Surface EMG: what it measures and how to read it in archery studies

What it records

When the nervous system activates a muscle, the muscle fibres produce small electrical impulses. Two electrodes placed on the skin over the muscle pick up the sum of those impulses, dampened by the tissue in between: this is surface electromyography, or EMG. In archery studies the electrodes are discs about a centimetre across, set a couple of centimetres apart along the line of the fibres, after the skin has been shaved, lightly abraded and cleaned; the signal is measured in microvolts.

The signal is good at telling you three things: when a muscle switches on or off, how hard it is working relative to a reference, and how its frequency content shifts as the muscle tires. It does not tell you directly how much force the muscle is producing, because the link between amplitude and force varies from one muscle to another and with fatigue, and it only sees muscles close to the skin clearly. On the archer's forearm, where the finger flexors and extensors are small and packed side by side, an electrode over one muscle also picks up some activity from its neighbours. This is known as cross-talk, and it is why electrode placement matters so much.

The earliest EMG study of archery in the registry dates from 1971, when Kamei and colleagues recorded 14 muscles of the neck, shoulders and arms in three kyudo practitioners of different standards: a fourth dan, a second dan and a beginner. Since then the most studied muscles have been the finger flexors and extensors of the drawing hand, the deltoid, trapezius, biceps and triceps; in recent years the rotator cuff and the trunk muscles have been added.

From signal to number

The raw signal swings above and below zero, so the first step is to rectify and smooth it. The result is a curve, the envelope, that rises and falls with muscle activity; its area over a time interval is the integrated EMG, and a root mean square over a moving window gives the RMS value. Then comes the choice of time window, and here no two archery studies are alike: two seconds around the release split into 100 ms intervals in Martin's 1990 study, one second either side of the clicker snap in Ertan's 2003 study, from 400 ms before to 800 ms after the snap in Simsek's 2014 study, the whole aiming phase in Jacquot's 2025 study.

To know where the clicker snap falls on the trace, you need a synchronisation signal: a switch under the clicker, an accelerometer on the bow, a microphone or a camera. Each marks a slightly different instant, and the difference matters when comparing timings across studies.

Normalisation

Raw amplitude depends on where the electrodes sit, on the thickness of the fat under the skin and on how the skin was prepared, so the microvolts of two people, or of the same person on two different days, cannot be compared as they stand. A 2024 Polish study shows this well: in the first measurement, taken before the hundred arrows of the protocol, the middle deltoid of the fourteen men in the sample gave a mean of 184.6 µV with a standard deviation of 126.0. That is why almost every study converts amplitude into a relative value.

The reference most often used in archery studies is the maximal voluntary contraction: before shooting, the archer pushes as hard as possible against resistance with each muscle, and activity during the shot is expressed as a percentage of that maximum. This was the choice of Martin in 1990, of Ertan's group in Ankara, of Shinohara in 2018 and of Jacquot in 2025. It only works if the maximal contraction really is maximal, in every muscle and every person, as the authors of a 2024 fatigue study point out; and it depends on the position in which it was measured.

Even after normalisation, the differences between people remain large. Among the fifteen archers studied by Martin, most of them on the US team, the coefficient of variation of mean activity between archers was around 52% for both the finger flexors and the extensors, while across six shots by the same archer it dropped to 15-16%. An archer looks very much like himself or herself and much less like anyone else, and that weighs on every comparison between groups.

The reference can also give itself away. In Simsek's study the beginners, in the 400 ms before the click, activated their finger flexors and extensors beyond their own maximum: a value above 100% means that the contraction measured beforehand was not the true maximum for that muscle in that position.

There are two other routes. When the shape of the curve matters more than its level, studies use indices that do not depend on scale, such as the variance ratio chosen by Soylu in 2006: it is 0 when successive shots produce identical curves and 1 when they bear no resemblance. When the interest is fatigue, the frequency of the signal is used instead of its amplitude: Leroyer in 1993 expressed the mean frequency of the deltoid as a percentage of its starting value and saw it fall steadily, by 0.72% per second on average, while seven archers held at full draw until exhaustion, for 52 seconds on average.

What it has shown in archers

The results are told in chronological order further down, in the section on archery EMG. Here they are grouped by what the EMG measured, because each kind of result has its own weak points.

Timing

This is the most robust use, because the moment a muscle switches on or off does not depend on normalisation. In his most skilled archers Nishizono saw, in 70-100% of shots, a brief silence in the deltoid about 110-120 ms after the clicker snap, with the release at 170-180 ms. Ertan, working with ten elite archers, ten beginners and ten people who had never shot, found that all of them opened the fingers by contracting the extensor, and that the extensor peaked about 100 ms after the click in the elite group, 200 ms in the beginners and 300 ms in the non-archers. Using the same tool, the Ankara group went on to validate a chronometer for clicker reaction time.

