The Gold Standard of Visual Arts | Why a Cropped Pattern Can Fit More Than One Repeat Length

Imagine a fictional border divided into equally spaced cells along one straight direction. Each cell contains one of several distinguishable motif states. We write A or B to record those states, retaining orientation and colour only if the chosen classification treats them as relevant.

Our first crop shows six complete consecutive cells: A B A A B A. Their order and spacing are supplied exactly. No cell is missing between them, and the crop does not contain a hidden break or a change of scale.

These conditions form a deliberately simple one-dimensional model. A real artwork can contain continuous shapes, hand variation, damage, perspective distortion and two-dimensional arrangements. The model is useful when its cell record has been justified; it does not establish that every border should be forced into equally spaced letters.

The Visual Arts subject guide connects observation, composition and purposeful experiments. Here the first task is to record what is visible before assigning a rule to what continues outside the image.

Contents · Next chapter

The letter A should stand for a stated visual equivalence class. Does a reflected motif count as the same state? Does changing its colour create a new state? Does a small hand-drawn variation remain inside the class or become another label?

Different choices can produce different apparent periods. If an upright motif and its reflected version alternate, treating both as A can erase the alternating orientation. That does not make either record inherently wrong; it makes their questions different.

Before searching for repetition, write the classification rule and preserve uncertain cells. A colour reading or compressed image may not support a confident distinction. The guide to colour readings and the reflectance spectra they hide explains a separate reason that a matching colour record does not identify every material property.

For this exercise, A and B are exact, visibly distinct states. The arithmetic begins after that stipulation. It does not decide whether two actual motifs should receive the same letter or establish a historical connection between works that happen to produce similar symbol sequences.

A controlled classification change shows the consequence. In the later example A B A C A B A, suppose a coarse record merges B and C into one state X. The row becomes A X A X A X A, which has period two. The original finer record does not, because B and C differ. Merging classes can create a compatible repeat while preserving all equalities that were already present. It cannot establish that the finer visual distinctions never mattered. Keep both records if the distinction is relevant to the artwork: one may describe a broad shape rhythm, while the other describes an orientation or colour sequence. The two records answer different observational questions.

Contents · Previous chapter · Next chapter

For a finite row of n cells, a positive shift p is compatible with the crop when every pair of observed cells p positions apart has the same state. In symbols, cell i must equal cell i + p whenever both positions lie within the crop.

For the six-cell row, testing p = 3 compares positions 1 and 4, 2 and 5, and 3 and 6. All three comparisons agree: A with A, B with B, A with A. Thus three is a period of this finite record.

The test compares only positions actually present. It says nothing directly about position 7 or any later cell. A phrase such as the pattern repeats every three cells can therefore overstate the result if the evidence is just this crop.

Use a precise sentence: the six observed cells are compatible with a three-cell period under the stated classification. That wording preserves the positive finding while leaving the unobserved continuation open. A smaller claim can be more useful because it identifies exactly what another image would need to test.

Contents · Previous chapter · Next chapter

Test p = 1 on A B A A B A. Positions 1 and 2 disagree, so a one-cell period fails. For p = 2, positions 2 and 4 disagree: B is not A. That candidate also fails.

The three-cell candidate passes the comparisons already listed. A four-cell candidate fails because positions 1 and 5 contain A and B. A five-cell candidate compares only positions 1 and 6; both are A, so it passes.

The compatible shifts below the crop length are therefore three and five. A list of repeated-looking chunks might reveal three quickly while overlooking five, because the five-cell test has only one overlapping comparison.

This is a useful lesson about evidence density. Passing a test with one required equality is weaker information than observing several complete repeats, even though both satisfy the formal compatibility definition. Record how many comparisons were available and which cells were compared. A candidate’s numerical size does not by itself tell us how strongly the crop constrains it.

Contents · Previous chapter · Next chapter

The six-cell example can be checked without relying on visual intuition. Keep the positions numbered and record one decisive comparison for each failed candidate.

A failed candidate needs only one contradictory pair. A passing candidate needs every required pair to agree. Checking a few favourable pairs cannot establish the full finite-period condition.

The ledger also preserves the scope: these are shifts compatible with one exact crop. It is not a catalogue of the artist’s intended repeat blocks, and it does not show that the smallest compatible shift is the true fundamental period of the complete border. Those questions require a continuation model and further evidence.

Contents · Previous chapter · Next chapter

Repeat the block A B A indefinitely to form A B A A B A A B A and so on. Its first six cells match the observed crop exactly.

