Begin by asking why a rabbit recurrence is not already the same story as a sunflower’s golden spiral. Move through recurrence and phyllotaxis in Pisa, Paris, and Leipzig; matter and diffusion in Dundee and Manchester; local rules in Utrecht and Cambridge; scale and growth grammars in Yorktown Heights and Calgary; then physical, numerical, organismal, and molecular tests in Paris, Los Alamos, Kyoto, Bern, Freiburg, and Barcelona. Four experiments vary angle, diffusion, branching, and neighbor rules while keeping visible that similar pictures need not share a cause.
QUESTION FOR THE ROUTE
Instead of searching for a formula describing a finished pattern, how can we distinguish and test the local rules, materials, boundaries, growth, and environments that keep producing it?
WHAT THIS RIVER DOES NOT CLAIM
The river is not geography. Distance downstream stands for time passing, and the light turns from dawn to dusk as the centuries go by. The objects by each stele are symbols of the kind of event and of how each century band wrote and calculated; they do not reconstruct any real artefact. The land around each stop sketches the natural geography of the scene’s real place, and an iconic building appears only if it already stood in that year. Each scene keeps its real place and evidence basis; open it on the map to read where it happened. This route does not claim that nature obeys one law called the golden ratio, fractals, or the Turing equations. Fibonacci’s rabbit problem is not retroactively a botanical discovery, a computed picture is not biological proof, L-system symbols are not one-to-one genes, and the Game of Life is not actual life. The line is an edited comparison among sequences, geometric observation, physics, chemistry, computation, and molecular experiments—not a proven single chain of transmission; publication, research base, and experimental sites remain distinct at each pin.
WHAT YOU SEE ON THIS RIVER
- Diagram in the sky
- A phyllotaxis spiral set at the golden angle
- Emblem at the source
- A disc of seeds set in spirals
- The real place around each stop
- Around each stele the land takes on the natural geography of that scene’s real place — sea or lake, plain, hills or mountains, the colour of the ground and its common trees — and, where one defines the place, its landform: a volcano, snow peaks, granite domes, a mesa, dunes, a fjord, islands, a rock hill, a gorge or loess terraces. The water near the stop takes the colour of the real river or sea, and the haze the place’s climate. A small globe on the stele marks where it is, with the route from the previous place. Where a city has an iconic building that already stood in the scene’s year, its schematic silhouette rises behind the stop and is named on the card. The land follows today’s terrain and climate as a sketch and the silhouettes are not measured reconstructions. Between stops the river itself stays symbolic.
- A figure board at every stop
- Each board draws the mathematics of that scene. When the boat arrives, the construction is drawn in and the key result rises in red. The drawings are schematic reconstructions, not historical manuscripts.
- Century bands along the banks
- 500–1449 · Stone stele · paper lanterns · single-arch bridge · watermills and villages · lateen boats
- 1750–1899 · Cast-iron plaque · gas lamps · iron truss bridge · factory chimneys, railway and steam train · steamboats
- 1900–1969 · Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft
- 1970 onward · Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites
Where the century band changes, the boat passes under a bridge of the new band. Villages, mills, factories, pylons and towers stand for the technology of each century, not for any real place or architectural style.
01·1202 CE·Pisa(basis: Composition)
Counting Rabbit Reproduction with a Recurrence — Fibonacci’s Liber Abaci
Leonardo of Pisa included an idealized rabbit-breeding problem in the 1202 Liber Abaci. A sequence in which the two preceding terms produce the next compressed growth into a rule where earlier states generate a later state. He did not study plant phyllotaxis or the golden angle, however, and real rabbit populations do not ignore seasons, death, and resources as the problem does.
- Pause and ask
- What does a rule that makes the next generation from the preceding two reveal about growth, and what realities does it erase?
- How thinking changed
- Shift from a finished table of population counts to a recursive process in which earlier states generate the next state.
- What we cannot claim
- Fibonacci did not investigate plant spirals or the golden angle. The rabbit problem omits death, seasons, and resource limits, so it is not a law of real ecological growth.
