Law 11 and the Armpit Line: Re-reading the Offside Law Eight Years On

Minute 64, Kazan, 30 June 2026. Kylian Mbappé collects the ball near the halfway line, drives four touches past three Argentina players and finishes into the near corner. The stadium erupts, France lead 3-2, and in the press tribune I write an analysis blog within twenty minutes. My conclusion: the goal is offside, because in the frame I am watching, Mbappé's shoulder is ahead of the last Argentina defender's shoulder.

The next morning I receive a short reply from a specialist in the Laws: you are arguing with a version of the Law that has already expired, and the arm is not considered when determining offside.

It took me three months to understand that the arm does not belong to the offside law. In those three months I rebuilt fifty offside situations from the 2026 World Cup, frame by frame, line by line. The outcome was not that I got one goal wrong. The outcome was that I had argued with my eyes while what I actually needed was a written proposition. A mistake is a footnote; only silence is a verdict.

Law 11 is the offside provision in the Laws of the Game issued by IFAB. IFAB was founded in 1886, comprises the four football associations of England, Scotland, Wales, Northern Ireland and FIFA, and is the only body empowered to amend the Laws. A new edition of the Laws takes effect from 1 July each year, a small detail that sits at the root of a good number of disputes, because it means a competition finishing late may be officiated under a different text from one that has just kicked off.

The history of Law 11 is a history of loosening and tightening. The 1863 rules outlawed almost any player ahead of the ball. In 2026, IFAB reduced the minimum number of defenders from three to two, and the 2026-26 season saw goals in England's top division surge. In 2026 the Law moved the other way: a player level with the second-last opponent was no longer offside, which is why the phrase "level is onside" exists today. Between 2026 and 2026, IFAB clarified what it means to take part in active play; the 2026/14 edition added the phrase "gaining an advantage from that position".

The amendment that matters most to me is the 2026/22 edition. For the first time the Law states it in a sentence: the hands and arms of all players, including goalkeepers, are not considered when determining offside. The boundary of the arm is defined as the bottom of the armpit. Before that, from 2026, referees had been instructed to apply that spirit, but the text had not been rewritten. The three years from 2026 to 2026 were three years in which a working instruction lived alongside an old provision, and an expired clause still says more than an infinite promise.

When a referee determines offside, the player's body is divided into two zones. The zone that counts comprises head, torso, legs and shoulder. The zone that does not count comprises hand and arm, measured from the bottom of the armpit downwards. The core point is that the same piece of anatomy can carry two different legal definitions in two different Laws, and that is the origin of almost every dispute supporters describe as "VAR drawing a ridiculous line".

Take the shoulder. Under Law 11, the shoulder counts. Under Law 12, a ball striking the shoulder is not a handball offence, because the arm boundary also ends at the bottom of the armpit. Two Laws share one anatomical line but use it for two different purposes: one to establish position, the other to establish conduct. When a defender raises an arm and the ball strikes the deltoid, supporters see handball, the Law sees shoulder. When a forward rotates the shoulder while accelerating and the shoulder passes the defender, supporters see offside, the Law also sees offside, but the drawn line will sit at a point the naked eye never notices.

Across the seasons I have spent watching football in Brisbane, I have given a lot of time to the acceleration phases of wide players. What I learned is that at high speed the shoulder leads. The arms swing back, the shoulder rotates forward, and at the final instant before the pass is played, the part that protrudes beyond the defender is usually the tip of the shoulder rather than the foot. The image supporters see is visually correct, but it does not correspond to the part of the body the Law permits for determining position.

At the 2026 World Cup in Qatar, a major tournament used semi-automated offside technology for the first time. The system comprises twelve cameras mounted under the stadium roof, tracking twenty-nine data points on each player fifty times per second, plus a sensor inside the ball operating at five hundred times per second. According to FIFA's published data, the average time for an offside check fell to around twenty-five seconds, compared with roughly seventy seconds under the earlier manual process.

On 22 November 2026, Argentina lost 1-2 to Saudi Arabia and had three goals disallowed, including offsides so narrow that no television camera angle could show them to the naked eye. That was one of the matches that introduced the phrase "semi-automated offside" to a global audience on the scoreboard graphic.

Technology does not adjudicate. It supplies data; the referee still decides, and that decision turns on a question that is not technical at all: which instant is the instant a team-mate first touches the ball. On a long pass, the gap between the frame of the ball leaving the foot and the next frame can correspond to tens of centimetres of movement by a runner. The dispute shifts from "was it offside" to "which frame did the ball leave the foot". That is a harder dispute, and more cameras do not answer it.

The broadcast line is not generated automatically in every case. In competitions without a semi-automated system, a technician must select a frame, calibrate the stadium's perspective, and project a vertical line from a reference point. The error in that process has three parts: frame-selection error, perspective-calibration error, and resolution error. On a line separated by a few centimetres, all three can be larger than the result. That is why the semi-automated system matters: it replaces frame selection with continuous tracking data.

There is a technical difference between goal-line technology and offside technology. The goal line is a binary question: has the ball fully crossed, and the answer does not depend on timing. Offside is a continuous question: everything depends on the instant the ball leaves the foot, and that instant is not a point but an interval. A system that measures position to the millimetre still fails to settle the matter if its unit of time does not match the Law's unit of time.