This construction establishes that a complete periodic design with period three is compatible with the crop. Because the block is not constant and the smaller shifts one and two fail within the visible cells, three is also the smallest period of this constructed infinite sequence.

The reasoning moves in two stages. The crop passes a finite compatibility test. Then an explicit continuation demonstrates that one possible whole design has the proposed period. Neither stage establishes that the actual unseen border follows that continuation.

For a learner, draw the next three cells under this hypothesis and label them predicted, not observed. If a later photograph shows something else, revise the hypothesis while preserving the original crop. A useful reconstruction keeps its evidence and additions distinguishable.

The repeat block also has a phase relative to the crop. In the three-cell continuation, choosing the first observed cell as a block boundary writes ABA. Moving the notional boundary one cell writes BAA; moving it again writes AAB. These cyclic block descriptions can represent the same continuing border with different choices of origin. The crop edge is an image boundary, not evidence that the artist began a repeat unit there. Record where a proposed block starts and distinguish that convention from its length. If two reconstructions differ only by this chosen boundary while predicting the same cells everywhere, they are not the same kind of competing hypothesis as the three-cell and five-cell continuations that disagree at position seven. The clearer the predicted continuation, the easier it becomes to identify a new observation that would genuinely challenge it.

Contents · Previous chapter · Next chapter

Now repeat the block A B A A B. Its sequence begins A B A A B A B A A B. The first six cells are again A B A A B A, matching the same crop.

This continuation differs from the three-cell one at position 7: the three-cell model predicts A, while the five-cell model predicts B. Both explanations fit the current six observations, yet they make different predictions outside them.

The five-cell block is nonconstant and has prime length five. A complete infinite repetition of it cannot have a smaller fundamental period that also divides five unless that period is one; the visible nonconstant states exclude one. It therefore supplies a genuinely different fundamental-period construction.

The educational point is underdetermination. The same finite evidence can support multiple complete patterns. Choosing a short description can be useful for a design task, but it is not equivalent to discovering the original construction. State whether the goal is to invent a coherent continuation, compress the observed record or infer what was actually present outside the crop.

Contents · Previous chapter · Next chapter

Any multiple of a complete pattern’s period is also a period. A design repeating every three cells also repeats after six, nine and further multiples of three. Those larger shifts do not become its smallest repeat.

The six-cell block A B A A B A can be repeated as a block, but the resulting infinite sequence still has fundamental period three. Describing it as a six-cell repeat is compatible with one block choice; describing six as its smallest period would be wrong for that construction.

This distinction connects to the existing guide on textile patterns, symmetry transformations and repeats. That article explains primitive repeats, repeat vectors, colour and geometric transformations. Here we isolate what a finite crop can identify about a whole sequence.

When reporting a result, specify whether the number is a compatible shift, a chosen repeat-cell size or a certified smallest global period. The same number can appear in all three discussions while carrying a different claim in each.

Contents · Previous chapter · Next chapter

If a proposed period p is at least the crop length n, there are no observed pairs p cells apart. The finite compatibility condition is then vacuously satisfied: there is no required comparison within the crop to contradict it.

That does not mean every such number is established as the smallest global period. It means the crop alone cannot reject it merely through overlapping-cell comparisons. A compatible periodic extension and its fundamental period remain separate matters.

For the first crop, a seven-cell block A B A A B A C, with C distinct from A and B, gives a complete continuation matching all six observations. The single C in each repeated block prevents a smaller global repeat: its positions recur exactly seven cells apart.

More generally, a sufficiently long block can contain the visible prefix and one unique C outside it, producing a long-period continuation. The original crop does not reveal that unseen marker. This explicit construction shows why an upper limit on the true repeat length cannot be inferred from a short matching fragment without additional constraints.

Contents · Previous chapter · Next chapter

A period test can be drawn as a graph on observed positions. Join positions whose states must match under a candidate shift. If several shifts are proposed simultaneously, include the edges required by all of them.

For the six-cell crop and shifts three and five, the three-step edges join 1–4, 2–5 and 3–6. The five-step edge joins 1–6. Following these edges makes positions 1, 3, 4 and 6 one equality group, while positions 2 and 5 form another.

Assign A to the first group and B to the second. The resulting sequence is exactly A B A A B A. The two different groups explain how both candidate shifts can coexist without forcing every cell to be the same.

