A Cookbook for Sailing from A to B

Preparation time: longer than the app says. Serves: one crew. Pairings: rosé.

Every recipe begins with a promise. Combine these ingredients, follow these steps, and something predictable will emerge from the oven at the end. This is the social contract of cooking: effort in, soufflé out. The variables are controlled. The flour does not change direction at 3 AM. The eggs do not have an opinion about your departure time.

Sailing from A to B should be equally straightforward. You are at A. You wish to be at B. A child can draw the line. Google Maps will draw it for you, with an estimated arrival time calculated to the minute, a suggested coffee stop, and a passive-aggressive reroute if you deviate by more than 200 metres.

But Google Maps assumes a road. A road does not rearrange itself based on thermal gradients, planetary rotation, and the gravitational influence of the moon. The sea is not a road. And the line between A and B — that obvious, kindergarten-simple, geometrically irreproachable straight line — is almost never the answer.

This is a recipe for the rest of it.

Ingredients

• 1 boat (any size, any condition, yours)
• 1 set of sails (quantity and ambition may vary)
• 1 engine (optional, but see below)
• 1 captain
• 1 crew (that does what the captain says)
• 1 departure date (more important than you think)
• Weather forecasts (plural; trust none individually)
• 1 set of boat polars (supplied by the manufacturer’s marketing department)
• Sea state (not included in the polars; ask the sea)
• 1 routing app (to process all of the above incorrectly)
• Years of experience in your boat (replaces all of the above)

Method

The Captain

This is the most important ingredient in the recipe and the one no app can quantify.

There are, broadly, two kinds of captain. The first decides, before the lines are cast, that the boat will sail aggressively. The weather is an opponent. The passage is a race — against time, against the forecast, against the part of the captain’s psyche that wonders whether a quieter life might have involved accounting. This captain will hammer upwind in 30 knots at 2 AM, change sails four times in a watch, keep the crew on a rotation that would make a submarine commander wince, and arrive at B twelve hours before the other boats — damp, exhausted, and radiating the particular satisfaction of someone who has won a competition that nobody else entered.

The second captain opens the rosé before the harbour walls have disappeared behind the stern.

Between these two archetypes lies every possible sailing trajectory, every possible arrival time, and every possible state of crew morale. The polar diagram of the boat is theoretical. The polar diagram of the captain is the one that actually matters. No algorithm has yet been developed to model whether the person at the helm will choose the optimal tack angle or the tack angle that doesn’t interrupt lunch.

Both captains will arrive at B. The question is when, in what condition, and whether anyone aboard is still speaking to anyone else. This is not a variable that appears in routing software. It should be.

The Crew

The crew is the captain’s instrument — in theory.

In practice, the crew is a collection of individuals with private opinions about sail changes at 3 AM, personal relationships with the rosé supply, and a finite tolerance for being told that the next tack will be “the last one, I promise.” A crew aligned with the captain — truly aligned, not merely resigned — will execute manoeuvres with conviction, maintain the watch system with discipline, and trust the routing decision even when the routing decision appears to involve sailing away from the destination for six hours.

A crew not aligned with the captain will do something subtly worse than mutiny. They will comply — slowly. They will trim the genoa — approximately. And when the captain goes below to sleep, they will open the rosé, bear off five degrees because “it’s more comfortable,” and undo three hours of VMG (Velocity Made Good — the component of your speed in the direction you actually want to go) without ever raising their voice or violating a single instruction.

The finest routing algorithm in the world is powerless against a crew that has decided, collectively and silently, that arriving four hours later is an acceptable price for not gybing in the dark.

The Starting Date

If the captain is the most important human variable, the departure date is the most important physical one. You don’t have to be a Phoenician — who were crossing the Mediterranean by the stars three thousand years ago — or to have read Joseph Conrad’s The Shadow Line [7] to know that the departure date is everything. Choose correctly and the universe cooperates. Choose poorly and the universe has other plans — some of which are fatal.

