. If one shape has an invariant the other lacks, no deformation links them. In the 19th century, Camille Jordan showed that a certain count of independent loops on a surface is an invariant of this kind (O'Connor and Robertson 1996c).

Same or Different: Two Test Cases

A square and a circle are the same. Each is one piece, each surrounds one region, and neither has loose ends. Rounding the square's corners turns it into the circle.

A circle and a figure eight are different. Both are one piece, but the figure eight surrounds two regions and the circle only one. There's a second way to see it. Remove any single point from a circle and it stays in one piece, an arc. Remove the crossing point of the figure eight and it falls into two pieces. "What happens when you remove a point" is itself an invariant, so no deformation can turn one shape into the other.

The Technical Name

Two shapes that are topologically the same are called homeomorphic. Strictly, this means there is a one-to-one match between the points of the two shapes that keeps nearby points nearby, in both directions. Stretching and bending without tearing or gluing is the intuitive picture of such a match. One subtlety: being homeomorphic doesn't depend on how a shape sits in space. A loop of string tied in a knot and a plain circle are homeomorphic, as loops, even though you can't untie the first into the second without cutting.

Conclusion

Topology keeps a shape's connections and throws away its measurements. Two shapes count as the same when one can be continuously deformed into the other, and they're shown to be different by finding an invariant on which they disagree. The number of pieces, the number of holes, and the presence of edges are the first invariants most people meet.

Key Terms

• Continuous deformation: A change of shape made by stretching, squashing, or bending, with no cutting, tearing, or gluing.
• Invariant: A feature or number that stays the same when a shape is stretched or bent without tearing or gluing.
• Hole: A gap that runs all the way through a shape, like the opening in a ring or the space inside the letter O.
• Homeomorphic: Topologically the same: two shapes are homeomorphic when there is a one-to-one match between their points that keeps nearby points nearby, in both directions.

References

• O'Connor, J. J., and E. F. Robertson. 1996c. "A History of Topology." MacTutor History of Mathematics Archive, University of St Andrews. Last updated May 1996.
  • Free: MacTutor: A history of Topology
• Richeson, David S. 2008. Euler's Gem: The Polyhedron Formula and the Birth of Topology. Princeton, NJ: Princeton University Press.
  • Free: None
  • Free companion: O'Connor and Robertson 1996c

Introduction

The skin of a ball and the skin of a doughnut are both smooth, unbroken, and edgeless, yet no amount of stretching turns one into the other. A strip of paper with a half-twist looks almost the same as a plain loop, yet it differs in a way an ant walking on it would discover.

This reading sorts surfaces in two ways: by how many holes they have, and by whether they have two sides.

Surfaces and Their Holes

A surface is a shape that, close up, looks like a piece of a flat sheet. It is the skin of an object, not the solid inside. The surface of a ball, the surface of a doughnut, and a sheet of paper are all surfaces.

Surfaces don't have to be closed. A sheet of paper has one edge running around its border. A tube, like the side of a tin can with both lids removed, has two edges, one at each end. Counting edges is one way to tell surfaces apart.

Three closed surfaces, ones with no edge, give the basic pattern:

A torus is the surface of a ring doughnut: a closed surface with one hole through it. Here "hole" means a gap that runs all the way through the shape, not the hollow space inside it. A beach ball is hollow, but its skin is still a sphere with no hole.

Topologists call the number of holes in a closed surface its genus. A sphere has genus 0, a torus genus 1, and a double torus genus 2. Genus is an invariant: a feature or number that stays the same when a shape is stretched or bent without tearing or gluing. That's why a coffee mug's surface, with its single handle, counts as a torus.

The Band with One Side

Take a strip of paper, give one end a half-twist, and tape the ends together. The result is a Möbius band: a strip joined with a half-twist, which has only one side and only one edge. Run a finger along the middle and you return to your starting point on what seemed to be the other face of the paper, without ever crossing the edge. Trace the edge and you find there is only one, a single loop twice as long as the strip.

Johann Benedict Listing described this band in 1861, and August Ferdinand Möbius published his own description in 1865 (O'Connor and Robertson 1996c). The two worked independently. The band is named for Möbius.