- This place
- Pisa was the writing base for Liber Abaci’s reconstruction of Mediterranean calculation for Latin readers; the rabbit problem was one among its commercial-arithmetic and algebraic problems. (Arno banks · pines and olives · flat plain · 43.7°N 10.4°E)
- Figure board
- Each grown pair (●) bears a new pair (○) every month and new pairs mature a month later: 1, 1, 2, 3, 5, 8 pairs.
- On the river
- A desk holding a written record · 500–1449 (Stone stele · paper lanterns · single-arch bridge · watermills and villages · lateen boats)
02·1837 CE·Paris(basis: Publication)
Reading Leaf Spirals through Fractions and Lattice Arrangements — the Bravais Brothers
Louis and Auguste Bravais described leaves and seed arrangements around stems using spirals, fractions, and intersecting curve families in French botanical work of the 1830s. Connections between observed arrangements and fractions close to Fibonacci ratios made phyllotaxis a quantitative subject. This was neither a theorem that every plant optimizes the golden ratio nor a single 1837 discovery of one golden angle.
- Pause and ask
- When leaves and seeds around a stem become fractions, spirals, and lattices, which regularities appear and which variations disappear?
- How thinking changed
- Move from naming a plant’s finished appearance to geometry that measures and classifies the order and angular relations of its primordia.
- What we cannot claim
- Observed arrangements are not reduced to one golden angle. The Bravais work is not a universal theorem that every plant optimizes the golden ratio.
- This place
- Parisian natural-history journals, societies, and specimen networks made it possible to publish and debate the Bravais brothers’ observations as comparable diagrams and fractional series. (Seine banks · broadleaf trees · flat basin · 48.9°N 2.4°E · landmark: Notre-Dame de Paris (1250), Dôme des Invalides (1706))
- Figure board
- Unrolled onto a lattice, the leaves around a stem form two crossing spiral families; leaf 8 sits right above leaf 0 after three turns (3/8).
- On the river
- A stack of books · 1750–1899 (Cast-iron plaque · gas lamps · iron truss bridge · factory chimneys, railway and steam train · steamboats)
03·1868 CE·Leipzig(basis: Publication)
Asking Whether a New Leaf Appears in the Widest Remaining Gap — Hofmeister
Wilhelm Hofmeister’s 1868 book organized observations that a new leaf primordium tends to appear away from older primordia at the growing tip. Rather than attaching a fixed number sequence to nature, the idea opened a route from local spacing on a growing surface to angular order. Leipzig marks publication; his principal professorial bases were Heidelberg and Tübingen.
- Pause and ask
- Can spirals appear from the local placement of a new primordium in the widest gap, without prescribing a golden angle in advance?
- How thinking changed
- Shift from a sequence of fixed angles to a process in which primordium size, distance, and birth order on a growing surface select an angle.
- What we cannot claim
- The ‘widest gap’ is a productive empirical rule, not a complete single law for every plant tissue. Leipzig is not misrepresented as his laboratory.
- This place
- The Leipzig pin marks the book’s 1868 publication. Hofmeister’s observation and professorial bases remain distinct from the publication site. (Riverside woods · broadleaf trees · flat plain · 51.3°N 12.4°E)
- Figure board
- As older primordia are pushed outward by growth, the new one appears in the widest empty gap at the tip.
- On the river
- A stack of books · 1750–1899 (Cast-iron plaque · gas lamps · iron truss bridge · factory chimneys, railway and steam train · steamboats)
04·1917 CE·Dundee(basis: Teaching / position)
Comparing Cells, Shells, and Bones through Forces and Transformations — D’Arcy Thompson
In Dundee, D’Arcy Thompson’s On Growth and Form compared biological shapes through physical forces, changes of scale, and transformed coordinate grids. Deforming a grid from one organism’s outline toward another shifted attention from names to processes and constraints. The approach did not invalidate genetics, development, or natural selection, and similar outlines do not prove identical causes.
- Pause and ask
- Does connecting similar outlines with a deformed coordinate grid explain their causes, or does it create a question for comparison?
