Generating a Phyllotaxis Pattern by Hand: What the Numbers Mean

The golden spiral & phyllotaxis calculator on this site turns one number — how many seeds to place — into a full set of coordinates, using the exact same rule sunflowers, pinecones, and daisy heads follow. The rule itself is short enough to write on an index card, and walking through it by hand, point by point, makes it far more concrete than reading the formula in the abstract. None of it requires anything more advanced than a square root, a modulo operation, and basic trigonometry — the elegance is entirely in what the rule produces, not in any hidden complexity behind it.

The recipe in full

Every point in the pattern is generated the same way. For seed index i, starting from 0: the radius is scale × √i, where scale is a spacing constant you choose. The angle is i × 137.5077640500° (the golden angle), taken modulo 360° so it stays within one full turn. Converting that radius-and-angle pair into ordinary x/y coordinates uses standard trigonometry: x = r·cos(θ), y = r·sin(θ), with θ converted from degrees to radians first. That's the entire algorithm — one square root, one multiplication, one modulo, and two trigonometric calls per point, repeated as many times as there are seeds.

The first seven points, worked through

Running that recipe with scale=1 produces, for i=0 through 6: seed 0 sits at the origin (radius 0, undefined angle, so it's just placed at x=0, y=0). Seed 1: radius=√1=1, angle=137.507764°, giving (x,y) = (−0.7374, 0.6755). Seed 2: radius=√2≈1.4142, angle=(2×137.507764) mod 360 = 275.015528°, giving (0.1236, −1.4088). Seed 3: radius=√3≈1.7321, angle=52.523292° (after wrapping past 360° once), giving (1.0538, 1.3746). Seed 4: radius=2, angle=190.031056°, giving (−1.9694, −0.3484). Seed 5: radius≈2.2361, angle=327.538820°, giving (1.8867, −1.2002). Seed 6: radius≈2.4495, angle=105.046584°, giving (−0.6359, 2.3655). Every one of those figures is the calculator's own live output, not a separately typed table — plotting just these seven points already starts to suggest the outward-spiraling arrangement that becomes unmistakable at higher seed counts.

What the scale parameter actually controls

Scale is a pure multiplier on every radius, with no effect on any angle: doubling scale from 1 to 2 doubles every single radius in the pattern while leaving all the angular relationships between points completely unchanged, which is exactly what you'd want from a parameter meant to control physical size (pixels, millimeters, whatever unit a specific use calls for) rather than the pattern's underlying structure. Multiplying scale by 10 turns seed 1's coordinates from (−0.7374, 0.6755) into (−7.374, 6.755) — the same direction, ten times farther out. Because every radius scales by the identical factor, the pattern's overall shape, spiral-arm counts, and relative point spacing are entirely determined by the golden-angle rule and the square-root growth; scale only ever decides how large the final result renders.

Why the radius grows as a square root, specifically

The square-root choice isn't arbitrary, and it's worth understanding why rather than treating it as a fixed part of the recipe. The goal is for each new seed to occupy, on average, the same amount of planar area as every seed before it, rather than the same radial distance. The area of a disk of radius r is πr², so if the cumulative area used by the first i seeds is meant to grow in direct proportion to i (one seed, one unit of area, consistently), then r² must be proportional to i, which means r itself must be proportional to √i. Checking this directly against the calculator's own output confirms it: the area enclosed by the outermost seed, divided by the seed count, settles toward a single constant value as the count grows — roughly 2.51 at 5 seeds, 2.99 at 21 seeds, 3.09 at 55 seeds, and 3.12 at 144 seeds, closing in on exactly π as the seed count increases. That convergence to a single constant is the whole point: it means the pattern really does distribute area evenly per seed, in the limit, rather than merely looking evenly spaced.