The graph is a representation of equality constraints, not a model of aesthetic preference. It helps a learner check transitive implications that are easy to miss when comparing isolated pairs. If one connected group contains two confidently different observed states, the proposed combination of periods is incompatible with the record.

Contents · Previous chapter · Next chapter

Extend the same six-cell prefix by one observed cell. If shifts three and five are both still required, shift three connects position 4 to 7, while shift five connects position 2 to 7.

The existing record has A at position 4 and B at position 2. The seventh cell would therefore need to be both A and B. With those states distinct, no seventh observation can preserve both period claims.

Under the three-cell continuation, position 7 is A and the five-cell claim fails. Under the five-cell continuation, position 7 is B and the three-cell claim fails. A different state would reject both of these particular hypotheses.

This is stronger than saying that a longer crop is generally useful. It identifies a precise observation at which two surviving explanations disagree. The learner can choose the next evidence deliberately. In an actual artwork, collecting that evidence still depends on access, image coverage and whether the classification and cell-spacing assumptions remain valid across the newly observed region.

Contents · Previous chapter · Next chapter

A general result known as the Fine–Wilf periodicity theorem gives a sufficient overlap length. If a finite sequence has periods p and q and its length is at least p + q − gcd(p, q), it also has period gcd(p, q).

For three and five, that threshold is 3 + 5 − 1 = seven. A seven-cell sequence with both periods must have period one, so it must be constant under the exact symbol model. Our nonconstant six-cell example sits just below that threshold.

Jeffrey Shallit’s university-hosted Fifty Years of Fine and Wilf presents the mathematical context and the 1965 result of Fine and Wilf. The labelled border constructions and direct equality checks here are original teaching examples.

Do not read the theorem backwards as a promise that one more cell always identifies the actual global design. It constrains simultaneous finite-period claims. A different, longer global period can still match the enlarged crop unless further evidence or a justified candidate limit excludes it.

Contents · Previous chapter · Next chapter

The period theorem concerns exact equality along a consecutive finite sequence. It does not directly apply to a row with missing cells treated as though they were absent positions, or to approximate photographic similarities without a stated tolerance model.

The symbols also need a consistent equivalence rule. If A and B are later merged into one broad motif class, a previously nonconstant record can become constant. The changed conclusion would reflect a changed classification, not a contradiction in the mathematics.

Likewise, the number of visible repetitions and the number of sampled cells are different quantities. Counting one large motif as a cell in one image and two smaller parts as cells in another changes the index scale on which p and q are measured.

A useful evidence note states cell definition, sequence length, candidate shifts, equality criterion and any missing or uncertain positions. These conditions should appear before a global claim is made. A theorem can strengthen an argument only when the observed record actually matches the mathematical object to which the theorem applies.

Contents · Previous chapter · Next chapter

Suppose the third cell of the six-position border is damaged. The recorded positions are A, B, unknown, A, B, A. The physical position of the unknown cell remains part of the evidence.

Deleting that position would produce A B A B A, a different five-cell sequence compatible with period two. In the original indexed record, period two fails because positions 2 and 4 contain B and A. Removing the gap has created a false alignment.

Under period three, the unknown third cell would need to match position 6, which is A. That is a conditional reconstruction. It becomes an observed A only if independent evidence establishes its state.

For a practical record, retain missing positions and distinguish confirmed states from values predicted by a model. A damaged motif does not shorten the distance between the neighbouring cells. This is a useful transfer principle for cropped photographs, conservation records and design documentation: preserving the location of missing information can be as important as preserving the information that remains visible.

Contents · Previous chapter · Next chapter

Suppose a faint cell might be A or B. A period test should not quietly choose whichever label makes the preferred repeat work. Keep both admissible classifications until the evidence resolves them.

For the first crop, uncertainty at a position involved in a contradictory pair can change which shifts are excluded. A better photograph or a clearer account of orientation may therefore be more useful than a longer sequence whose states remain ambiguous.

The graph representation helps. Mark uncertain vertices and ask whether a proposed equality group contains a confirmed contradiction, a compatible assignment or an unresolved classification. These are three different outcomes.

Do not attach numerical probabilities to the alternatives unless the evidence supports such a model. Saying that two labels remain possible does not make them equally likely. The simple exact-symbol examples here demonstrate logical compatibility. Actual image uncertainty may need a documented classification procedure and suitable review. The conclusion should reflect that procedure rather than the confidence created by a neat string of letters on the page.