- How thinking changed
- Read biological form not as the result of a taxonomic name but as a deformable record left by forces, materials, scale, and growth rates.
- What we cannot claim
- A grid transformation describes a difference in form; it neither proves a shared developmental cause nor discards genetics and evolution.
- This place
- Dundee’s university, natural-history specimens, and mixture of classics with mathematical physics gave D’Arcy Thompson a durable base for comparing form across disciplines. (Firth of Tay · broadleaf trees · hills · 56.5°N 3.0°W)
- Figure board
- An outline drawn on a square grid, then the grid smoothly deformed: the same coordinates carry it to a related but different outline.
- On the river
- A lectern and a board · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
05·1952 CE·Manchester(basis: Publication)
Showing How Diffusion Can Break Uniformity — Turing’s Morphogenesis
At the University of Manchester, Alan Turing showed mathematically that reacting chemicals diffusing at different rates can amplify a small initial disturbance into a spatial pattern. The result reversed the intuition that diffusion only smooths differences and supplied conditions under which local activation and longer-range inhibition could make spots or stripes. It proposed a possible mechanism; it did not identify the molecules behind every animal pattern.
- Pause and ask
- Why can diffusion, which normally smooths differences, amplify a small disturbance into spots or stripes when two substances react?
- How thinking changed
- Move from an external blueprint stamping a pattern to dynamics in which a uniform state becomes unstable and amplifies spatial differences.
- What we cannot claim
- The 1952 paper proposed a possible mathematical mechanism. It did not show that every biological pattern comes from the same two-chemical equations or identify the actual molecules.
- This place
- The University of Manchester’s environment of mathematics, computing, and biological questions supported Turing’s move from postwar computation to equations of chemical morphogenesis. (River Irwell · broadleaf trees · flat basin · 53.5°N 2.2°W · landmark: Manchester Town Hall (1877))
- Figure board
- From a nearly uniform state, only some wavelengths of a small disturbance grow into a pattern; the positive band of the growth-rate curve picks them.
- On the river
- A stack of books · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
06·1968 CE·Utrecht(basis: Publication)
Writing Rules for What Cells Tell Their Neighbors — Lindenmayer
Botanist Aristid Lindenmayer used parallel rewriting rules in Utrecht to represent cell division and state changes in filamentous organisms. An L-system compressed the question ‘what does each part become alongside its neighbors at the next step?’ into a short grammar. Biological growth modeling came before the later beautiful tree graphics, and its symbols are not literal one-to-one genes.
- Pause and ask
- If every part executes its own rule in parallel, how can length and branching grow without a central designer?
- How thinking changed
- Shift from a finished plant picture to a temporal grammar in which symbols are rewritten in parallel to generate structure.
- What we cannot claim
- L-systems did not begin as decorative tree-drawing algorithms, and their symbols and rules are not one-to-one copies of genes or cell signals.
- This place
- Utrecht’s botanical research environment was Lindenmayer’s durable base for connecting formal languages with interactions among biological cells. (Canals and river · broadleaf trees · flat land · 52.1°N 5.1°E · landmark: Dom Tower of Utrecht (1382))
- Figure board
- Every cell of a filament is rewritten at once by a → ab, b → a, so it grows 1, 2, 3, 5, 8 cells long with no central design.
- On the river
- A stack of books · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
07·1970 CE·Cambridge(basis: Main activity)
Watching Birth, Death, and Motion on a Grid That Is Not Alive — Conway’s Life
John Conway’s Game of Life lets each square in a grid live or die in the next generation using only the states of its eight neighbors. Oscillators, moving gliders, and structures that appear after long delays emerge from a few local rules, making it possible to test whether order requires a global blueprint. Despite its name, it is not a model of real cells or ecosystems, and visual complexity alone is not life.
- Pause and ask
- When three neighbor-count rules produce moving, oscillating, and long-lived structures, where should we say the complexity resides?
- How thinking changed
- Move from putting complex instructions inside each entity to emergence produced jointly by simple local updates and an initial arrangement.