Consider the Atlantic crossing. Leave Cape Verde in November, and the northeast trades will carry you to the Caribbean in two to three weeks of downwind sailing so pleasant that the main challenge is boredom. A coconut tossed into the water from the beach in Mindelo, carried by the same trade winds and the North Equatorial Current, will wash up somewhere in the Caribbean roughly around the same time you do — without a crew, without a routing app, without a single sail change, and without ever questioning its life choices. You will arrive marginally faster than the coconut. You will also arrive with a story about how you sailed the Atlantic. The coconut will not care.

Leave Cape Verde in August, and you will discover what the Intertropical Convergence Zone thinks of your schedule. The trip may be much faster — in the sense that a 40-knot squall will accelerate you briefly in a direction you did not choose — or much slower, in the sense that three days of equatorial calms will leave you drifting in a mirrored sea, questioning every decision you have ever made. Or you may not arrive at all — depending on how loosely you define “arrival” and how seriously the hurricane season takes its work.

The same principle applies at every scale. The Mediterranean thermal breeze builds after noon and dies after sunset. A 40-mile passage timed to ride this pattern is sailing. The same passage started at 0600 into glassy calm is motoring while pretending to sail, which is a third category of human endeavour that the recipe acknowledges but does not endorse.

The Sails

Sails are the most visible difference between two boats at sea and the one most directly controlled by the captain. One boat may carry ten sails — a wardrobe of canvas for every angle and every weight of breeze, switched rapidly as conditions evolve, trimmed with the obsessive precision of someone who believes that an extra eighth of a knot is a moral imperative. Another boat may have left the dock with whatever was already on the furler and whatever was accessible in the sail locker without moving the fenders.

The first boat will arrive faster. This is aerodynamics, not opinion. A Code 0 on a broad reach produces forward force that a mainsail alone simply cannot. An asymmetric spinnaker, properly set, turns a 4-knot downwind crawl into a 7-knot sleigh ride.

But sail changes require the crew. The crew whose alignment with the captain may be, at this point in the passage, deteriorating. The captain who changes sails six times in eight hours will extract maximum performance from the boat — provided the crew agrees that maximum performance is the objective. If the crew has silently reclassified the objective as “survival” or “lunch,” the sixth sail change will be executed with the enthusiasm of a tax audit.

Unavoidably — the mathematics is clear — the captain who optimises sail configuration for every shift will arrive sooner. If the crew agrees. The recipe takes no position on crew management. Your relationships are your own.

The Engine

Here is the ingredient that divides sailors into tribes more cleanly than politics, religion, or opinions about anchoring technique.

Consider a simple physical fact. If between A and B there exists a single region of one hundred metres — one hundred metres — with zero wind, then a sailboat without an engine cannot complete the passage. Zero wind means zero force. Zero force means zero acceleration. The boat will sit in that patch of glassy water until something changes — the wind, the current, the crew’s patience, or the captain’s principles. Physics does not negotiate with purity.

Now add an engine. The captain starts the Yanmar, crosses the hundred metres in forty-five seconds, shuts it down, and continues sailing. Arrival is now guaranteed. One hundred metres of diesel solved a problem that no amount of sail trim, routing optimisation, or motivational speaking could address.

From this minimum case — the hundred metres that separate the possible from the impossible — extends the entire spectrum of engine use. Motorsailing into a headwind to maintain schedule. Motoring through a calm to reach the anchorage before dark. Running the engine for eight hours because the wind died and the captain has a restaurant reservation that was made three weeks ago and cannot be cancelled without consequences that exceed the cost of diesel.

The variables are: the captain’s philosophy, the fuel tank capacity, and the owner’s willingness to pay for diesel at Mediterranean marina prices — which are set by people who understand that a sailor without wind is a negotiator without leverage.

The purist, who would rather drift for three days than start the engine, will arrive eventually — or not. The pragmatist will arrive on Tuesday. The motorboater, who hoisted the sails once for a photograph in 2019, will arrive first and will not understand what all the fuss was about. The recipe accommodates all three. It judges none of them. Visibly.

The Weather Forecasts

Many sailors assume this is the most critical ingredient. It is probably not.