Which Way Is Up?

The one-sidedness of the band has a more precise version. A surface is orientable if you can consistently say which side is "up" everywhere on it. Here is a test. Draw a small circle with an arrow running clockwise, and slide the drawing around the surface without lifting it. On a sphere or a torus, the arrow always comes back clockwise. On a Möbius band, carrying it once around the band brings it back running counterclockwise.

So spheres and tori are orientable, and the Möbius band is not. Möbius himself expressed the band's strangeness this way: its surface can't be covered with small triangles that are all oriented compatibly (O'Connor and Robertson 1996c).

Why One-Sidedness Matters

Every small patch of a Möbius band looks exactly like a small patch of an ordinary loop. A tiny creature living on the band could measure lengths and angles all day and find nothing unusual. The difference shows up only when it travels all the way around and comes back mirror-reversed.

This is a pattern that runs through topology. Some properties of a shape can't be detected by any local measurement, however careful. They belong to the shape as a whole. Orientability is one of them, and genus is another.

Conclusion

Closed surfaces can be told apart by their genus, the number of holes they have, and surfaces of every kind can be sorted by whether they are orientable. Both properties are invariants. Both are also global: you can't detect them by examining one small region.

Key Terms

• Surface: A shape that, close up, looks like a piece of a flat sheet; the skin of an object rather than the solid inside.
• Torus: The surface of a ring doughnut: a closed surface with one hole through it.
• Genus: The number of holes in a closed surface.
• Möbius band: A strip joined with a half-twist, which has only one side and only one edge.
• Orientable: Describes a surface on which you can consistently say which side is "up" everywhere.

References

• O'Connor, J. J., and E. F. Robertson. 1996c. "A History of Topology." MacTutor History of Mathematics Archive, University of St Andrews. Last updated May 1996.
  • Free: MacTutor: A history of Topology

Introduction

Count the corners, edges, and faces of a box and combine them in the right way, and you get 2. Do the same for a pyramid, a soccer ball, or a lopsided crystal, and you get 2 again.

This reading shows how the count works, why stretching and subdividing can't change it, what happens when a solid has holes, and what Euler's own argument got wrong.

Counting a Cube and a Tetrahedron

A polyhedron is a solid bounded by flat polygon faces that meet along straight edges and at corners called vertices. Three kinds of part can be counted:

• A vertex is a corner of a polyhedron, where edges meet. Write V for the number of vertices.
• An edge is a straight segment where two faces of a polyhedron meet. Write E for the number of edges.
• A face is one of the flat polygons that bound a polyhedron. Write F for the number of faces.

In 1750 Leonhard Euler noticed that for every polyhedron he tried, vertices minus edges plus faces came to 2 (O'Connor and Robertson 1996c). In symbols:

For the cube, 8 - 12 + 6 = 2. For the tetrahedron, 4 - 6 + 4 = 2. You can check a solid of your own the same way. A pyramid with a square base has 5 vertices, 8 edges, and 5 faces, and 5 - 8 + 5 = 2. The number V - E + F is called the Euler characteristic of the polyhedron: vertices minus edges plus faces, a number that tracks how many holes a solid has.

Why Stretching and Subdividing Don't Change It

Stretching a cube into a long box, or leaning it over, changes lengths and angles but not the counts. The Euler characteristic is an invariant: a feature or number that stays the same when a shape is stretched or bent without tearing or gluing.

Subdividing doesn't change it either. Draw a diagonal across one face of the cube. You've added one edge, and you've split one face into two, so you've added one face. The changes cancel: E goes up by 1 and F goes up by 1, so V - E + F is unchanged. Put a new vertex in the middle of an edge and the same thing happens: V goes up by 1 and E goes up by 1. However finely you cut up the faces, the total stays at 2.

When a Solid Has Holes

In 1813 the Swiss mathematician Simon Lhuilier noticed that Euler's formula fails for solids with holes through them. A square picture frame built from wooden beams, for example, gives 0, not 2. Lhuilier showed that a solid with g holes gives

so 2 for no holes, 0 for one hole, and -2 for two holes. MacTutor calls this the first known result on a topological invariant (O'Connor and Robertson 1996c).

