Mallorca's wind is a clock, not a season: the island's own instrument says the hour you leave matters more than the month you book

Ask anyone who has ridden here what decides a day on this island and they will not say the gradient. They will say the wind — and then they will tell you which month to come in, because that is the only handle anybody has ever been given on it. The whole market answers the wind question with a calendar.

The island keeps an hourly wind record and has done since 1973. Read as a rider would read it, it says the calendar is the wrong handle.

In July this station is calm in 61.9% of hours at 08:00 and in 0.6% of hours at 16:00. The mean goes from 1.8 m/s (6 km/h) to 6.1 m/s (22 km/h). That is a swing of 4.4 m/s inside one day. The widest gap between any two months at the same hour, anywhere in the year, is 2.6 m/s.

The hour you leave is worth about 1.7× the month you book.

The sea breeze runs to a timetable, and it runs all year

A sea breeze is not weather in the ordinary sense. The land heats faster than the sea, the air over it rises, and air comes in off the water to replace it. It is a machine driven by the sun, so it starts at a time of day rather than on a day, and it starts at roughly the same time over the whole island because the island shares a sun.

Here is when it starts, by the record’s own rule: the first hour after 06:00 at which the wind comes from the sea more often than it does not come at all.

How we got this Hourly surface observations from NOAA's Integrated Surface Database, cut to the island's bounding box, assessed by counting, and aggregated by month and local hour over two ten-year eras. One observation per local hour, METAR preferred over SYNOP. Directions are the direction the wind blows FROM, in sixteen points; shares are of all hours in the cell, so calm, variable and the sixteen sectors sum to 100.Computed from 87,017 hourly observations over 3,644 days at Palma de Mallorca (LEPA), 2015–2024, with 86,167 more from 1975–1984.Computed 16 August 2026.

Read the second column down. It never leaves the two hours between 11:00 and 13:00 — not in July, and not in January either. The market states this as a summer fact; the record says the machine runs in every month there is, and what changes across the year is how hard it blows once it has started, not when it starts.

Now read the third column. At 08:00 this station is calm in more than half of all hours in every month of the year. The lowest figure in the column is 57.8%. A rider who is rolling before nine is not riding in a lighter wind — most mornings, they are riding in no wind at all.

From April to October it does not shift. It reverses.

The last column but one is the one worth stopping on. It is the angle between the average direction the wind blows from at 08:00 and at 16:00, taking the short way round the compass. A number near 180° means the afternoon wind comes from precisely where the morning wind was going.

7 months of the year sit at 170° or more. April to October is not a season with a bit of a sea breeze in it — it is a compass that turns end for end every single day, from NNE/NE/ENE at breakfast to SSW/SW/WSW by mid-afternoon. The other 4 months — January, February, November, December — do not do this at all: their mornings already blow from NW/NNW/WNW and stay roughly there. March is the month the machine switches on, at 136° — a NNE morning that has become a SW afternoon by four, not yet the full reversal the summer runs.

This is where the most-repeated piece of wind advice in cycling comes apart on this island.

What replaces it is not a direction. It is an hour. The wind here is not a coin you toss at the start of a ride; it is a schedule you can read in advance, and the only real lever on it is what time you are where.

The wind that slows you down is not the wind that stops you

Two different winds get called “the wind on Mallorca”, and conflating them is why the market’s direction advice is so muddled.

The first is the sea breeze: reliable, moderate, and it is the one you meet on almost every ride. 57.0% of all hours at 16:00 blow from SSW/SW/WSW.

The second is what the island is famous for. Of the 786 hours in this record at Beaufort 6 or above — 10.8 m/s, 39 km/h, the point at which a headwind stops being weather and becomes the ride — 30.7% come from W alone and 25.6% from the four northern points. The sea block contributes 16.5%.

So the two questions a rider actually has get two different answers. How much wind will I be riding in this afternoon? — a sea breeze, from the south-west, arriving on schedule. Will there be a day this week I should not go out at all? — a different wind, from the other half of the compass, and it is a winter question: March reaches Beaufort 6 on 12.1% of days and August on 1.6%.

Which loops finish into it

On a loop the wind is not a tax, it is a redistribution. Every kilometre ridden into it is very nearly matched by one ridden with it, and across the 16 published routes measured here the whole-loop headwind share lands between 21.4% and 35.0% — a number that tells a rider nothing.

What is not nothing is where in the ride the headwind falls, because the sea breeze is a clock. A rider who starts at eight is in the last quarter of a long loop at exactly the hour the wind is at its peak. So the measurement is the share of the final quarter ridden within 45° of straight into the afternoon block — and it ranges across these routes from almost nothing to more than half.

How we got this Each track is cut into segments, each segment given a bearing, and a segment counts as into the wind when it travels within 45° of straight into the sea block. The published figure is the share of the final quarter of the distance, in the direction the route's own publisher draws it.Computed from 16 published routes with a measured track, 7 of them inside the basin.Computed 16 August 2026.

Serra de Tramuntana - Sud finishes with 57.1% of its last quarter pointed into the sea breeze, and its whole-loop figure — 29.9% — hides that completely. Fars de Ponent finishes with 2.1%. The two are 7.3 points apart on the whole-loop figure and 55.0 points apart on the one that describes the afternoon — which is the whole argument for measuring the quarter rather than the loop.