Contents · Previous chapter · Next chapter

A shift of three cells is an index distance. Converting it into centimetres requires a justified cell spacing and image scale. A photograph alone may not provide either accurately.

If an original model supplies equal cell spacing of 2 centimetres along the border, a three-cell translation corresponds to 6 centimetres. If the scale is unknown, the index result remains meaningful while the physical distance is unresolved.

Perspective can also change apparent spacing across an image. Equal pixel gaps are not automatically equal physical distances, and unequal pixel gaps do not automatically prove that the original border lacks regular spacing.

The guide to recovering composition from detail images at different scales develops the neighbouring reconstruction problem. Keep its scale and overlap questions separate from our exact symbol-period question. A valid period in a carefully indexed row does not establish an object’s physical size, camera geometry or the complete arrangement outside the photographed region.

Contents · Previous chapter · Next chapter

A row can repeat regularly while the arrangement of rows changes. The first row of a textile might match the next after a horizontal shift, a reflection or a colour change. A one-row crop cannot determine all those two-dimensional relationships.

To infer a plane repeat, record enough rows and identify how cells correspond across them. The existing textile-pattern guide explains repeat vectors and cell choices. Our one-dimensional tests do not replace that geometry.

Similarly, a border can contain a regular local section inside a larger nonperiodic design. Passing the finite test in that section establishes compatibility there, not global periodicity of the whole artwork.

State the observation region and the claim’s reach. A report can say that one row supports a candidate shift while leaving row-to-row structure unresolved. That is a useful finding. Turning it into a complete reconstruction without additional evidence would hide the actual reason for confidence. The artistic decision may still be to invent a continuation, but it should be described as a new design choice rather than a recovered fact.

Contents · Previous chapter · Next chapter

The graph used here joins positions that must share a motif state. It does not describe which shape lies in front of another. Graphs with the same appearance can represent different relationships.

The guide to overlap cues and one global layer order studies a different kind of constraint. Its directed ordering relationships cannot be substituted for the equality edges in a period test.

Likewise, fading a group versus fading its overlapping shapes concerns compositing contributions. A partially visible motif may prompt that separate investigation before a confident cell label is assigned.

These connections help a learner choose the next route without collapsing distinct questions. First identify whether the uncertainty concerns motif state, sequence position, physical scale, occlusion or digital composition. Each needs evidence suited to its mechanism. A general impression that the pattern looks wrong gives less direction than locating the first uncertain relationship.

Contents · Previous chapter · Next chapter

A new original crop contains A B A C A B A, with A, B and C distinct. Test candidate periods four and six. Then ask whether the crop has period two, the greatest common divisor of four and six.

For period four, compare positions 1–5, 2–6 and 3–7. They agree. For period six, compare positions 1 and 7; they also agree. Period two fails because positions 2 and 4 contain B and C.

The crop length is seven. The sufficient Fine–Wilf overlap length for four and six is 4 + 6 − 2 = eight, so the non-two-periodic crop does not contradict the theorem. It falls one cell short of the required length.

This independent example changes the candidate shifts, their common divisor and the number of motif states. A learner who can explain the comparisons has transferred the method beyond the first six-cell illustration. Merely computing the greatest common divisor without checking the length condition would produce an unsupported conclusion.

Contents · Previous chapter · Next chapter

Repeat the four-cell block A B A C. Its first eight cells are A B A C A B A C. The eighth cell is C.

Repeat the six-cell block A B A C A B. Its first eight cells are A B A C A B A B. The eighth cell is B. Both continuations match the seven observed cells, then disagree at the next position.

If both period claims were required at length eight, period four would force position 8 to match position 4, which is C. Period six would force it to match position 2, which is B. Since those labels are distinct, no eighth cell can preserve both claims for this prefix.

Again, this identifies a discriminating observation between two hypotheses. It does not establish that either hypothesis is the real global design. A longer repeat with another unseen state remains possible. Keep the list of tested candidates separate from the claim that the list contains every possible model of the artwork.

Contents · Previous chapter · Next chapter

For a paper-based investigation, create the first crop on six cell cards. Ask two learners independently to propose a periodic continuation and record the rule. Compare predictions before revealing a seventh card.

Next, use the changed seven-cell crop and require a comparison ledger. Give one learner the four-cell hypothesis and another the six-cell hypothesis. Ask them to identify the first position at which their predictions differ.

