"Keep the shoulder down" is one of the first corrections an archer hears, and it usually comes with another about the same spot: the bow elbow should be rotated so that it does not point at the ground and the string clears the forearm. The third, more general, puts the shoulders on the line of the arrow at full draw, without the bow shoulder rising. Anyone who has shot for a few years also knows the problem behind all three: a bow shoulder that rises on its own, sometimes during the draw, more often towards the end of the aim or after a lot of arrows.
There are fewer measurements of this part of the body than you might expect: the X-rays of expert archers are half a century old, the comparison between aligned and tilted shoulders involves two archers, and the studies disagree on how the work is shared between the two sides of the body. The claims that come out of them are mostly reported in the registry, meaning they rest on little and small evidence, or else contested, and none links the height or rotation of the bow shoulder to the score of the individual arrow.
What the bow shoulder holds
At full draw the bow pushes into the hand with the whole force of the draw, and that push reaches the shoulder through the straight arm. To stay where it is, the arm has to hold its abduction, the angle that keeps it raised out to the side, against a load that does not ease while the arrow is on the string. This is the mechanism a group of Japanese orthopaedic surgeons reasoned about in 1976, when they compared, with photographs, electromyography and X-rays, four university archers with more than three years of practice and six young men who had never held a bow.
In the beginners the picture was one every coach knows: the bow arm did not reach the abduction it needed, the left shoulder was pushed up, the trunk and neck leaned back and, with the bow held away from the body, the neck bent forward to aim along the arrow as you would with a rifle. The authors link these faults with a mechanical argument. With the shoulder rotated outwards the deltoid, the strongest muscle for holding the arm up, does not work effectively, and the arm is held up almost entirely by the front of the deltoid and the biceps; so it gives way to the pressure of the bow, abduction decreases and the shoulder blade is pushed upwards. The dropping arm in turn lowers the point of the arrow, and to bring it back onto the target the beginner leans back.
It is a plausible chain, but it remains an argument: there are no figures and the sequence is not measured. The registry therefore says only that in a beginner, posture faults come together, a reported claim; whether they spread from one part of the body to the others, starting from shoulder rotation, nobody has tested.
Rotation of the shoulder and forearm
The X-rays in the same study look at the bow arm from the side, at full draw. In the experts the radius and ulna, the two bones of the forearm, stayed parallel: the forearm was supinated, turned upwards, and the shoulder was in neutral rotation, turned neither out nor in. In the beginners the two bones crossed at the elbow, with the forearm pronated, the shoulder rotated outwards and the crease of the elbow facing up. The authors also describe the order of the movements: the expert straightens the elbow from a shoulder already in neutral rotation, the beginner straightens it with the shoulder rotated outwards. They add that doing it deliberately is harder than it looks, and is often the first hurdle for a newcomer, but that once the movement is understood even a beginner manages it easily.
To explain why it matters, the study goes back to strength: with the shoulder in neutral rotation the deltoid takes its full part in holding the arm up, and with the forearm supinated the biceps contributes to abduction as well. On some of the subjects, without saying how many, the authors measured the shoulder's abduction force in the two positions and found it about a fifth higher in neutral rotation than in external rotation, with no statistics.
Since then there have been few measurements of bow shoulder rotation. In 2025 eighteen high-level Bulgarian archers shot six arrows at 18 m with inertial sensors on their shoulders: external rotation of the left shoulder was the most variable parameter of all, with standard deviations of up to 53 degrees, and according to the author those who let it rotate outwards at the release sent their arrows further from the centre, but no statistical test backs this up. The claim that the bow shoulder works better in neutral rotation with the forearm supinated therefore remains reported: it has a mechanical logic and the X-rays of a single study from 1976, not a well-measured result in today's archers.
The shoulder line and the two sides of the back
The rule about keeping the shoulders on the line of the arrow has a single measurement behind it. In 1994 an Italian group studied twelve federation archers with motion analysis, electromyography and a force plate, and two high-level archers among them, with almost identical FITA scores of 1,350 and 1,330, shot in two different ways. Two tenths of a second before the clicker went off, the first had an almost perfectly aligned shoulder line, the back muscles on both sides working at the same level and the biceps barely active. The second had the shoulder line tilted relative to the bow arm and the spine at an angle, with the left side of the back more active than the right and the biceps very active.