From polar coordinates to something you can actually plot

Radius-and-angle (polar) coordinates are the natural language for describing a rotate-and-grow rule, but almost every drawing surface — a canvas element, an SVG document, a plotting library — expects x/y (Cartesian) coordinates instead, which is why the conversion step matters in practice, not just in the formula. The conversion itself is ordinary trigonometry, but it's easy to trip on the units: the angle has to be converted from degrees to radians (multiply by π/180) before it's handed to cos() or sin() in virtually any programming environment, since those functions universally expect radians. Skipping that conversion, or applying it in the wrong direction, is a common, entirely avoidable source of a phyllotaxis pattern that renders as visual nonsense instead of a spiral — worth checking first if a from-scratch implementation doesn't produce the expected shape.

A note on why the precision differs between columns

Looking closely at the calculator's output, the radius and angle figures carry six decimal places while the final x and y coordinates carry only four — a deliberate choice, not an inconsistency. Radius and angle are intermediate values that get fed into cos() and sin() before producing the final coordinates, so rounding them too aggressively before that conversion would let a small error compound through the trigonometry and show up amplified in the final position. Keeping extra precision on the inputs to a calculation, and only rounding the final displayed output down to a "reasonable for reading" number of decimals, is a small but genuinely important habit in any multi-step numerical pipeline — round too early, at any stage, and the error doesn't disappear, it just becomes invisible until it's already contaminated everything downstream of it.

What happens if the angle gets truncated

It's worth demonstrating, concretely, why this site's calculator carries the golden angle out to ten decimal places (137.5077640500°) rather than rounding it to something tidier like 137.5° for convenience. Generating the identical pattern with the angle truncated to just one decimal place looks nearly identical to the correct version at first — at 100 seeds, the truncated version's minimum angular gap (2.5°) is actually still respectably close to the correct version's (1.81°). But 137.5° is secretly a simple fraction of a circle (137.5/360 reduces to 55/144), and simple fractions are exactly what the golden angle is supposed to avoid. By 500 seeds, the truncated version's minimum gap has collapsed to 0° — points landing essentially on top of each other — while the correctly precise version's gap has continued shrinking smoothly to 0.427°, exactly as expected. The failure isn't gradual or subtle; it's a sudden collapse once the seed count catches up to the truncated angle's hidden rational denominator, which is precisely the mechanism described elsewhere on this site for why "close to the golden angle" isn't good enough — here it is, breaking a real generated pattern rather than staying an abstract number-theory claim.

Choosing a seed count worth generating

The calculator accepts any seed count up to 5,000, a practical ceiling rather than a mathematical one — the golden-angle rule itself places no upper limit on how many points can be generated, but a browser rendering tens of thousands of individual points starts running into ordinary performance limits that have nothing to do with the underlying math. Lower seed counts (10 to 50) are useful for seeing the rule mechanically, one rotation at a time, the way the worked example above does. Counts in the hundreds start to visually resemble the botanical patterns this site discusses elsewhere — a moderately dense sunflower-style seed head is commonly modeled with seed counts somewhere in the many hundreds to low thousands, well within the calculator's range. There's no single "correct" seed count to pick; it depends entirely on whether the goal is inspecting the mechanism or generating something that reads, at a glance, like an actual seed head.

Where this recipe gets used in practice

Because the algorithm is short, deterministic, and produces well-distributed points with no randomness involved, it shows up well outside botany illustration whenever a design or visualization task needs a fixed number of points spread evenly across a disk: generative art pieces that want an organic-looking but fully reproducible dot arrangement, data visualizations that need to place a set of markers or labels around a circular layout without overlap, and radial UI patterns that want visual variety without the rigid look of a grid. None of those uses depend on any claim about golden-ratio beauty; they depend on the same concrete, checkable packing property demonstrated throughout this site — a fixed rotation that never lets points cluster, applied to a plain practical layout problem instead of a seed head.

Generate it yourself

The golden spiral & phyllotaxis calculator runs this exact recipe for any seed count and scale you choose, and shows the resulting coordinates directly rather than just a picture, which makes it straightforward to check any of the worked figures above, or to pull the numbers into a drawing tool of your own. Every figure worked through in this post came from that same tool, run at the specific inputs described, rather than typed in from a separate reference.