The Gold Standard of Knot Theory

The gold standard of knot theory is an argument that distinguishes a change in drawing from a change in the knot itself. Knot theory studies closed loops in three-dimensional space and the ways they can be deformed without cutting or passing a strand through another. Its central questions are concrete: are two loops equivalent, and what evidence could establish the answer?

This root guide connects the eduKate branches on topology, algebra, geometry, graph theory and mathematical reasoning. It begins with diagrams and permitted moves, then builds an original worked calculation using three colours. The arithmetic is accessible; interpreting what it proves is the more important lesson.

The scope is classical smooth or polygonal knots and links in three-dimensional space. The discussion uses the usual unrestricted ambient deformation model. Physical cord adds thickness, friction and length constraints, so observations about a classroom model need their assumptions stated.

The educational purpose follows The Gold Standard of Education: a learner should carry the reasoning into unfamiliar problems. Numerical exercises and teaching records below are original worked material. Linked university notes provide actual knot diagrams and theorems for further study.

Featured photograph: mathematics study materials in daylight. It illustrates mathematical study. For knot diagrams used in tracing and move exercises, follow the illustrated UCLA introduction.

Start with a loop, rather than a length of string with two free ends. In the classical model, a knot is an embedded circle: it closes on itself and does not intersect itself in space. The unknotted circle is included in the subject, despite its everyday name suggesting an absence of knotting.

Two such loops are considered equivalent when an allowed deformation of the surrounding space carries one onto the other. Informally, the loop may bend and stretch while remaining unbroken and avoiding strand passage. The surrounding-space condition matters because an arbitrary change in a drawing can conceal an illegal event.

For a classroom demonstration, join the ends of a loose cord and keep the join visible. Ask the learner to identify which manipulations preserve the agreed model. Opening the join, moving an end through a loop and reconnecting it changes the problem. A successful physical rearrangement must be interpreted in light of the rules actually followed.

The introductory material in Olga Radko’s UCLA knot theory course gives the guiding equivalence question and a route towards numerical and polynomial invariants.

Write the task in one sentence before experimenting: “Decide whether these closed loops are equivalent under deformation without strand passage.” That statement prevents a practical success at untying an open rope from being presented as a solution to a different mathematical question. It also makes clear which observations another learner would need to reproduce.

Contents · Next chapter

A knot diagram represents a spatial loop through a planar projection together with crossing information. At an ordinary crossing, one strand passes over the other. They do not meet at a shared point in space. A drawing that omits the over–under choice leaves important information unspecified.

Trace one strand through a crossing by continuing along its own path. Do not turn onto the other strand as though the crossing were a junction. In a diagram, a short visual gap often marks the underpassing strand; it is a drawing convention, not an actual cut in the loop.

A useful first task is an audit rather than an identification. Check that every strand continues, every crossing has an unambiguous over–under assignment and there are no accidental endpoints. Then trace a full circuit and mark the return to the starting point. Only after this audit should the learner name the object.

Saul Schleimer’s Warwick lecture notes distinguish a projection from a diagram and provide examples to inspect. Use those figures when practising the tracing task; a verbal description alone does not supply the missing crossing data.

For transfer, compare this with a road-map crossing. A bridge and an intersection may look similar in an oversimplified map while permitting different journeys. The analogy helps explain why a representation needs the information relevant to its purpose. It does not make the mathematical crossing a physical road junction.

Previous chapter · Contents · Next chapter

Reidemeister moves describe three kinds of local change in a classical diagram. Type I introduces or removes a small curl. Type II introduces or removes an appropriate pair of crossings. Type III slides a strand across a crossing while preserving the relevant over–under relationships.

These are specific local configurations, not permission to erase any crossing or cancel any neighbouring pair. The portion outside the chosen region remains fixed during the local replacement. A learner should check the whole local pattern before applying the name of a move.

The Reidemeister theorem relates diagram manipulation to the spatial equivalence problem: equivalent classical knots have diagrams connected by a finite sequence of these moves together with planar deformation. The UCLA introductory lecture diagrams provide a visual reference for the moves.