Finally, introduce one missing or uncertain cell without changing the physical index positions. Require the learners to revise their conclusions while retaining the original observation record. This tests whether they preserve evidence rather than silently repairing it to suit a favoured pattern.

The task can remain entirely synthetic. No historical artwork needs to be reconstructed, and no cultural attribution follows from the letter sequence. The successful learner can describe the observed crop, the candidate rule, the predicted extension and the precise observation that would challenge it. That explanation is more informative than a page filled with an attractive repeated motif but no account of why it was chosen.

Contents · Previous chapter · Next chapter

Consider the sentence: “The crop contains ABA twice, so the original border’s smallest repeat was definitely three.” The visible repetition supports one compatible continuation. It does not exclude the five-cell or longer constructions already shown.

Another claim says: “Five is a period because the first and last cells match, so the whole border repeats every five.” The finite test has only one required comparison at that shift. It does not verify all positions outside the crop.

A third says: “The two candidate shifts have greatest common divisor one, so every visible cell must be identical.” That conclusion needs the appropriate finite-length condition. The six-cell counterexample shows the condition cannot be omitted.

Repair each explanation by naming the observed region and the exact test passed. Then distinguish a compatible hypothesis from a recovered global fact. The aim is not to make pattern recognition hesitant. It is to make the reasoning inspectable enough that a later image can confirm, reject or refine the proposed continuation without rewriting what was originally seen.

Contents · Previous chapter · Next chapter

For a repair route, begin with small numbered cell cards and one candidate shift. Point to every pair that must agree. Use a contradictory pair to reject a candidate rather than relying on whether the row feels repetitive.

For a stabilisation route, test several shifts and construct at least two complete continuations for one crop. Separate passing a finite test from certifying a fundamental period. Retain predicted cells in a different notation from observed cells.

For an extension route, draw equality graphs for two proposed periods. Explore when the graph joins all positions or splits into several classes. Verify the changed four-and-six example, then investigate another pair with its length condition stated explicitly.

The artistic route can use the same evidence differently. A designer may deliberately choose the shortest compatible repeat for economy, or a longer one for surprise. That choice is legitimate when described as a creative decision. Learning transfers when the student can explain both the mathematical possibilities and the purpose behind the selected design.

Contents · Previous chapter · Next chapter

Does the shortest fitting repeat identify the artist's original repeat?

No. It identifies a short compatible model under the stated cell record. The original whole design can contain distinctions outside the crop.

Can a larger repeat also be valid?

Yes. Multiples of a true period are also periods, and a finite crop can fit different continuations with different fundamental periods. Keep those two reasons distinct.

Why preserve a missing cell's position?

Deleting it changes which observed cells are compared at a given shift. That can create a period the original indexed record does not support.

Does a one-row result establish a wallpaper pattern?

No. Row-to-row relationships and two-dimensional repeat vectors need further evidence. A local repeating section also need not make the whole design periodic.

Does mathematical similarity establish common cultural origin?

No. A repeated symbol record does not establish provenance, influence, meaning or authorship. Those questions require their own historical and contextual evidence.

Contents · Previous chapter · Next chapter

Return to the Visual Arts subject guide for observation, intention and revision. For the larger geometry of repeating designs, read textile patterns, symmetry transformations and repeats.

For uncertain image evidence, choose the route matching the problem: detail images at different scales, colour readings and hidden spectra, global layer order from overlap cues or group and individual fading.

The wider routes are How X Works and The Gold Standard of Education: learning transfers into life. The practical transfer is to preserve the difference between a visible fragment and the larger structure a reconstruction proposes.

Contents · Previous chapter · Next chapter

The specialist mathematical source is Jeffrey Shallit’s University of Waterloo presentation Fifty Years of Fine and Wilf, checked on 23 October 2026. It presents the finite-period context and identifies Fine and Wilf’s 1965 paper, Uniqueness Theorems for Periodic Functions. Use the linked exposition for the theorem’s stated hypotheses, then distinguish that general result from the particular cells and observations in the original exercises here.

The letter crops, continuations, missing-cell example and direct equality arguments are original educational constructions. They do not describe a particular artwork or certify a reconstruction, attribution or material history.

Before accepting a repeat claim, check the cell definition, observed positions, candidate shift comparisons and scope of the proposed continuation. If two explanations fit, state what they predict differently. A useful next observation is one that can distinguish surviving possibilities, rather than another image chosen only because it resembles the preferred reconstruction.

Contents · Previous chapter · Return to Visual Arts