The authors add a reading that bears directly on the rising shoulder: the second archer's asymmetry might come from trying to keep the left shoulder down in the last part of the aim. It is their interpretation, not a measurement, but it suggests that a shoulder held down by force comes at a cost. In the same group, after a fatigue protocol of sixty 20-second holds at full draw, five archers out of twelve, three intermediate and two high-level, raised the activity of the upper back muscles on the left side, and in them the authors saw a link with the alignment of the shoulder line relative to the bow arm, described rather than quantified.
The claim that with the shoulder line aligned the two sides of the back work at the same level, and with the shoulders tilted they do not is reported: it rests on two cases set side by side, and the same study notes that each archer has a muscular configuration of their own.
When archers are asked what they think about, balance between the two sides comes to the fore. In an open-ended questionnaire, 65 competitive Korean archers described what had led to their shots in the X: for drawing and anchoring, good balance on both sides gathered 50 mentions, against 14 for correct arm position and 8 for the absence of shaking. These are mentions, not people, and they tell us what archers from a single school remember, not that thinking about balance makes anyone shoot better; so the claim that in the draw, competitive archers say they think about the balance between the two sides more than about a steady arm is reported.
How hard the bow side works
You often hear that the bow arm just has to hold, while the string side does the work, but electromyography of both sides tells a more complicated story. The most complete study, on five Russian recurve masters of sport, with 18 muscles per side and 180 shots at 18 m, calculates an asymmetry coefficient for each muscle, negative when the bow side dominates and positive when the string side does. Over the whole shot the string side dominates in the forearm, the biceps, the erector spinae and even the tensor fasciae latae at the hip; the bow side dominates in the serratus anterior, the pectoralis major, the latissimus dorsi and the upper trapezius.
The breakdown by phase is what concerns the bow shoulder. During the draw the bow side dominates in the muscles that raise the bow and bring it onto the target, namely the upper and lower trapezius, the middle deltoid and the serratus anterior, with coefficients between -0.18 and -0.23. The middle deltoid then becomes symmetrical and in the follow-through dominates on the string side, while the serratus anterior, which holds the shoulder blade against the ribcage, goes the other way: its bow-side dominance rises to -0.66 in the last part of the draw and to -0.83 at the release, then falls to -0.36 in the follow-through.
Not every study points the same way. In nine recreational archers, using a recurve of 0.45 kg draw weight shot at just over six metres, the two sides worked at similar overall levels with different muscles, and the bow arm used more of the lateral deltoid and the triceps. This is why the claim that in the shot the drawing side works harder than the bow side is contested. On one point, though, the measurements agree: the bow arm is not passive. As early as 1993, in ten Japanese club archers, the bow-side deltoid was the most active of the six muscles recorded during aiming, and in eight elite French archers, in the three for whom the figure is given, the middle deltoid of the bow arm worked at between 46.5% and 65.5% of maximum contraction while aiming, a figure the registry keeps in the entry on the shoulder muscles that work hardest while aiming.
The work also carries on after the release: in the 1976 recordings the bow shoulder muscles of the experts stayed active through the follow-through, while in the beginners they cut out at the release and the bow swung down or to the side. Whether this goes with better arrows is a different question, on which the studies of individual arrows point in different directions: the claim that keeping the shoulder muscles active through the follow-through is linked to better arrows is contested.
Why the shoulder rises
Taken together, these measurements do not give a measured cause, but they show where to look: at rotation, if you follow the 1976 argument; at a shoulder held down by force, if you follow the reading of the Italian comparison; at fatigue, which after the holds at full draw changed the activity of the upper back muscles on the left side in five archers out of twelve.
