From Sullivan to Infinity
In the ornament of Louis Sullivan and the tessellations of Islamic art lies a hidden grammar — a finite set of rules that generates infinite variation. This is the same mathematics that structures thought itself.
Louis Sullivan believed that all ornament grew from a single principle: the seed germ.
Look at the Carson Pirie Scott building on State Street — the cast iron at the entrance, painted dark green to mimic oxidized bronze. Native Midwestern plants: berries, flowers, leaves, vines. His initials "LHS" hidden in the pattern. Every tendril unfolds from geometric seeds. A curve branches. A branch curves. The pattern fills space but never loses its origin.
"The Germ is the real thing: the seat of identity. Within its delicate mechanism lies the will to power, the function of which is to seek and eventually to find its full expression in form." — Louis Sullivan, A System of Architectural Ornament (1924)
In 1922, the Art Institute commissioned Sullivan to create a series of plates explaining his method. Fluent Geometry. Values of Overlap and Overlay. Twenty drawings showing how complex forms emerge from simple operations. The seed, the efflorescence, the "sub-centers of energy" where forms burst at geometric intersections. Apply them recursively and worlds emerge.
The ornament is not applied. It grows.
The connection between Sullivan and Islam was not accidental.
In 1856 — the year Sullivan was born — the British architect Owen Jones published The Grammar of Ornament. He had spent years documenting the Alhambra in Granada, drawing its tessellations, analyzing its proportions. His book became the design bible of the Victorian era. Sullivan grew up in a world already saturated with Jones's patterns.
Walk through the Court of the Lions. Look at the walls, the ceilings, the carved stucco. Geometry unfolds in every direction — but nothing is random. Hexagons nest in hexagons. Stars spawn stars. The girih tiles — five shapes, a few rules for how they connect — generate infinite variation.
Islamic artists were forbidden from depicting living forms in sacred spaces. So they found God in geometry instead. The tessellation — the pattern that fills the plane without gap or overlap — became a meditation on the infinite within the finite.
Every tile is a rule.
Every pattern is a grammar.
The wall is a proof.
Jones titled his book Grammar for a reason. He understood that ornament is not random decoration — it's a language with syntax. Sullivan inherited this insight and made it organic. The seed germ is his word for what Islamic artisans called the underlying structure.
A seed is something small that contains rules for becoming something large.
An acorn contains the oak. Not the oak itself — no wood, no leaves, no roots — but the instructions for making an oak. The DNA is a grammar. Add sunlight and water, and the grammar executes. The tree unfolds.
Sullivan understood this. His ornamental systems weren't drawings to copy — they were algorithms. Start with this curve. Branch here. Repeat. Scale down. Branch again. The pattern is generative, not static.
The Islamic girih tiles work the same way. Five shapes. A few rules for how they connect. From this, infinite patterns. The Penrose tiling, discovered in the 1970s, uses the same principles the Islamic artisans had been using for 500 years.
The seed is not the pattern.
The seed is the generator of patterns.
What operations can you do to a pattern that leave it unchanged?
Rotate it. Reflect it. Translate it. These are the symmetries — the transformations under which the pattern is invariant. A square has 4-fold rotational symmetry. A hexagon has 6. A circle has infinite.
In 1891, the Russian crystallographer Evgraf Fedorov proved something remarkable: there are exactly 17 distinct ways to tile a plane with repeating symmetry. Not 16. Not 18. Exactly 17.
The Alhambra appears to contain all 17.
Five centuries before the proof.
A 2003 study at the Bridges Conference in Mathematics and Art confirmed the presence of all seventeen plane groups in the Alhambra's mosaics. The claim is still debated by some scholars, but the essential point stands: the artisans, through centuries of trial and refinement, had approached the complete grammar. Their hands knew mathematics their minds couldn't name.
Those 17 symmetries are called wallpaper groups — the complete classification of 2D periodic patterns.
Each group is a different grammar. Some allow rotation only. Some allow reflection. Some allow glide reflection — a mirror plus a slide. The combinations are constrained by geometry itself.