This is why the Euler characteristic counts as a landmark. It is a single number, found by counting, and no deformation can change it. Two solids with different Euler characteristics can't be stretched into each other, whatever they look like.

Gaps in Euler's Argument

Euler's proof assumed that the solids were convex, meaning that the straight line between any two points of the solid stays inside it (O'Connor and Robertson 1996c). Many polyhedra aren't convex, and some, like the picture frame, don't satisfy the formula at all. Euler's argument didn't say which solids it covered or why.

Later mathematicians spent decades finding exceptions and repairing the statement. The modern version says exactly which solids give 2, and it treats the Euler characteristic as a property of the surface rather than of one particular way of cutting it into faces. Poincaré extended the idea to shapes in any number of dimensions, which is why the number is also called the Euler–Poincaré characteristic (Richeson 2008).

Conclusion

For any polyhedron without holes, V - E + F = 2, and for a solid with g holes it is 2 - 2g. Stretching and subdividing leave it unchanged, so the count measures something about the shape that measurement can't. Euler's original proof had gaps, which later mathematicians found and closed.

Key Terms

• Vertex: A corner of a polyhedron, where edges meet.
• Edge: A straight segment where two faces of a polyhedron meet.
• Face: One of the flat polygons that bound a polyhedron.
• Euler characteristic: Vertices minus edges plus faces, a number that tracks how many holes a solid has.
• Invariant: A feature or number that stays the same when a shape is stretched or bent without tearing or gluing.

References

• O'Connor, J. J., and E. F. Robertson. 1996c. "A History of Topology." MacTutor History of Mathematics Archive, University of St Andrews. Last updated May 1996.
  • Free: MacTutor: A history of Topology
• Richeson, David S. 2008. Euler's Gem: The Polyhedron Formula and the Birth of Topology. Princeton, NJ: Princeton University Press.
  • Free: None
  • Free companion: O'Connor and Robertson 1996c

Introduction

Euler's rule says that for a solid without holes, the number of vertices minus the number of edges plus the number of faces is 2. A rule like that is worth testing by hand, on solids you can picture, including one built to break it.

This reading works through four solids, counting each part with care, and shows how the result changes when a hole runs through the solid.

The Problem

You have four solids. Each is a polyhedron: a solid bounded by flat polygon faces that meet along straight edges and at corners called vertices.

1. A tetrahedron: a pyramid with a triangular base, so four triangular faces in all.
2. A cube.
3. An octahedron: two square-based pyramids glued base to base, giving eight triangular faces.
4. A "picture frame": a square frame built from four rectangular beams, with a square hole through the middle.

For each solid, count the vertices V, the edges E, and the faces F, and compute V - E + F. Then explain the pattern. The quantity V - E + F is called the Euler characteristic: vertices minus edges plus faces, a number that tracks how many holes a solid has. In 1813 Simon Lhuilier showed that a solid with g holes has Euler characteristic 2 - 2g (O'Connor and Robertson 1996c).

Working It Through

Step 1: Count the tetrahedron

A tetrahedron has 4 corners, so V = 4. Every pair of corners is joined by an edge. There are 6 pairs among 4 corners, so E = 6. It has 4 triangular faces, so F = 4.

You can check the edge count another way. Each of the 4 triangles has 3 sides, which makes 12 sides in total. Every edge is shared by exactly 2 faces, so each edge was counted twice: 12 \div 2 = 6.

Step 2: Count the cube

A cube has 4 corners on top and 4 on the bottom, so V = 8. It has 4 edges around the top, 4 around the bottom, and 4 vertical edges joining them, so E = 12. It has 6 square faces, so F = 6.

The shared-edge check works here too: 6 squares with 4 sides each give 24 sides, and 24 \div 2 = 12.

Step 3: Count the octahedron

Picture the octahedron as a square "equator" with one point above it and one below. That's 4 vertices around the middle, plus a top and a bottom, so V = 6. There are 4 edges around the square, 4 edges running up to the top point, and 4 running down to the bottom point, so E = 12. The upper pyramid has 4 triangular faces and the lower one has 4, so F = 8. The square where the pyramids were glued is inside the solid, so it isn't a face.