- What we cannot claim
- ‘Life’ in the Game of Life is a metaphor. It does not model real cellular metabolism, evolution, or environment, and complex motion alone is not a test for life.
- This place
- Cambridge’s community around combinatorial games, logic, and mathematical play supported Conway’s construction and circulation of a cellular automaton that could be tested by hand. (River Cam · willows · flat fen edge · 52.2°N 0.1°E)
- Figure board
- Rules that see only eight neighbours (born on 3, survive on 2 or 3) move a glider one cell diagonally every four generations; a bar oscillates.
- On the river
- A place of ongoing work, marked only by the route’s emblem · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
08·1975 CE·Yorktown Heights(basis: Main activity)
Measuring Roughness That Reappears across Scales — Mandelbrot
At IBM, Benoit Mandelbrot proposed the name ‘fractal’ in 1975 and gathered coastlines, clouds, and branches into a mathematical language for roughness that persists under changes of scale. It made structures missed by integer dimension comparable. Natural objects have finite sizes and materials, so they are not infinitely exact self-similar fractals, and similar branching need not imply a shared cause.
- Pause and ask
- Even when branches and vessels look similar under magnification, should we not also measure the scale at which similarity stops?
- How thinking changed
- Shift from geometry of smooth lines, surfaces, and integer dimensions to geometry that measures how length and roughness change with scale.
- What we cannot claim
- Natural trees, lungs, and vessels have finite cells and materials, so they are not exactly self-similar across infinite scales. Fractal dimension is a descriptor, not a complete cause.
- This place
- IBM Yorktown Heights supplied computation, visualization, and cross-disciplinary research conditions for Mandelbrot to compare scaling phenomena as one program. (Woods · oak and maple · rolling hills · 41.3°N 73.8°W)
- Figure board
- Measuring a curve whose wiggles repeat across scales gives a sloped log–log line, which flattens below its finest wiggle.
- On the river
- A place of ongoing work, marked only by the route’s emblem · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
09·1990 CE·Calgary(basis: Data release)
Drawing Branches and Leaves from a Short Growth Grammar — The Algorithmic Beauty of Plants
Przemysław Prusinkiewicz and Aristid Lindenmayer connected L-systems, turtle graphics, branching angles, and growth stages in Calgary to generate plant forms without copying a finished picture. The process made repeated rules and developing structure visible. These models are tools for investigating plant development, not literal programs that replace genes, hormones, gravity, light, material limits, and environmental response.
- Pause and ask
- When a short branching grammar makes a plausible plant, how do we separate visual resemblance from resemblance in the actual growth mechanism?
- How thinking changed
- Move from storing a finished shape as coordinates to regenerating it by executing growth stages, branching angles, and changes in thickness.
- What we cannot claim
- A procedural tree is not a literal translation of a genetic program. Without light, gravity, hormones, mechanics, and damage response, it cannot fully explain a real plant’s causes.
- This place
- The University of Calgary’s computer-science and plant-modeling work, together with open Algorithmic Botany materials, expanded L-systems into reproducible visual experiments. (Bow River · snowy Rockies · aspen and spruce · 51.0°N 114.1°W)
- Figure board
- Running the rule F → F[+F]F[−F]F once, twice and three times with a turtle grows branches at the same angle δ.
- On the river
- Glowing data columns · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
10·1992 CE·Paris(basis: Experiment)
Testing Whether Repelling Drops Can Produce Fibonacci Phyllotaxis — Douady and Couder
Stéphane Douady and Yves Couder placed magnetic-fluid drops one after another around a circular rim so that they repelled one another. Changing a condition analogous to growth rate produced transitions among spiral arrangements, including Fibonacci-related families, showing how local placement can select order without a global golden-ratio blueprint. The apparatus was not a plant shoot apex and did not establish the arrangement of every species.
- Pause and ask
- If sequential placement and mutual repulsion produce Fibonacci spiral transitions, is the golden ratio a cause or an outcome?
- How thinking changed
- Move from finding ratios in a finished sunflower to a physical experiment that varies a growth-like condition and watches arrangements be selected.