Make a move record with a before drawing, an after drawing and the move type. Ask a partner to trace the strands through both pictures. If a crossing has been switched rather than transformed by a permitted move, the record needs correction.

This is an early proof lesson. A sequence of verified moves can demonstrate equivalence. A sequence containing one unjustified replacement cannot be repaired by having many correct steps elsewhere. The valuable habit is to inspect the transition, not merely whether the final picture looks simpler. Later algebraic methods will check features that survive every legal transition.

Previous chapter · Contents · Next chapter

An invariant takes the same value on equivalent objects. It may be a number, a polynomial or another mathematical structure. To use a diagram-based calculation as a knot invariant, one needs a reason that the result survives the permitted changes of diagram.

The logic is one-directional. If two knots have different values of a valid invariant, they cannot be equivalent. If their values agree, that particular test has not distinguished them. Equality proves equivalence only when the chosen invariant is known to be complete for the class being considered.

Try an original analogy with containers. The number of objects inside a sealed container remains unchanged when it is moved around. Different counts establish different contents in at least that respect. Equal counts do not establish identical contents. The analogy illustrates the logical structure without pretending that counting objects classifies knots.

A candidate measurement can fail the invariant test. Counting all visible crossings in the current picture changes when a Type I curl is added. The loop remains equivalent, so the displayed crossing count cannot serve as an invariant of the underlying knot.

The connected Ledger of Invariants develops this transferable idea. For every proposed test, record the object being measured, the allowed transformations and the claim that survives. That short record makes it much harder to confuse a convenient feature of a representation with an established property of the represented object.

Previous chapter · Contents · Next chapter

A link consists of one or more disjoint closed components; a knot is the one-component case. The number of components is preserved by the allowed deformation. Two diagrams with different component counts therefore represent different links immediately.

Component counting is a tracing task, not a count of visible lobes. Start at a point away from a crossing and follow the same strand until it returns. Mark that circuit. Repeat from an unmarked strand until every part has been visited.

A single component can project into many crossings and apparent loops. Conversely, several components may be interlaced so that a quick glance suggests one complicated strand. The audit should follow continuity through each crossing without treating it as a junction.

For an original teaching record, suppose diagram P has two traced circuits and diagram Q has three. The component invariant settles their inequivalence. If another pair each has two, the count alone leaves the comparison unresolved. Having a quick test with a limited scope is useful when its limit is understood.

The UCLA notes on operations and elementary invariants continue into linking and colouring methods. The connection to graph theory also deserves care: a crossing becomes a graph vertex only under a specified graph representation. Changing that convention changes what a graph path means. Preserve the strand information needed for the original problem.

Keep the component labels separate from arc labels used later for colouring. One component can be divided into several colouring arcs because the arc convention breaks at undercrossings. Calling each arc a separate loop would change both counts. For a peer check, ask one student to produce a component trace and another to identify the colouring arcs on the same supplied diagram. Compare their boundaries and explain why the two partitions serve different purposes. This prepares the algebraic stage without assuming that every visible segment is a different physical piece. A complete loop remains connected while its diagram is organised into several analysis units.

Previous chapter · Contents · Next chapter

The crossing number of a knot is the minimum number of crossings among all its diagrams. A particular drawing supplies a candidate count. It does not automatically demonstrate that no diagram with fewer crossings exists.

Suppose an original exercise begins with a diagram having eight crossings. A verified sequence simplifies it to a diagram with five. This proves that the knot’s crossing number is at most five. It also shows that the initial picture was not minimal. The exercise has not supplied a lower bound of five.

To establish an exact minimum of five, another argument must rule out every smaller possibility. This is the familiar pattern of matching an upper bound with a lower bound. Constructing an example and proving optimality are different achievements.

The American Mathematical Society discussion of knot classification explains crossing number as a minimum and the wider search for invariants. It provides a useful reminder that an attractive standard picture still needs mathematical justification for any minimality claim.

A classroom report should distinguish the count before simplification, the count after simplification and the conclusion supported. Write “a five-crossing representation was obtained” when that is what was demonstrated. This wording preserves the accomplishment while keeping the remaining question visible. The same habit transfers to shortest-route problems, efficient algorithms and design optimisation: a good construction need not be the best possible construction.