Twenty years ago, obtaining a weather forecast offshore required an SSB (Single Side Band) radio, a weatherfax decoder, and the ability to interpret a synoptic chart that looked like a topographic map of someone’s migraine. The forecast was updated twice daily and had the resolution of a political promise.

Today, the forecast arrives on your Starlink every few hours, in full colour, at resolutions that would have made a meteorologist weep with joy in 2005. The ECMWF — widely regarded as the finest operational forecasting system on the planet — released its high-resolution model data to the public last year [1]. Free. For everyone. The same data that commercial routing services charge hundreds of euros to repackage in a prettier interface.

The GFS (Global Forecast System, the American model) runs every six hours. ICON (the German model) offers extraordinary coastal resolution. And the new generation of AI-based weather models — GraphCast, Pangu-Weather, FourCastNet — are producing 10-day forecasts that rival the deterministic models in skill, at a fraction of the computational cost [2].

Data is no longer the problem. For a standard passage in a well-forecasted region during the right season — which, if you were paying attention during the “Starting Date” section, is the only season you should be sailing — the forecast uncertainty is rarely the dominant source of error in your routing calculation.

The dominant source of error is the next ingredient.

The Polars

This is where the recipe falls apart. And this is, therefore, the most important section of this article.

Many sailors, when they hear the word “polars,” think of bears. Very few of them know how to read the diagram. This is a problem, because the polar diagram is the single most important piece of information about your boat’s performance — and if you can’t read it, you are navigating the optimisation problem with a blindfold on. So let us fix that.

A polar diagram is a chart that describes your boat’s theoretical speed at every combination of wind angle and wind strength. Here is one:

The wind comes from the top of the diagram. The vertical axis at the centre is 0° — dead upwind. The angles fan outward: 30°, 45°, 60°, 90° (beam reach), 120°, 150°, all the way down to 180° (dead downwind) at the bottom. The concentric rings represent boat speed — the further from the centre, the faster the boat.

You will notice this diagram has two halves: the left is labelled “Apparent Wind,” the right “True Wind.” This distinction is for the 1% of sailors who know what to do with it — and if you are not in that 1%, here is the shorthand: use true wind for route planning (it’s what the weather forecast gives you and what the routing app expects), and use apparent wind for sail trim while you’re actually sailing (it’s what the sails feel, what the telltales respond to, and what changes the moment your boat accelerates or decelerates). If that sentence made perfect sense to you, congratulations — you didn’t need this article. If it didn’t, use the right half of the diagram and move on. Nobody will judge you. The bears will pat you on the shoulder.

How to read it? Pick a wind speed — say, 12 knots of true wind. Find the corresponding curve (the blue curves are with jib, the red curves with spinnaker). Now trace that curve around from upwind to downwind. At every angle, the distance from the centre tells you how fast the boat should be going.

An example. You have 12 knots of true wind. You are sailing at 90° — beam reach, wind directly on your side. Follow the 12-knot blue curve to the 90° line. It reaches roughly 9.5 knots. That is your theoretical boat speed. Now look at the same 12 knots but at 45° — beating upwind. The curve has retreated toward the centre: about 7.5 knots. The boat is slower, because it’s fighting for every degree against the wind. Now look at 150° — a deep broad reach. The blue jib curve simply isn’t there. The marketing department chose not to show you what happens to a jib at that angle, because the answer is unflattering. But switch to the red spinnaker curve — if you bought one — and you’re back to 7.5 knots. This is why the captain with ten sails arrives faster: the spinnaker opens up an entire region of the diagram that the jib simply cannot reach — and that the brochure would rather you didn’t ask about.

And at 0° — dead upwind? Both curves touch the centre. Zero. The boat cannot sail into the wind. This is not a design flaw. This is physics.

The resulting shape looks vaguely like a butterfly wing — which is appropriate, because it is approximately as fragile and approximately as connected to reality.

Here is the problem.