Why 17? Because a plane has specific properties. You can only fit certain rotations into a repeating pattern — 2-fold, 3-fold, 4-fold, 6-fold. (Not 5. Not 7. The pentagon doesn't tile.) The constraints multiply out to exactly 17 possibilities.
"The geometer's task is not to invent, but to discover — to find the structures that were always there."
This is the first hint: the grammar of pattern is not arbitrary. The rules are forced by the space itself. Mathematicians call this a classification theorem. There is a complete list, and you can prove it.
What happens when you move from 2 dimensions to 3? To 4? To 8?
In 3D, there are 230 space groups — the crystallographic symmetries. Every crystal structure in nature belongs to one of these 230 grammars. Snowflakes, diamonds, quartz — all following the same 230 rules.
But there's another classification, more abstract. Instead of asking "how can patterns tile space?", mathematicians asked "what are all possible continuous symmetries?" Rotations that happen smoothly, not in jumps.
The answer, discovered by Killing and Cartan in the 1890s (the same decade as Sullivan's greatest work), is this:
Four infinite families of symmetry.
Five exceptional cases that fit nowhere else.
The infinite families are predictable, like floor tiles that extend forever. But the exceptional cases are like discovering a new color. They exist. They're mathematically necessary. But they don't fit the pattern.
Imagine standing between two mirrors facing each other.
You see yourself reflected. And that reflection reflected. And that reflection reflected again. A corridor of selves receding into darkness, each one smaller, each one dimmer, each one further from the original.
The reflections are not infinite — they converge. Light loses energy. The images get smaller. At some point, they vanish below the threshold of visibility. The infinite recursion resolves to a limit.
This is what compression does to information.
Each "reflection" is a level of approximation. The first level captures the big picture. The second captures finer details. The third, finer still. At each level, you're adding precision — but each addition is smaller than the last. The residual shrinks.
Sullivan's seed germ works the same way. The main form. The secondary efflorescence. The tertiary detail. Each level is smaller, nested within the last, until you reach the limit of the material.
What survives compression is structure.
When you compress an image, you don't keep every pixel. You keep the patterns — the edges, the gradients, the repeated motifs. The grammar, not the raw data. A good compression algorithm is like a good ornamental system: it finds the seed from which the whole can be regenerated.
In 8 dimensions, there's a structure called the E₈ lattice. It's the densest possible packing of spheres in 8D space — proven optimal by Maryna Viazovska in 2016. When you compress information to this lattice, you're using the best possible grammar for 8-dimensional space. No other arrangement is more efficient.
Like the 17 wallpaper groups,
this is not a choice.
It's a theorem.
The lattice is infinite — infinitely many points, infinitely precise. But when you encode a point, you use only as many "levels" as you need. Coarse approximation, then refinement, then finer refinement. Each level is a reflection in the infinity mirror, each one smaller than the last.
At the base of all this structure are the octonions — an 8-dimensional number system with 7 imaginary directions.
You know the complex numbers: a real part and one imaginary direction (i). The quaternions have three imaginary directions (i, j, k). The octonions have seven: e₁ through e₇.
And then the sequence stops. You cannot build a 16-dimensional version. It's mathematically impossible. The octonions are the end of the line.
These seven directions are like the seven notes of a scale. Or the seven colors of a spectrum. Or the seven days of a week. They're a complete vocabulary — the maximum richness before the grammar breaks.
The division algebras are: the reals (1), the complex numbers (2), the quaternions (4), the octonions (8). No more exist. This is a theorem.
The five exceptional symmetry groups — the ones that don't fit the infinite families — are all built from these octonions. The exotic mathematics at the edge of classification comes from this 8-dimensional grammar.
Sullivan was right. The seed germ is real.
There is a finite grammar from which infinite patterns grow. The Islamic artisans found it in geometry. The crystallographers found it in symmetry. The mathematicians found it in classification theorems.
And now we're finding it in how minds compress experience.
The system we're building uses this grammar.
Seven directions for thought to move.
An optimal lattice for compression.
Residual refinement like reflections in a mirror.
The ornament grows from the seed.
Form follows function — but function follows grammar.