An efficient way to assess an optimisation claim is to ask for two certificates. The construction certificate shows an admissible object achieving the proposed value. The exclusion certificate explains why a better value is impossible. In our fictional simplification exercise, the five-crossing drawing serves the first role. A list of failed attempts to draw four crossings does not automatically serve the second. Keeping the certificates separate helps a learner recognise partial progress accurately. It also clarifies what the next mathematical tool must contribute: either an improved construction or a justified obstruction to improvement, depending on the gap still present.

Previous chapter · Contents · Next chapter

Three-colouring assigns one of three labels to each arc of a knot diagram. For this convention, an arc continues through overcrossings and ends at undercrossings. At every crossing the three incident arc colours must be either all the same or all different. A nontrivial colouring uses more than one colour.

Represent the colours by 0, 1 and 2, and calculate modulo 3. If the overpassing arc has label b and the underpassing labels are a and c, the rule is a + c ≡ 2b modulo 3. Here congruence means the two sides differ by a multiple of three.

Check original assignments. With a = 0, b = 1 and c = 2, both sides are 2 modulo 3, so the crossing passes. With a = 0, b = 0 and c = 1, the sides are 1 and 0, so it fails. With all three labels equal, the condition always passes.

The UCLA lecture on three-colourings connects this rule with linear algebra over the three-element field. The benefit of numbers is that the same local test can be checked systematically.

Before calculating a whole diagram, define the arcs and verify one crossing at a time. Decorative colouring of arbitrary strand pieces is not the mathematical construction. The chosen units and local equations determine exactly what the count will mean.

Previous chapter · Contents · Next chapter

For a standard three-arc trefoil diagram, label the arcs a, b and c. The three crossing equations can be written as 2a ≡ b + c, 2b ≡ a + c and 2c ≡ a + b modulo 3. Consult the linked university diagrams when matching these variables to actual arcs.

Because 2 ≡ −1 modulo 3, each equation is equivalent to a + b + c ≡ 0. Thus one may choose a freely in three ways and b freely in three ways; c is then determined uniquely as the residue of −a − b. The total number of assignments is 3 × 3 = 9.

For example, a = 2 and b = 2 force c = 2, giving a monochromatic assignment. With a = 2 and b = 0, c = 1, giving three different colours. Direct substitution checks both examples against every equation.

Three of the nine assignments are monochromatic: all 0, all 1 or all 2. The remaining six are nontrivial. Named colours remain distinct in this count; we have not identified assignments obtained by permuting their names.

This is an original step-by-step derivation of the standard result under the stated convention, rather than an experimental count of physical knots. It connects naturally to Linear Algebra: a diagram supplies equations, and their solution set carries information. The crucial interpretation still requires the invariance theorem, not merely successful arithmetic.

The counting convention matters when checking an answer from another learner. If they report one essentially different nontrivial pattern after treating all colour renamings as equivalent, they have answered a different counting question from the nine labelled assignments here. Compare the definitions before deciding that the arithmetic conflicts. Our six nontrivial assignments are the six permutations of 0, 1 and 2 among the three arcs. They all satisfy the same equation, but they remain distinct functions from the labelled arcs to the named colours. This is a small, concrete example of why counting objects and counting equivalence classes produce different totals.

Previous chapter · Contents · Next chapter

The total number of three-colourings is unchanged under Reidemeister moves. Establishing that fact requires checking that valid colourings correspond across each type of move. The local rule is designed to make those checks work; a pattern that happens to look attractive would not suffice.

A crossing-free diagram of the unknot has one continuous arc and therefore three assignments, one for each named colour. Since the count is invariant, every diagram of the unknot has that same total. The trefoil calculation gives nine. Different invariant values establish that the trefoil is not equivalent to the unknot.

Notice the structure of the argument. We calculated convenient diagrams, invoked a theorem about all permitted diagram changes, and then compared the results. The convenient picture is a place to calculate, while the theorem is what allows that result to say something about the knot type.

Louis Kauffman’s introduction to knot theory develops the three-colouring argument and further invariants. Read its local-move discussion when moving from the worked calculation to a proof of invariance.