From the 2026/22 season, position-tracking created a new skill for defenders. Defenders used to be taught to watch the forward's shoulder. Now they must learn to place the body so that an opponent's lower armpit does not pass beyond their own nearest scoring part. At youth academies the drill of four defenders moving along an imaginary string still exists; the string has simply been lowered to armpit height. For a wide defender in the A-League, this change is really a change of body posture in the instant the ball is played, and it takes hundreds of sessions to become reflex.

The principle that "level is onside" arrived in 2026 as a way to reduce disputes, and it worked. Before 2026, a referee had to decide whether a forward was nearer to the goal line than the second-last opponent, an ambiguous question. After 2026, the question became a comparison of two concrete positions. But that principle also pushed referees into a new mode of thinking, in which the boundary is no longer a blurred band but a line. When data is complete, that line is clear. When data is incomplete, the line disappears, and the referee falls back on judgement.

Law 11 has one further layer that is rarely discussed. The Law does not only establish who is in an offside position; it establishes when that position becomes an offence. A player two metres beyond the line who does not take part in play commits no offence. The referee must separate three states: interfering with an opponent, challenging an opponent, and gaining an advantage from that position. Those three states are why a forward standing completely beyond the line can still be legal if the ball is played to the opposite flank.

According to FIFA data published after the 2026 World Cup, VAR was used to review hundreds of incidents and to overturn dozens of decisions, with the accuracy of key match decisions rising from around 95% to above 99%. What interests me is not the percentage but the fact that most reviewed incidents concerned offside and handball, two Laws sharing one problem: both rest on an anatomical boundary that cannot be seen.

In the V.League, VAR began appearing in selected matches from 2026 and has expanded in subsequent seasons. The biggest issue is not the line-drawing software but the camera infrastructure. A narrow offside needs a frontal, elevated, long-lens angle. A stadium with few cameras produces fewer situations that can be conclusively judged, and when data is insufficient the referee is forced to keep the on-field decision. In the A-League, where VAR has operated since 2026, disputes over drawn lines have followed the same trajectory: technology does not erase the argument, it moves the argument down to the level of definition.

When VAR first arrived in Vietnam, the immediate supporter reaction was to distrust the drawn line. I understand that reaction, and I think it is healthy. A line shown on television is a reconstructed image, not a source image. But distrust of an image should not become distrust of the Law, because those are two separate questions. Whether the line is accurate is a technical question, testable against camera data. Whether the line is reasonable is a question about the Law, and the answer sits in Law 11, not in the software. In Vietnam, the communication problem is larger than the technical one: when a VAR decision is delivered, most supporters receive only the final conclusion, not the line and not the source frame. A conclusion without an accompanying image is a conclusion that cannot be verified, and a conclusion that cannot be verified will always be doubted, however correct it is.

For viewers who want to assess a narrow situation themselves, I suggest a fixed sequence: find the frame in which the ball leaves the foot, identify the nearest defender's foot, identify the forward's shoulder, then identify the forward's armpit. If the only part beyond the defender sits below the armpit line, the situation is legal. I did this with a notebook and a pencil for three months, and it taught me more than any television debate.

I do not think supporters need to read all sixty pages of the Laws of the Game. But I think every viewer should know five points: the arm does not count, the shoulder does, level is onside, position only becomes an offence when a player takes part in play, and the final decision belongs to the referee on the field. Those five points explain most contested incidents in any competition, from the V.League to the Premier League.

The arm not being offside is not a loophole to be plugged. It is the logical consequence of a different question asked long before: which parts of the body can score a legal goal. If a goal scored with the hand is not recognised, then the hand cannot create a positional advantage either. IFAB is simply making two Laws consistent with each other. Supporters remember goals; I remember clauses.

The second counter-intuitive point is that the more accurate technology becomes, the more disputes arise rather than fewer. Once the error of the human eye is removed, the only remaining error is the error of the text. An offside of two centimetres is not a camera fault; it is the fault of a Law with no tolerance. Arsène Wenger proposed the principle that the whole body must be beyond, commonly called daylight, from 2026, and FIFA has trialled it at selected youth competitions since 2026. That is the only way to introduce tolerance into a Law that technology cannot make tolerant on its own.

Precedent archaeology helps here. When Christian Eriksen collapsed in the 43rd minute of Denmark against Finland on 12 June 2026, I reconstructed the Newcastle against Aston Villa case of 2026 for comparison and had to say plainly that I had heard at least three different versions of the 2026 story, none fully matching another. A precedent retold in many ways is a precedent missing its text. Law 11 is not in that condition. It is specific, it is clear, and that is precisely why it irritates: a clear sentence denies anyone the right to interpret by feel.

If FIFA accepts a tolerance principle, Law 11 will undergo its biggest amendment since 2026. If it does not, we will live with a system in which every goal can be re-examined within thirty seconds, and supporters will learn not to celebrate immediately. That is a cultural change larger than any technical one.

Three things can be done, and none of them requires amending the Laws of the Game. Publish the edition of the Laws a competition is applying immediately before the opening round, so supporters know which text they are watching. Show the frame and the drawn line on the stadium screen during the wait, rather than only the conclusion. And define a minimum tolerance, small enough not to encourage players to stand beyond, large enough that a drawn line does not decide a match.

Eight years after Kazan, every time I see a shoulder ahead of a defender's shoulder, I still pause for a beat. Before pointing at anyone, I ask myself whether I have read the whole contract. A match can be paused, but the responsibility of the person holding the whistle cannot.