Patterns and activation levels

Martin found two different ways of releasing among his fifteen archers, each consistent from shot to shot: eight relaxed the flexors and extensors together, seven contracted the extensor just before the release. Thirteen years later, in Ertan's study, all of them contracted the extensor. Simsek and colleagues, testing twenty-seven recurve archers at three levels, found that the elite group used the forearm muscles less and the posterior deltoid and middle trapezius more, while the lower trapezius worked harder in the beginners and intermediate archers. Shinohara and Urabe, testing thirty male archers, found the opposite for the same muscle: in the elite group the lower trapezius worked harder, and the biceps and triceps less. On the bow arm, Ertan found a more active finger extensor in elite archers, which in the author's view keeps them from gripping the handle.

Shape and repeatability

In one of the four experiments by Clarys and colleagues in Brussels, the patterns of the three Olympic archers were identical from shot to shot in 92% of the muscles recorded, those of the six Belgian national archers in 41%, and those of the six beginners in none. In 2006 Soylu used the variance ratio to measure the repeatability of forearm curves and found it was related to FITA score (r = 0.72), but in a sample that included ten people who had never shot, as discussed below under reading errors.

Fatigue

In an Italian study from 1994 twelve archers from the Italian federation shot ten arrows before and after six sets of holds at full draw: the score did not change, lateral sway of the bow rose by 20% on average, and electrical activity tended to increase, most of all in the biceps of the less skilled archers. In 2024 a Polish group tested twenty traditional archers before, halfway through and at the end of a hundred arrows, using maximal contractions as the test.

Competition distances

Most measurements have been made in the laboratory or indoors, but Olympic distances have been there from the start. In Clarys's first experiment eight skilled archers shot at 18, 25, 50, 70 and 90 m: activity rose significantly only between 25 and 50 m, and did not increase in proportion to distance. In 2025 Jacquot and colleagues recorded 14 muscles in eight top French archers, 72 arrows each at 70 m with their own equipment: across the group, seven muscles in both arms and the trunk changed their activity with score during aiming, while other parameters changed only in some archers, and not always in the same direction: aiming time, for example, was shorter on one archer's worst shots and longer on another's.

Common reading errors

The first is reading activity as force or as quality. More microvolts, or a higher percentage of maximal contraction, mean the muscle is more active, not that it is pulling harder or working better. Nor does a difference between groups tell you what to train: Shinohara and Urabe conclude that the lower trapezius needs training, but theirs is a one-off comparison between archers of different levels, not a training study.

The second is comparing groups that are too far apart. In the 2005 study of skill indices, Ertan, Soylu and Korkusuz found a correlation of 0.74 between an EMG index and the ranking of FITA scores, but the sample included seven elite archers, six beginners and ten people who had never shot, who were assigned a score of 250. There is a second reason not to take that 0.74 as a measurement: the four time windows the index is computed over were not fixed in advance, the authors chose them by maximising the squared correlation with the ranking of the scores, and without validation on separate data a number arrived at that way is the most that fits that sample. The 2006 repeatability study uses a group with exactly the same characteristics, and there the significant differences are between archers and non-archers, not between elite archers and beginners; the authors themselves write that the index is useful for groups but not for assessing individuals. A correlation built on people who have never shot says little about the difference between two club archers.

The third is counting arrows as if they were people. Several studies analyse dozens or hundreds of arrows from a handful of archers as independent observations, which makes even small differences significant. Erkek and Tinazci's two papers on four archers do this, and so does Jacquot, who compares 26 parameters across the arrows of eight archers.

The fourth is taking the average for everyone's technique. Martin found two release patterns in the same group, Edelmann-Nusser and Gollhofer very different techniques in nine German national team archers, and Jacquot individual strategies among top archers: a group-average curve may resemble none of the archers who make it up.

The fifth concerns timing. In the same group of twenty elite archers, extensor EMG gave a clicker reaction time of 146 ms and the mechanical chronometer 177 ms: the muscle switches on before the release, while the chronometer stops only once the arrow is already moving. When comparing timings across studies you need to know which instant was marked.

The last concerns conditions. Some studies do not measure real shooting: in Lin's 2010 study ten novices held various draw positions for seven seconds with a 26 lb recurve without shooting, and in the Polish fatigue study the EMG came from maximal contractions performed before, halfway through and after the hundred arrows. These measurements are useful, but they do not describe the muscles during a shot at 70 m. Fatigue also needs care: amplitude can rise or fall, and the authors of the Polish study warn that its trend is not always consistent; a downward shift in frequency is a more specific sign.

What to check in an archery EMG study

• which muscles, on which side, and where the electrodes were placed (ideally following a stated procedure such as the SENIAM recommendations);
• how amplitude was normalised, and in what position the maximal contraction was measured;
• which time window was analysed, and how the clicker snap was synchronised;
• how many archers and how many arrows, and whether arrows were treated as independent observations;
• the archers' level and discipline, and whether non-archers were included in the comparison;
• distance and conditions: real or simulated shooting, indoors or outdoors;
• whether the result applies to the group or to each archer individually.

Samples in these studies range from a single archer to around thirty people. How the registry weighs design, sample and journal is explained in How we read a study.

@misc{campagna2026_enresearchmethodssurfaceemg,
  author = {Campagna, Matteo},
  title = {Surface EMG in archery: what it measures and how to read it},
  year = {2026},
  howpublished = {Research, matteocampagna.com},
  url = {https://matteocampagna.com/en/research/methods/surface-emg},
  note = {accessed 2026-10-16}
}