That leaves draw weight, the first suspect when the shoulder rises, and here the data come from a single group: thirteen university students shot bows of 20, 25 and 32 lb, in the same session and always from the lightest to the heaviest. In the women, going from 25 to 32 lb, the trunk arched back more than twice as much, from about -8 to -17 degrees, while in the men no comparison was significant, and flexion-extension and rotation of the bow shoulder showed no difference between draw weights or between the sexes. With six or seven people per group a null result does not rule out an effect, and the claim that with a heavier bow the angles of trunk, shoulder and elbow change is contested. How far the bow shoulder rises, in degrees or centimetres, none of these studies has measured.
Feeling where the shoulder is
Many archers correct the shoulder by feel, and there are a few measurements of how reliable that feel is. Thirty women club archers in Tehran, lying down with their eyes closed, had to find an angle of external rotation of the drawing shoulder again: at 60 degrees they stopped on average almost six degrees short, with individual errors ranging from 18 degrees short to 5.6 beyond, and at 30 degrees the mean was close to zero, but with errors in both directions. In seventeen right-handed children at the end of a beginners' course, the angles of the right shoulder, the drawing one, reproduced with eyes closed at 60 and 90 degrees tended to fall short of the one requested. These are small measurements, in regional journals, taken outside the shooting posture, and the registry sums them up as follows: with their eyes closed, archers reproduce a shoulder angle with errors of a few degrees, a reported claim.
For the bow shoulder there are even fewer measurements. The only comparison with non-archers, on twenty-two recurve archers aged 15 to 19, puts the archers slightly ahead, but it does not say whether its figures are errors or reproduced angles and appears in a journal with questionable editorial standards; so the idea that archers have a more accurate sense of bow-shoulder position than non-archers remains a question the registry lists as not studied. In thirty professional compound archers, followed through twelve weeks of balance exercises, the error in finding the bow shoulder raised to 125 degrees again was between three and four and a half degrees, and those who reduced it most in flexion improved their score most, but in only one of the six measurements; in the seventeen children on the beginners' course there was no link with their arrows. The claim that a finer sense of shoulder position goes with better scores is therefore contested.
Where the measurements stop
Almost everything that is taught about the bow shoulder rests on samples of two to eighteen archers. No study has set the height or rotation of the bow shoulder at full draw beside the score of each arrow with a statistical test, and nobody has tried correcting it in a group of archers to see whether the arrows group more tightly.
The health side should be kept apart as well. Studies of archers put the shoulder first among the sites of overuse injury, a reported claim, but they often place it on the drawing side and in the drawing phase, and none of them links a high or rotated bow shoulder to an injury. The data on pain are on the page about injuries in archery and in the entry on the shoulder as the most common site of overuse injury.
What I take to the shooting line
These measurements do not offer a shoulder position to copy, and none of them comes from an experiment that changed an archer's shoulder and then measured the score. I keep that in mind in the three points I take from them.
The first concerns rotation, and it has the soundest logic, even though the evidence is old. In the experts' X-rays the shoulder was already in neutral rotation when the elbow straightened, and the forearm stayed supinated; in the beginners the elbow straightened with the shoulder rotated outwards. So when the shoulder rises, I would look first at the order in which the bow arm settles during set-up, and only then at the height of the shoulder at full draw: by the reasoning of that study, if the rotation comes late the shoulder has to hold the bow in the position where the deltoid works least well.
The second concerns how to correct it. Pushing the shoulder down by force in the last part of the aim is what, in the interpretation of the authors of the Italian comparison, may have unbalanced the second archer's back, so I would look for a position in which the shoulder stays low without having to be held there. When it rises also matters: if it happens only in the last ends of a session, fatigue should be considered before changing the technique, since after the holds at full draw it was the activity of the upper back muscles on the left side that changed; if it appears after an increase in draw weight, the little data there is does not allow the bow to be ruled out.
The third concerns checking. An error of a few degrees in finding a shoulder angle again means that the feeling of having the shoulder down is not enough to know where it is. A photo or video from the front, at full draw and always from the same angle, shows the shoulder line better than feel does, and comparing the first and last ends tells you whether the shoulder rises as you tire. I cover this on the page about video analysis and self-coaching, and it is part of the repeatability of the shot. Nobody has measured what a shoulder a few degrees lower is worth in points, so the aim I set myself is for the shoulder to stay the same from one arrow to the next, rather than at a particular height.