Why a kolam.
Design systems borrow visual traditions all the time, and almost always they borrow the surface — a palette, a motif, a texture. This one borrows a grammar. That distinction is the whole argument for the choice, and it only holds because the kolam is one of the few folk traditions that is genuinely, formally rule-governed rather than merely patterned.
Why a kolam
Four properties, each of which turned into a load-bearing part of the system rather than a decorative reference.
1 · It is a rule set, not a style
A kolam is specified before it is drawn: a dot count, a grid shape, a rule about how the line may move. The composition follows from the specification. That is the same relationship a design token has to a component, which is why the metaphor survives contact with implementation instead of stopping at the mood board.
→ became principle 01 and the entire lattice.
2 · It has a correctness condition
This is the rare part. Most visual traditions can be done well or badly but not wrongly. A kolam can: in the looped form, if the line breaks, the drawing has failed. A tradition with a pass/fail condition can be tested, and a design system that can be tested is a different kind of object from one that can only be reviewed.
→ became principle 02 and the generator's zero-open-endpoints guarantee.
3 · It is generative
One grid produces unlimited valid, distinct patterns. An identity built on it does not need an asset library — it needs a function and a seed. Very few visual traditions can be compressed to five numbers without losing what makes them recognisable.
→ became the generator.
4 · It is meant to be erased
Drawn at dawn, walked over all day, swept away, drawn again. A tradition whose value is in the drawing rather than the artefact has something useful to say to an industry whose systems mostly die of accretion — nothing is ever removed, so eventually nobody can tell which version is current.
→ became principle 04 and the deprecation policy.
The honest part
A rationale that only lists advantages is a sales pitch. Three things should be said plainly.
- The mathematics came after the practice, and mostly from outside it. Kolam artists were not applying graph theory. Researchers recognised structure in what was already being done — which is a real intellectual contribution, and also not the same as the tradition being "secretly mathematical". The order matters.
- Choosing a tradition with formal structure is partly self-serving. Kolam suits a design system unusually well, which makes it easy to mistake convenience for respect. The cultural handling rules exist because that mistake is the default.
- “Kolam” here means the looped form. The word covers a whole family — including pulli kolams drawn as separate strokes between dots, which are not one continuous line. Everything in this documentation says simply kolam, and everywhere it makes a claim about continuity it means the looped, single-line form the generator draws. Naming that once is more readable than qualifying it forty times, but it is a simplification and this is where it is declared.
- This system captures a narrow slice. Square kolams on odd grids. Not padi kolams, not freehand, not the diamond and looped-mesh forms, not any of the coloured festival work. What is documented here is one construction, described precisely — not the tradition, described adequately.
How it changed
The chronology is genuinely contested, and the disagreement is more interesting than either answer.
Sangam-era references
Some writers trace floor-decoration practices in the Tamil region to Sangam literature. The references are indirect and describe decorated ground rather than the specific dot-lattice construction, so they establish an ancestry for floor art rather than for the kolam as practised now.
Āṇṭāḷ
The poems of the Vaishnava saint-poet Āṇṭāḷ are commonly cited as the earliest clear literary reference to the practice. This is the point most accounts treat as firm ground.
The word itself
Gift Siromoney argued that common threshold patterns are not very ancient — that the practice of decorating the floor in this way may go back about six hundred years and not more. He noted that the word kolam, used for patterns drawn on the floor, appears for the first time in the 16th-century Kuravanji Madurai Meenatchiammai Kuram, and a little later in the 17th-century Kutrala Kuravanji.
Formalisation
Gift and Rani Siromoney, at Madras Christian College, treat kolam designs as two-dimensional picture languages with definable syntactic rules, and show that many complex patterns are generable by context-free array grammars. The model becomes known as the Siromoney matrix grammar. This is the moment the tradition acquires a formal description — and the moment this design system becomes possible.
Ethnomathematics and computation
Marcia Ascher's work places kolam alongside other non-Western mathematical practices. Prusinkiewicz and Hanan apply L-systems to South Indian folk patterns. Naranan and collaborators construct kolams from Fibonacci and generalised Fibonacci sequences. Yanagisawa and Nagata contribute numerical and diagrammatic analyses and digitalisation studies.
Flour to vinyl
Rice flour — traditionally also a small daily act of feeding ants and birds — gives way in many households to stone powder, chalk and synthetic colour, which last longer and feed nothing. Daily ephemeral drawing gives way in urban apartments to painted permanent designs, printed stickers and moulded plastic mats. Competitions and Margazhi-month displays professionalise a domestic practice. Software, including this, digitises it.
What changed
- Material — rice flour to stone powder, chalk, acrylic, vinyl.
- Permanence — daily and swept to painted and kept.
- Setting — the earth threshold of a house to a tiled apartment landing, a competition floor, a screen.
- Who draws — a near-universal daily practice among women in many households, now unevenly transmitted and often performed rather than lived.
- Why — protective and auspicious ritual, increasingly also aesthetic and civic display.