Three things this table is not, all of which matter more than the numbers in it. It is the publisher’s direction of travel — ride the loop the other way and head and tail swap. It is a quarter of the distance, not of the time, so on a mountain loop the last quarter is ridden later than this implies, which makes the figure conservative rather than flattering. And it is Palma’s compass: the 9 routes with any point beyond 35 km of the station are excluded rather than estimated — Fars de Migjorn (41.8 km), Ermites del Raiguer (43.1 km), Llogarets del Raiguer (47.3 km), Serra de Tramuntana nord (47.3 km), Serra de Tramuntana - Central 1 (50.5 km), Llogarets del Llevant (56.2 km), Ermites del Llevant (56.8 km), Far de Formentor (61.4 km), Fars del Llevant (66.3 km). An onshore breeze at Port de Pollença is a north-easterly, and this record cannot see it.

What the market says, held against the instrument

19 claims from 7 publishers, each proved word for word against that publisher’s own page on the date beside it, then tested against the record. 8 hold. 4 are true and unquantified — the number is what this page adds to them. 5 differ. 2 are about a coast this instrument cannot see and are carried untested rather than dropped.

How we got this a statement about Mallorcan wind made by a publisher in the survey and proved word for word against that publisher’s own page on the date recorded beside itComputed from a page making at least one claim about wind on Mallorca — its timing, direction, season or strength — addressed to a visitor rather than to a forecaster, and readable without an account, a paywall or a CAPTCHA.Computed 25 August 2026.

The pattern in that table is worth naming. Almost nobody writing about this island is wrong about the wind. What they are is unquantified — and the two claims that do differ are both cases where a sentence that is true elsewhere was carried here unchanged: a headwind band borrowed from the windier tenth of afternoons, and an out-and-back rule that assumes the wind stays where it was.

Fifty years, two eras, and a comparison this page will not turn into a trend

The same station has been reporting since 1973, which makes a before-and-after possible: ten complete years in 1975–1984 against ten in 2015–2024, 86,167 hours against 87,017.

What survives that caution is the part this page is actually built on, and it is the more interesting half: the shape of the day has not moved. The afternoon block in July held 77.3% of hours forty years ago and 77.3% now. Whatever the instrument was doing at the calm end, the timetable it recorded is the same timetable.

What this record cannot see

1 of 6 passes the membership rule — 2,000 observations carrying a wind speed, in at least 8 distinct years. The count is published rather than the conclusion, because a survey that reports only what it found cannot be checked for what it missed.

• One compass, one coast. The rose on this page is measured at Palma de Mallorca, on the bay of Palma. A sea breeze blows onshore, so the same physics gives a south-westerly here and a north-easterly at Port de Pollença. The hour is the island's; the direction is not, and no number here is converted from one coast to another.
• Sea level, not the ridge. The station is 8 m above the sea. The one station in the database on the Serra — ALFABIA MTN RANGE, at 1015 m — reports a median of 868 usable hours a year against 2,000 needed, so the mountain this island is ridden for has no wind record here at all.
• No minutes, no watts. Turning a headwind into time lost needs a drag model, and this site's standing rule is that a second rider model is a second rider. Exposure here is geometry — which way you are pointing, in which hour — and the cost of it is left to the legs it happens to.
• Gusts are not in this. Every figure is a sustained speed. The database carries gusts in an optional section whose coverage here is thin, and a gust figure drawn from a thin section would look like a measurement without being one.

And what would close the biggest of those gaps is named rather than left as a silence:

• AEMET OpenData — hourly observations from the Balearic station network — would answer the compass at Port de Pollença, Alcúdia and the east coast — every claim this survey files as unreached. Not here because: a free API key issued to an email address, which is a human signup and is in the owner queue; the portal is public and nothing here is being worked around.
• NOAA ISD — the Alfabia mountain range station, 08303099999 — would answer what the wind does on the ridge, which is where this island is ridden. Not here because: nothing a human can fix — the station is in the record and was counted: it fails the membership rule on its own numbers, and the count is published in `stations`.

Know before you ride

• Set the alarm, not the calendar. The quiet window ends between 11:00 and 13:00 in every month of the year. Rolling out at seven rather than ten is worth more than choosing a different week — 1.7× more, measured at this station across the hours between 07:00 and 20:00.
• Plan the last hour of the ride, not the first. On a loop the headwind share barely varies; the final quarter varies from 2.1% to 57.1%, and the final quarter is the part you ride at four in the afternoon.
• Do not use the morning wind to choose a direction. For 7 months of the year it will have reversed before you turn round, and most mornings there is not enough of it to choose by in the first place.
• The hard days are a winter thing. March reaches Beaufort 6 on 12.1% of days; August on 1.6%. If a week is going to be blown out, it is not the week you booked for the heat.
• Everything above is measured at sea level on the bay of Palma. The hour is the island’s; the direction is this coast’s. Read the regions guide for what changes as you move across the island, and the 312 guide for the ride where all of this arrives at once, in the second half, at exactly the wrong hour.

How we got this A station is usable when it delivers at least 2,000 observations carrying a wind speed in at least 8 distinct years. 1 of 6 passes. The others are published with their counts rather than omitted, including the ridge station, which reports too thinly to use.Computed from 6 stations inside the island's bounding box, each counted.Computed 16 August 2026.