- What we cannot claim
- The magnetic-drop apparatus is not a plant shoot apex; it tests a possible physical principle for Fibonacci arrangements, not every plant’s actual molecular and cellular mechanism.
- This place
- Parisian experimental physics combined magnetic-fluid drops, a rotating apparatus, and imaging to make a local-placement hypothesis for phyllotaxis manipulable. (Seine banks · broadleaf trees · flat basin · 48.9°N 2.4°E · landmark: Eiffel Tower (1889), Notre-Dame de Paris (1250))
- Figure board
- Drops placed one by one at the centre repel into a spiral; lowering the growth-like condition G shifts them from 180° alternation into a spiral.
- On the river
- An experiment stand with a swinging pendulum · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
11·1993 CE·Los Alamos(basis: Main activity)
Finding Spots, Mazes, and Waves in a Two-Chemical Grid Calculation — Pearson
At Los Alamos, John E. Pearson numerically explored the Gray–Scott reaction–diffusion equations and mapped how small changes in feed and removal rates separate spots, stripes, waves, and self-replicating structures. The same equation family can yield radically different patterns under different parameters and initial states. These are not Turing’s exact original equations, and a computed image alone does not prove a particular organism’s chemistry.
- Pause and ask
- When a few parameter changes split the same two equations into spots, mazes, and waves, what should be compared before the pictures?
- How thinking changed
- Move from attaching one formula to one pattern to computation mapping families of patterns across parameters, initial conditions, boundaries, and numerical resolution.
- What we cannot claim
- The Gray–Scott system is not Turing’s exact 1952 equations. Even when numerical patterns resemble skin, molecules, reaction rates, and tissue boundaries require separate validation.
- This place
- Los Alamos National Laboratory’s nonlinear-computation resources supported Pearson’s repeated Gray–Scott simulations across broad parameter ranges and classification of pattern regimes. (Pajarito Plateau · mesas · ponderosa pines · 35.9°N 106.3°W)
- Figure board
- Moving a little in the plane of feed rate F and removal rate k splits the same two equations into spots, mazes or waves.
- On the river
- A place of ongoing work, marked only by the route’s emblem · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
12·1995 CE·Kyoto(basis: Experiment)
Comparing Moving Angelfish Stripes with Reaction–Diffusion — Kondo and Asai
Shigeru Kondo and Rihito Asai observed new stripes being added and older stripes shifting as angelfish grew, then compared that time-dependent behavior with reaction–diffusion simulations. Testing movement over time was stronger than matching a static picture. Kyoto marks Kondo’s institutional base while Asai’s Seto Marine Biological Laboratory was in Shirahama, and the study did not yet identify a specific molecular pair.
- Pause and ask
- Does evidence for a model become stronger when it matches stripe motion and insertion during growth, rather than one static picture?
- How thinking changed
- Move from visual resemblance in a finished pattern to discriminating a dynamic model through a new prediction about position changes during growth.
- What we cannot claim
- The 1995 result strongly supported reaction–diffusion as a viable mechanism but did not identify an actual activator–inhibitor molecular pair. It is not written as a single-site Kyoto experiment.
- This place
- The Kyoto pin marks Kondo’s university base, while the scene separately identifies Asai’s Seto Marine Biological Laboratory and fish-observation base in Shirahama. (Kamo River · cedar and maple · mountain basin · 35.0°N 135.8°E · landmark: Tō-ji Five-storied Pagoda (1644))
- Figure board
- At three stages of growth, existing stripes drift apart and new stripes (red) are inserted into the widening gaps.
- On the river
- An experiment stand with a swinging pendulum · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
13·2003 CE·Bern(basis: Experiment)
Looking for New Leaf Positions in Auxin Flow Rather Than a Golden-Angle Gene — PIN1
Researchers at the University of Bern showed that the orientation of the auxin transporter PIN1 and local auxin accumulation precede the positions of new primordia in Arabidopsis. Phyllotaxis narrowed from one abstract angle to a developmental process combining cell-to-cell transport, tissue growth, and the influence of existing primordia. The result did not discover a ‘golden-angle gene,’ and other arrangements remain possible across species and conditions.