For a teaching check, ask the learner which step would fail if recolouring were allowed differently at every diagram without a fixed rule. The numerical comparison would lose its common meaning. A valid test needs both a precise calculation procedure and a justification that the procedure measures the property claimed.

Previous chapter · Contents · Next chapter

A successful distinguishing test is still a limited test. Equality of three-colouring counts does not establish equivalence of knots. There are nontrivial knots whose total count is three, matching the unknot.

The ETH Zürich solutions on knot colourings give explicit examples and discuss the behaviour under mirroring. They are useful follow-on material because they show both a method’s power and the cases it leaves unresolved.

In an original reasoning exercise, imagine a valid invariant I with I(P) = 7, I(Q) = 7 and I(R) = 11. The values prove that R differs from P and from Q. They do not tell us whether P and Q are equivalent. No additional arithmetic on those three numbers recovers information that the invariant discarded.

The next step can be a different invariant, a constructive equivalence sequence or a more detailed analysis. The choice depends on the actual knots and the intended question. Repeating the same inconclusive test does not add an independent reason for equivalence.

This is a useful connection to Logic. “Equivalent implies equal invariant” and “equal invariant implies equivalent” are different statements. A learner who can explain that distinction has gained a reasoning tool that applies far beyond topology, including classification systems, diagnostic tests and summaries of complex data.

Previous chapter · Contents · Next chapter

Orientation assigns a direction of travel to each component. Mirroring changes the spatial handedness represented by a diagram; it can be represented by switching every over–under crossing. These are different operations and should have different entries in a working record.

A clockwise arrow on a page is not a complete definition of a knot’s handedness. The shape may be drawn in another position, and the spatial embedding carries information beyond the arrow. State which operation is being performed before interpreting its effect on an invariant.

For an oriented diagram, each crossing has a sign under a fixed convention. The writhe is the sum of those signs. If an exercise supplies five positive crossings and two negative ones, the writhe is 5 − 2 = 3. This arithmetic assumes the signs have already been assigned correctly from the diagram.

Writhe changes under a Type I move, so it is not an invariant of an ordinary unframed knot under all the moves used here. The AMS discussion of knots and writhe explains this distinction and provides illustrated conventions.

A useful checking question is whether a proposed measurement survives the exact equivalence relation in the problem. A quantity may be meaningful for a more restricted kind of diagram change while failing for the broader model. This is why the initial scope statement matters throughout the investigation, not just in the introduction.

Previous chapter · Contents · Next chapter

Trying moves can establish equivalence when a complete valid sequence reaches the target. Failing to find a sequence within a time limit establishes a fact about that search. It does not by itself prove that no sequence exists.

Record the starting diagram, allowed moves, search order, maximum number of steps and any restriction on intermediate complexity. A search that allows only immediate simplifications examines fewer possibilities than one that permits temporary increases in crossing count.

For an original example, a program explores all states within twelve moves of its input under a documented representation and finds no crossing-free diagram. The supported conclusion concerns that explored region. An impossibility claim would require a further theorem or a complete decision procedure with verified assumptions.

Even the claim “all states within twelve moves” needs care. Equivalent encodings, duplicate states and the implementation of local moves must be handled consistently. A bug that omits one permitted move can make the explored set smaller than intended.

Ask a learner to compare two reports: “No solution was found under these limits” and “No solution exists.” Identifying the missing justification is part of the mathematical task. The linked Combinatorics and Discrete Mathematics roots extend the study of finite structures and systematic cases. Knot theory supplies a memorable setting in which a search result must be interpreted at its actual strength.

Previous chapter · Contents · Next chapter

A mathematical loop has no prescribed material stiffness or positive thickness in the classical model used here. A physical cord does. Stretching, bending and moving a real sample therefore involve constraints that an equivalence theorem may deliberately ignore.

Suppose two classroom models represent the same knot type but one is made from thick cord and the other from a flexible narrow strip. A particular rearrangement may be easy for one and awkward for the other. The difficulty does not contradict topological equivalence; the physical task has additional variables.

Keep two records. The topological record specifies components, crossings and permitted strand passage. The material record specifies length, thickness, stiffness, friction and the manipulation attempted. This prevents a physical obstruction from being described as a topological proof.