The polars were generated by the boat manufacturer. A naval architect designed the hull, ran a CFD (Computational Fluid Dynamics) simulation — the same technology that designs Formula 1 cars and aircraft wings — and produced a theoretical performance envelope for a boat that exists only in a computer. This computer boat has a clean bottom. It has the sails specified in the original sail plan. It weighs exactly the displacement listed in the brochure. It has no dinghy on davits, no air conditioning compressor, no four cases of Provençal rosé in the bilge, and no barnacles.

Your boat has all of these things.

The manufacturer does not know that you installed air conditioning. They do not know that your antifouling was last applied two seasons ago and is now providing a thriving habitat for marine organisms that have, in biological terms, won the lottery. They do not know that you replaced the original genoa with a heavier cruising sail, or that your waterline is four centimetres lower than designed because of the generator, the watermaker, and the complete works of Patrick O’Brian in hardback.

They don’t know. And — here is the critical part — they don’t want to know. The polars are what the marketing department asked for. The boat should be fast. On paper. In a brochure displayed at the Düsseldorf Boat Show next to a photograph of the boat sailing at 30 degrees of heel in flat water with a crew of models velcroed to the deck who have never touched a winch handle.

Your actual polars — the ones that describe the boat you actually own, in the condition it actually is — are somewhere between 50% and 80% of the published figures [8]. Nobody publishes those. So don’t be surprised if your trip takes twice as long as the app predicted.

But let us, for a generous moment, pretend the polars are correct. Let us enter Cinderella’s carriage and accept the manufacturer’s numbers. Even in this fairy tale, there is a pumpkin waiting.

The polars assume flat water.

The Sea State

Every sailor who has beaten into a head sea at 2 AM — the bow slamming into each wave with a violence that rattles the cutlery, the hull shuddering, the off-watch crew waking with each impact and lying in the dark wondering whether the keel bolts were inspected this decade — knows the difference between the polar diagram and reality.

The polar diagram was drawn in a CFD simulation where the water is flat. Mathematically, beautifully, impossibly flat. No chop. No swell. No cross-sea reflecting off the cliffs of Bonifacio at precisely the frequency that maximises discomfort.

In the real ocean, waves add resistance — each wave face decelerates the hull. They degrade sail shape — the pitching motion constantly changes the apparent wind angle. And they destroy crew performance — seasick, exhausted, bruised people do not trim sails. They endure.

A 1.5-metre head sea can reduce upwind VMG by 30–40% compared to the flat-water polar [3]. A 2.5-metre sea can halve it. The optimal route in waves may look nothing like the route the polars suggest, because the polars live in a world where the sea is a mirror and the crew is tireless.

This is the moment when you wish the captain were the rosé one. But even the aggressive captain — the one who hammered upwind at 2 AM, who changed sails four times, who treated the passage as a personal vendetta against the Beaufort scale — even that captain, when the slamming starts and the off-watch crew are lying awake bracing against each impact, will quietly bear off ten degrees, ease the sheets, and let everybody sleep. Some battles are not worth winning. The sea will still be there in the morning.

The sea is not a mirror. The crew is not tireless. And the polars, which were already optimistic, are now fiction.

The App

And now the final ingredient: the software that promises to reconcile all of the above into a single, optimal route.

The routing app takes your inputs and performs, in seconds, a calculation that Francis Chichester did by hand with dividers and paper charts in the 1960s [4]. The method is elegant. The assumptions are heroic.

Here is what happens.

The app reads the weather forecast — the one from the ECMWF or GFS, updated every few hours. It reads your departure time. It takes the polars — the ones from the marketing brochure, because those are the ones in its database. It assumes a flat sea. It assumes a perfect crew. It assumes the sails that were on the boat when it left the factory, before the antifouling, before the air conditioning, before the rosé.

It removes everything that makes your boat your boat. And then it calculates.

From your position at A, at departure time, with the forecasted wind at that location and moment — how far can the boat travel in one minute, in every possible direction? This produces a ring of possible positions — an isochrone — one minute into the future. From each point on that ring, the app repeats: given the wind there, one minute later, where could the boat be in the next minute? A new, larger ring. And again. Minute by minute, the rings expand outward like ripples from a stone dropped into a pond — except the pond has currents, the ripples travel at different speeds in different directions, and the stone keeps changing its mind.