- Pause and ask
- If auxin transport direction and accumulation change before a leaf forms, which cellular process produces the abstract angle?
- How thinking changed
- Shift from a global number called the golden angle to feedback among membrane-transporter orientation, local hormone maxima, and tissue growth.
- What we cannot claim
- This work did not discover a ‘golden-angle gene.’ PIN1 and auxin are important mechanisms, not a solitary command that erases tissue mechanics, growth, and species variation.
- This place
- The University of Bern’s microscopy, molecular markers, and genetic perturbation tools let phyllotaxis hypotheses be tested at cellular scale in a living shoot tip. (Aare bend · Bernese Alps · broadleaf and fir · 46.9°N 7.4°E · landmark: Zytglogge (1771), Bern Minster (1893))
- Figure board
- In the cell mesh of the shoot tip, PIN1 transport arrows converge where auxin builds up, and a new leaf primordium follows there.
- On the river
- An experiment stand with a swinging pendulum · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
14·2006 CE·Freiburg(basis: Experiment)
Testing Whether WNT and DKK Select Hair-Follicle Spacing in Mouse Skin
Freiburg researchers altered WNT signaling and its inhibitor DKK while observing changes in periodic mouse hair-follicle spacing, connecting a reaction–diffusion-like mechanism with living tissue. Mathematical conditions, molecular candidates, and an observed pattern met in one experiment. Discriminating a unique model still requires further measurement, and tissue growth, mechanics, and additional signals cannot be ignored.
- Pause and ask
- If an activating signal and a farther-reaching inhibitor operate in real skin, should perturbing them change follicle spacing as predicted?
- How thinking changed
- Move from abstract activator–inhibitor curves to an in-vivo mechanism test combining gene expression, a secreted protein, and spacing changes in mouse tissue.
- What we cannot claim
- Observation and perturbation support a reaction–diffusion-like account without automatically selecting one unique mathematical model. Growth, mechanics, other signals, and competing accounts still require comparison.
- This place
- Freiburg’s molecular-development and skin-biology environment supported perturbing WNT and DKK, connecting them with tissue pattern, and comparing the result with mathematical models. (Dreisam riverside · fir forest · Black Forest · 48.0°N 7.8°E · landmark: Freiburg Minster (1330))
- Figure board
- Under periodic curves for short-range WNT and far-reaching inhibitor DKK, perturbing them tests whether follicle spacing d changes.
- On the river
- An experiment stand with a swinging pendulum · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
15·2014 CE·Barcelona(basis: Experiment)
Combining a Turing Network with Positional Signals in Digit Patterning — BMP, SOX9, and WNT
Researchers at Barcelona’s CRG perturbed BMP, SOX9, and WNT interactions in embryonic limbs and compared the results with models, supporting a Turing-type network in repeated finger and toe patterns. Positional signals along the limb and tissue growth also adjusted the final number and shape. This is strong biological support for a 1952 idea, not a conclusion that reaction–diffusion alone designs an entire hand.
- Pause and ask
- If self-organization makes a repeated pattern while positional signals orient the limb, can either one be called ‘the formula for a hand’?
- How thinking changed
- Move from two hypothetical chemicals to an integrated developmental model that perturbs a BMP-SOX9-WNT network while comparing positional information and growth.
- What we cannot claim
- The result is strong biological evidence for a Turing-type network, not proof that reaction–diffusion alone designs an entire hand. Positional signals, growth, and mechanics also matter.
- This place
- Barcelona’s CRG collaboration across developmental biology, genomics, and quantitative modeling combined molecular perturbations with spatial pattern data from embryonic limbs. (Mediterranean coast · pines and palms · hills · 41.4°N 2.2°E)
- Figure board
- A BMP–SOX9–WNT network lays repeated stripes (red) in the limb bud, while positional signals and growth adjust their number and shape.
- On the river
- An experiment stand with a swinging pendulum · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)