The broader research connections with physics and molecular structures are introduced in the UCLA course overview. Following those connections requires the specialist model and evidence for the application concerned. The classroom loop provides an analogy and an entry point.

Continue into Mathematical Modelling, Polymer Science or Molecular Biology according to the next question. A knot classification alone does not determine a material’s strength, motion or biological behaviour. The useful transfer is knowing which structural property has been established and which additional mechanisms still require investigation.

Previous chapter · Contents · Next chapter

Begin with a supplied university diagram and trace its components. Explain how the crossing notation affects the trace. Then identify one local Reidemeister move and draw the before and after states. Ask a partner to check the strand connections rather than judging the overall appearance.

Next use the trefoil equations from Chapter 8. Choose a = 1 and b = 2. The equation a + b + c ≡ 0 gives c = 0. Substituting into 2a ≡ b + c gives 2 ≡ 2, while 2b ≡ a + c gives 4 ≡ 1 modulo 3. The third equation also holds, since 0 ≡ 3.

For a counting check, explain why choosing a and b gives exactly one c. This establishes completeness of the nine assignments without listing all twenty-seven triples and hoping none were missed. Then identify the three monochromatic assignments and subtract them to obtain six nontrivial ones.

Finally change the question. If two diagrams each produce nine colourings, what can be concluded? Only that this invariant has not separated them. Ask the learner to propose the evidence needed for a stronger conclusion.

The sequence deliberately changes representation: spatial tracing, local diagram manipulation, modular equations and logical interpretation. A learner who can move between those forms has begun to understand the method. Keep the worked answer beside the reasoning question so a correct number does not conceal a mistaken explanation.

Previous chapter · Contents · Next chapter

Is every complicated loop genuinely knotted?

A complicated diagram can represent the unknot. Complexity in one picture does not establish the knot type. A valid simplification sequence or a suitable invariant can provide more useful evidence.

Is changing one crossing a Reidemeister move?

A crossing switch changes which strand passes over at that crossing. It is not one of the equivalence-preserving moves in this guide, although the resulting knot might happen to be equivalent in a particular case.

Does a larger colouring count mean a more difficult knot?

The count is an invariant with a specific definition. It is not a universal measure of visual complexity, physical difficulty or computational effort. State the property you want to compare before choosing a measurement.

Why use algebra for a spatial subject?

The equations encode constraints from the diagram. When their solutions survive legal moves, they provide a way to distinguish spatial objects through a calculation. The connection depends on the encoding and invariance argument.

Where should a beginner continue?

Use the directory below to revisit topology, practise modular algebra or explore the wider mathematics of invariants. The linked lecture notes supply diagrams for tracing and local-move work. Begin with one complete calculation and explain its limits before adding a more advanced invariant.

The gold standard is a conclusion with a checkable reason. A drawing supplies data, a move supplies a justified transition, and an invariant supplies a test that survives those transitions. The learner’s task is to keep those roles visible as the problem becomes more sophisticated.

Previous chapter · Contents · Explore connected articles

Sources, conventions and exercise notes

Institutional and author resources checked on 22 October 2026. This guide uses ordinary unframed classical knots and links, named colours and arithmetic modulo 3. Follow the source diagrams when assigning arcs or crossing signs.

• Olga Radko, UCLA — Introduction to Knot Theory course and reading routes
• UCLA Lecture 1 — knots, diagrams and Reidemeister moves
• UCLA Lecture 2 — elementary operations and invariants
• UCLA Lecture 3 — three-colourings and linear algebra
• Saul Schleimer, Warwick — Knot Theory lecture notes
• Louis Kauffman — Knot Theory
• Meike Akveld, ETH Zürich — solutions on colouring invariants
• American Mathematical Society — crossing number and classification
• American Mathematical Society — knots and diagram writhe

The calculation tables and teaching records were constructed for this guide. The trefoil count is derived under the stated standard colouring convention. Fictional simplification and search examples demonstrate how to interpret evidence; they do not claim a classification of an unspecified diagram.

Return to reading routes · Education: learning transfers into life