The app continues — hundreds of iterations, thousands of positions evaluated — until the wavefront reaches B. Then it traces backwards: which sequence of one-minute steps arrived first? That sequence, connected into a curve, is the optimal route.

Is this complex? It is vastly simpler than it sounds. It is Dijkstra’s algorithm, published in 1959 [5] — the same mathematics that routes packets across the internet and prices financial derivatives. Your passage from Palma to Cagliari is solved by the same maths as a hedge fund pricing a swap. The hedge fund has better air conditioning. But then, the hedge fund started with accurate inputs.

The routing app did not.

And the most difficult part is not the mathematics. The most difficult part is telling the app that the captain changed from the genoa to the Code 0 twenty minutes ago, that the spinnaker went up at sunset and came down in a hurry at 0200 when the wind shifted, and that somewhere around 3 AM the captain — the aggressive one, the one who doesn’t open the rosé — quietly bore off fifteen degrees because the slamming was so violent he became genuinely afraid the hull would crack, and he would rather arrive late than not arrive at all. The app does not know about fear. The app has never heard a hull groan. The app’s polars do not have a column for the sound that makes a captain change his mind.

And here is the irony. Of all the ingredients in this recipe — the boat, the sails, the engine, the captain, the crew, the experience — the app is by far the cheapest. It is, in fact, free. OpenCPN, an open-source chartplotter, includes a Weather Routing plugin that does exactly this: isochrone routing with GRIB (Gridded Binary) weather data and your boat’s polars [6]. The most mathematically sophisticated ingredient in the entire recipe costs nothing. The boat cost you a quarter of a million. The sails cost you twenty thousand. The engine maintenance costs you your patience. But the software that orchestrates all of it — the part that sounds the most complicated — is written by volunteers and available to anyone with a laptop. The universe has a sense of humour.

Serving Suggestion

You have assembled your ingredients. You have followed the method. The app has produced a beautiful curved line across the chart, annotated with ETAs (Estimated Times of Arrival), wind arrows, and a confidence interval that it does not deserve. The route is optimal — for a boat you don’t own, with sails you don’t have, in a sea that doesn’t exist, crewed by people who don’t tire, captained by an algorithm that has never opened a bottle of rosé.

Now you understand why forecasting the optimal path from A to B is difficult, and why the route so calculated rarely matches the one you actually sail.

So here is the real recipe. Replace all of the above — the forecasts, the polars, the app, the isochrones, the marketing brochures, the CFD simulations — with years of experience in your boat. Not a boat. Your boat. The one with the barnacles, the dodgy furler, the crew that won’t gybe after dark, and the engine that takes three attempts to start when cold.

After ten thousand miles, you don’t need the app. You are the app — running the same algorithm on biological hardware, corrected by every mistake you’ve made and every passage you’ve survived.

And if the app says 1400 and you arrive at 1800?

Relax. You arrive when you arrive. The rosé is already cold.

References

1. ECMWF (2025). “Free and open data: ECMWF’s open data initiative.” European Centre for Medium-Range Weather Forecasts.
2. Lam, R. et al. (2023). “Learning skillful medium-range global weather forecasting.” Science, 382(6677), 1416–1421.
3. Gerritsma, J., Onnink, R., & Versluis, A. (1981). “Geometry, Resistance and Stability of the Delft Systematic Yacht Hull Series.” 10th Chesapeake Sailing Yacht Symposium.
4. Chichester, F. (1964). The Lonely Sea and the Sky. Hodder & Stoughton.
5. Dijkstra, E.W. (1959). “A note on two problems in connexion with graphs.” Numerische Mathematik, 1(1), 269–271.
6. OpenCPN Weather Routing Plugin. https://opencpn.org/OpenCPN/plugins/weatherroute.html
7. Conrad, J. (1917). The Shadow Line: A Confession. J.M. Dent & Sons.
8. PredictWind. “Sail Routing Night Time Adjustment.” https://help.predictwind.com/en/articles/10084681-sail-routing-night-time-adjustment