Main menuGallery

Fibonacci phyllotaxis

This exhibit needs WebGL, which your browser or device doesn’t seem to support.
Try a recent Chrome, Edge, Firefox or Safari.

Fibonacci phyllotaxis

Each seed is planted one fixed turn after the last. Drag to orbit · scroll or pinch to zoom.

137.5078° per seed
spiral families:
no straight arms

Turn angle θ

137.5°
+0.0078°

Trap angles — simple fractions of a turn

Any simple fraction p/q of a turn sends every q-th seed to the same bearing, so the seeds collapse onto q straight arms and waste the space between them.

Head

600

Spiral families

Every seed is measured against its nearest neighbours. The gaps in the index that keep coming up are the spiral families — tap one to trace it.

 

The fractions that fail

p/q of a turnθarms

Each row is one Fibonacci number over the next but one. They close in on 1/φ² = 0.381966… — and the limit is the one turn that is not a fraction at all.

F(n+1) / F(n) → φ

ratiovalueθ = 360°/ratio²

 

Learn

Where’s the maths? This is number patterns and ratio — the same Secondary 1 topic as “find the next term” — wearing a very unusual costume. A sunflower builds its head with one rule repeated: put down a seed, turn through a fixed angle θ, push the older seeds outwards, put down the next seed. Seed number k therefore sits at bearing and at a distance ck from the centre, so the whole head is a sequence written in polar coordinates rather than in a list. Everything then depends on the single number θ. If θ is a simple fraction of a turn, the seeds line up into a few straight arms and most of the head is wasted; if θ is as far from every simple fraction as a number can be, the seeds never line up and the head packs evenly. That “worst-approximated” number is the golden ratio, which is why the spirals you can count on a real sunflower are Fibonacci numbers — 13 and 21, or 21 and 34, or 34 and 55. The spirals are not planted; they are what your eye makes of a lattice built by one turn repeated a few hundred times.

๐Ÿ“˜ On the Sec 1โ€“4 syllabus
  • Sec 1 ยท N5 Number patterns โ€” 5.5 patterns and the nth term (Fibonacci)
  • Sec 1 ยท N2 Ratio and proportion โ€” 2.1/2.2 ratios (consecutive Fibonacci ratios โ†’ golden ratio)
  • Sec 1 ยท G1 Angles, triangles and polygons โ€” 1.2 angles at a point (the 137.5ยฐ turn)
  • Sec 3/4 ยท G5 Mensuration โ€” 5.8 radians

Beyond the syllabus: the golden ratio's irrationality argument is enrichment.

The rule on screen is: seed k at angle , radius ck. The square root is not decoration. Area grows like radius squared, so if the k-th seed is to have the same room as the first, its distance from the centre must grow like √k. That one choice makes the density even; the angle θ then decides how the seeds are shared out around the ring. In radians the golden angle is 2π(1 − 1/φ) ≈ 2.3999 rad — the same turn measured the Secondary 3/4 way.

Why fractions fail. Suppose θ = p/q of a turn, say 25 = 144°. After q = 5 seeds you have turned through p whole turns and you are pointing exactly where you started, so seeds 1, 6, 11, 16, … all sit on one bearing. Only 5 bearings are ever used and the head becomes 5 straight arms with empty wedges between them. Tap the trap buttons: the number of arms is always the denominator.

The Fibonacci sequence is 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, … where each term is the sum of the two before it: F(n+1) = F(n) + F(n−1). Divide one term by the one before and the answers zig-zag in on a limit: 8/5 = 1.6, 13/8 = 1.625, 21/13 = 1.6154, 34/21 = 1.6190, … converging on the golden ratio φ = (1 + √5)/2 = 1.618034…. Turn that into an angle and you get 360°/φ² = 137.5078°, the golden angle.

Where the spirals come from. Nothing draws them. Seed k and seed k + m land close together whenever is nearly a whole number of turns — and for the golden angle the values of m that do that best are exactly the Fibonacci numbers. Your eye joins each seed to the neighbour m steps away and reads a curve. Two families are usually clearest, one clockwise and one anticlockwise, and their counts are consecutive Fibonacci numbers; the app finds them by measuring, not by assuming. Since F(n) + F(n+1) = F(n+2), the two counts always add up to the next family out.

Best of all the bad approximations. Every number can be squeezed between fractions. φ is the number that resists hardest: its continued fraction is 1 + 1/(1 + 1/(1 + …)), all ones, which converges more slowly than any other. Slow convergence means φ is never close to a simple fraction, so 137.5078° is never close to p/q of a turn, so the seeds never fall into arms. The plant is not doing number theory; plants that packed badly simply grew fewer seeds.

Try this: set Seeds to 600 and press Golden angle, then tap the two family buttons in turn and read the counts — you should get 34 and 55, and 34 + 55 = 89, which is the next family along. Now drag the Fine slider slowly from −2° to +2° without touching anything else. The packing falls apart and comes back within a fraction of a degree either side of 137.5078° — that is how sharp the golden angle is. Finally press (135°, which is 3/8 of a turn, one Fibonacci over another) and count the arms: 8, the denominator, and the head is mostly empty space.

Drag to orbit · scroll or pinch to zoom · drag the Coarse and Fine sliders to change the turn angle.