What did not
- The dot grid comes first. Every construction still begins by laying the pulli.
- The continuity rule for kolams. One line, no lifting, closed.
- Symmetry as the default and departure from it as an event.
- The threshold as the site. Whatever the material, it is drawn where inside meets outside.
The four invariants are exactly the four this system took. That is not a coincidence — they are what survives medium change, which is a decent test of what is structural.
The mathematics
Five distinct bodies of work, of which this system uses two.
Mirror curves and the dot lattice
The standard formalisation treats a kolam as a mirror curve: a line travelling diagonally through a lattice of dots, reflecting off the boundary and off "mirrors" placed between dots. The curve is not drawn freely — it is the deterministic consequence of where the mirrors are.
This is the family the generator belongs to. Its three-tile decomposition is the same construction expressed per-cell rather than as a travelling particle: a tile is a statement about which way the line turns in that square, and the arcs meet at cell-edge midpoints because the loop radius is exactly half the pitch. Same geometry, different bookkeeping.
A mirror curve with no crossings — every turn is a quarter arc around a dot.
One line or several — the gcd result
The question a kolam poses is whether the whole figure is a single closed curve or several disjoint ones. For the classical constructions this is not an aesthetic outcome; it is arithmetic.
When the grid parameters are coprime, the traversal closes into a single Eulerian circuit — one line, one loop, the ideal the form is named for. When the greatest common divisor of the parameters is greater than one, the same construction yields multiple disjoint closed loops, and the count is governed by that divisor. A practitioner choosing a dot count is choosing, whether or not in these terms, how many separate lines the finished drawing will contain.
Formal grammars
The Siromoneys' contribution was to treat kolam as a two-dimensional picture language. Where a one-dimensional grammar generates strings, an array or matrix grammar generates two-dimensional arrays of symbols — and many complex kolam patterns turn out to be generable by context-free array grammars, with new patterns produced by simple transformations: rotation, reflection, half-turn, conjugation.
The claim being made is precise and stronger than "kolams are mathematical". It is that the set of kolams is a formal language — that there exist production rules whose closure is the set of valid patterns. That is the intellectual result this entire system rests on, and it is worth naming its authors every time.
Two more, not used here
L-systems
Prusinkiewicz and Hanan applied Lindenmayer systems — parallel rewriting rules originally devised for plant growth — to South Indian folk patterns. L-systems capture the recursive character of kolam growth, where a larger pattern contains transformed copies of a smaller one, better than a per-cell tiling does.
Fibonacci constructions
Naranan and collaborators built square and rectangular kolams of arbitrary size from Fibonacci and generalised Fibonacci numbers. A pleasing result, and a caution: that kolams can be constructed this way is not evidence that they historically were.
What the system actually uses
| Mathematical idea | Used here? | Where |
|---|---|---|
| Mirror curves on a dot lattice | yes | The three-tile decomposition in the generator |
| Point symmetry groups (D4, D2) | yes | Tile-grid constraints, and the layout vocabulary in principle 03 |
| Closure / no open endpoints | yes | The boundary-closure proof |
| Eulerian single-circuit condition | no | Would over-constrain tile choice — closure is guaranteed, single-line is not |
| Array / matrix grammars | no | The intellectual foundation, but not the implementation |
| L-systems | no | — |
| Fibonacci constructions | no | — |
Two of seven. A system that claimed all of them would be more impressive and less true.
Elsewhere in the world
Threshold marks, endless lines and protective patterns recur across unconnected cultures. Some of the parallels are loose. Two of them are startlingly exact.
Across South Asia
The same domestic practice carries a different name and idiom in nearly every region — closely related traditions rather than distant parallels.
| Name | Region | Character |
|---|---|---|
| kolam | Tamil Nadu | Dot lattice; the looped, continuous-line form is the one this system uses |
| muggu | Andhra Pradesh, Telangana | Closely related dot-grid tradition |
| rangoli | Maharashtra, and the general term | Broader, often filled and coloured |
| mandana | Rajasthan | Painted on walls as well as floors, red ground with white |
| alpana / alpona | West Bengal | Rice paste, flowing freehand rather than lattice-bound |
| aripan | Mithila, Bihar | Finger-drawn geometric patterns in rice paste (pithar) |
| chowk / chowkpurana | Uttar Pradesh | Ritual squares |
| jhoti / chita | Odisha | Rice-paste threshold and wall drawing |
| kalamezhuthu | Kerala | Coloured powder, ritual figures rather than lattice patterns |
| sathiya | Gujarat | Threshold auspicious marks |
The two exact parallels
These are the ones worth taking seriously, because the correspondence is structural rather than thematic — the same mathematical constraints, arrived at independently.
sona · Chokwe, Angola
Sona (sing. lusona) are sand drawings of the Chokwe and neighbouring peoples of eastern Angola, north-west Zambia and parts of Congo, used in storytelling and initiation. They are mirror curves on a dot lattice — the same formal object as a kolam.