Check: 8 triangles with 3 sides each give 24 sides, and 24 \div 2 = 12.

Step 4: Count the picture frame

Figure: A square picture frame made of four beams with mitered corners, drawn at an angle so its 16 vertices are visible: 4 outer and 4 inner corners on the top face, and the same 8 on the bottom face. Top and bottom are each divided into 4 trapezoids, and the outer and inner walls are each 4 rectangles. Alt text: A thick square frame with a square hole in the middle, its top face split into four trapezoids by lines running from each inner corner to the matching outer corner.

Look at the top of the frame. It's a square ring, with 4 outer corners and 4 inner corners. The bottom is the same. So V = 8 + 8 = 16.

Each beam's top face is a trapezoid, a four-sided shape with two parallel sides of different lengths, because the beams meet at angled corners. The top of the frame therefore has 4 outer edges, 4 inner edges, and 4 short edges joining each inner corner to its outer corner: 12 edges. The bottom has another 12. Then there are vertical edges: 4 at the outer corners and 4 at the inner corners, 8 in all. So E = 12 + 12 + 8 = 32.

For the faces, count 4 trapezoids on top, 4 on the bottom, 4 rectangles on the outer walls, and 4 rectangles on the inner walls, around the hole. So F = 16.

Check: all 16 faces have 4 sides, giving 64 sides in all, and 64 \div 2 = 32. The answer isn't a slip. The frame really does give 0.

Step 5: Interpret the results

Three solids gave 2 and one gave 0. The difference isn't size or angle; you could make the cube enormous or the octahedron lopsided and the counts wouldn't change. The difference is the hole. The tetrahedron, cube, and octahedron have no hole through them, so g = 0 and 2 - 2g = 2. The picture frame has one hole, so g = 1 and 2 - 2g = 0.

The frame also explains why the count measures something real. You could build the frame from eight beams instead of four, or split each wall into smaller rectangles, and the counts of faces and edges would change. The total would still be 0, because each extra edge you add comes with an extra face or an extra vertex that cancels it (Richeson 2008).

Key Considerations

Count each edge once. Every edge of a polyhedron sits between exactly two faces. If you count edges face by face, you'll count each one twice. The shared-edge check, adding up the sides of all faces and dividing by 2, catches this mistake.

The common mistake: double-counting, or mislabeling the frame's faces. The most frequent error is counting the edges face by face and forgetting to divide by 2, which gives 64 edges for the frame instead of 32. A second error is to treat the frame's top and bottom as squares. The top of the frame is a square ring, and a square ring isn't a single polygon face: a face must be a polygon with no hole in it. Splitting the ring into four trapezoids is what makes it a legitimate set of faces. If you count the ring as one face, you get the wrong number of faces and edges, and the total comes out wrong.

What the formula assumes. Euler's own argument assumed the solid was convex, so that the straight line between any two of its points stays inside it (O'Connor and Robertson 1996c). The picture frame isn't convex: a line from one side of the hole to the other passes through empty space. The rule 2 - 2g is the corrected version that covers solids like it.

Summary

Counting carefully, with each edge counted once, gives 2 for every solid without a hole and 0 for the frame with one hole, exactly as 2 - 2g predicts.

Solid	V	E	F	V - E + F	Holes
Tetrahedron	4	6	4	2	0
Cube	8	12	6	2	0
Octahedron	6	12	8	2	0
Picture frame	16	32	16	0	1

1. Every edge count was checked by adding the sides of all faces and dividing by 2.
2. The three solids without holes give 2; the frame, with one hole, gives 0.
3. Both results match Lhuilier's rule V - E + F = 2 - 2g, with g the number of holes.

References

• O'Connor, J. J., and E. F. Robertson. 1996c. "A History of Topology." MacTutor History of Mathematics Archive, University of St Andrews. Last updated May 1996.
  • Free: MacTutor: A history of Topology
• Richeson, David S. 2008. Euler's Gem: The Polyhedron Formula and the Birth of Topology. Princeton, NJ: Princeton University Press.
  • Free: None
  • Free companion: O'Connor and Robertson 1996c