Paulus Gerdes documented the tradition and reconstructed its algorithms. The reported Chokwe aesthetic preference is for curves that (1) separate each dot from the others, (2) form a single monolineal Eulerian circuit, and (3) are symmetric.
sandroing · Vanuatu
Practised across some eighty language groups of central and northern Vanuatu, and inscribed on UNESCO's Representative List of the Intangible Cultural Heritage of Humanity in 2008 (proclaimed 2003). With one finger, the drawer traces a continuous meandering line on an imagined grid in sand, volcanic ash or clay.
The drawings are mnemonic as well as graphic — encoding ritual, mythology, kinship, song cycles, farming and craft knowledge. Marcia Ascher analysed their mathematical properties in the 1980s.
Looser parallels
Related by theme — endlessness, interlace, protection at a boundary — rather than by construction.
Celtic, Saxon, Norse interlace
Endless knotwork, widely read as apotropaic — protecting the wearer or owner from harm. Continuous and interlaced, but built from strapwork rather than generated on a dot lattice.
Girih
Interlaced strapwork in Islamic architecture — the term is from the Persian for knot — reaching complex tilings by the 11th century. Deeply mathematical, but the mathematics is tiling and symmetry rather than single-curve traversal.
Apotropaic threshold marks
In English vernacular buildings, hexafoils or "daisy wheels", mazes and diagonal lines were scribed near doors, windows and hearths — the vulnerable points — to catch or turn away evil. Same site, same intent, no continuous line.
What the recurrence does and does not mean
Three elements recur widely and independently: a threshold, a line without an end, and an ordering or protective intent. Where a tradition draws a single continuous line through a field of dots, it converges on the same mathematics — because there are only so many ways to do it.
That these are "the same tradition", or that one derived from another, or that they share a meaning. Sona encode narrative; Vanuatu sand drawings encode kinship and song; kolam is auspicious and domestic; a daisy wheel scratched by a Suffolk carpenter is defensive. Convergent structure is not shared meaning, and flattening them into a universal symbol erases the specific thing each one is.
What this system took
Stated plainly, so an adopter can judge whether the borrowing is proportionate.
Taken
- The dot lattice as a generating function for spacing and geometry.
- The meeting condition
r = P/2, which is the tradition's geometry, not an invention. - The continuity requirement, as both a pattern guarantee and a product principle.
- Symmetry as default, asymmetry as signal.
- Impermanence, as a release and deprecation model.
- Vocabulary — the Tamil terms, used with glosses throughout.
Deliberately not taken
- Ritual configurations. The generator produces secular looped patterns; a denylist of festival- and mourning-specific forms ships with it.
- Religious meaning. No claim is made that a generated pattern is auspicious, protective, or anything other than a figure.
- The claim of authority. This is one reading by a design system, not documentation of a living practice.
- Coloured festival forms, freehand traditions, and every regional variant listed above.
Sources
Where the claims on this page come from. Where accounts disagree, both are cited.
- Siromoney, G. & Siromoney, R. — “Array Grammars and Kolam”. The formalisation of kolam as a two-dimensional picture language generable by context-free array grammars. Semantic Scholar
- “Kambi Kolam as Algorithmic Pattern”, Algorithmic Pattern Salon. Context on the Siromoneys' 1970s work and the treatment of kolam as syntactic picture language. alpaca.pubpub.org
- Sahapedia — “Significance of Kolam in Tamil Culture”. Source for the contested chronology, including Siromoney's six-hundred-year reading and the 16th-century Madurai Meenatchiammai Kuram attestation alongside the Sangam and Āṇṭāḷ claims. sahapedia.org
- Gerdes, P. — Sona Geometry from Angola: Mathematics of an African Tradition (originally published in Portuguese, 1993). Documentation and algorithmic reconstruction of Chokwe sand drawings. Overview
- Chavey, D. — “Mathematical Experiments with African Sona Designs”, Bridges 2009. Source for mirror curves, common divisors, Eulerian cycles and the stated Chokwe aesthetic preferences. bridgesmathart.org
- UNESCO — “Vanuatu sand drawings”. Inscribed 2008 on the Representative List; source for the continuous line on an imagined grid, the ~80 language groups, and the mnemonic function. ich.unesco.org
- “Extending Hridaya Kolam to Multiple Loops: A Study of Non-Coprime Dot–Arm Structures”, arXiv. Source for the gcd result — where the divisor exceeds one, the construction yields multiple disjoint closed loops rather than a single Eulerian circuit. arxiv.org
- “Kolam Designs Based on Fibonacci Numbers” (Naranan et al.). Nieuw Archief voor Wiskunde
- Yanagisawa & Nagata — “Fundamental Study on Design System of Kolam Pattern”. Numerical and diagrammatic analysis, and digitalisation of kolam construction. ResearchGate
- Britannica — “Rangoli”, and Memeraki, “Floor Paintings of India”. Sources for the regional names table. britannica.com · memeraki.com
- Historic England — “What Are Witches' Marks?” Source for apotropaic threshold marks at doors, windows and hearths. historicengland.org.uk