Shot consistency in archers: Zhang 2024 study

Twenty-eight Chinese archers at three levels each shot twelve arrows at 30 metres with cameras and electrodes on the drawing arm. What was compared is not the movement but how much the movement changed from one arrow to the next, measured as a coefficient of variation.

In the draw phase, the elite and intermediate archers varied less than the beginners in the activation of five muscles out of eight: the flexor digitorum superficialis, with coefficients of 9.8 and 10.7% against 20.5; the biceps, 10.0 and 13.0 against 21.9; the posterior deltoid, 9.0 and 11.7 against 21.5; the middle deltoid, 9.4 and 13.1 against 23.5; the middle trapezius, 8.6 and 11.5 against 22.5. The extensor digitorum, the long head of triceps and the latissimus dorsi did not separate the groups.

In the aiming phase no muscle separated the groups, and every coefficient sat between 41 and 63%, four or five times the draw-phase figures. It is the most striking number in the paper and nobody comments on it: at full draw, muscle activity varies enormously from one arrow to the next, at any level.

On the joint angles, the only measure that came out is shoulder abduction: coefficients of variation of 1.1, 0.8 and 1.2% at the peak and 1.0, 0.8 and 1.5% at the trough, with the two trained groups marked as lower than the beginners. Elbow extension gave no differences, and the variation of the trough of shoulder extension was so scattered (18.5 plus or minus 61.5 in the elite archers, 15.7 plus or minus 50.6 in the intermediates, 28.5 plus or minus 77.8 in the beginners) that the measure says nothing.

The eight elite archers are the whole of the Chinese national team training between 2018 and 2020 for the Tokyo Games; the twelve intermediates and eight beginners come from provincial squads. Mean age was 19.6 years plus or minus 3.8 in the elite group, 17.4 plus or minus 1.3 in the intermediates and 16.0 plus or minus 1.3 in the beginners; draw weight 46.3 plus or minus 4.8, 42.1 plus or minus 5.2 and 42.3 plus or minus 4.2 pounds; years of practice 4.9 plus or minus 2.0, 3.5 plus or minus 1.9 and 2.2 plus or minus 0.8. None had had an upper limb injury in the preceding six months.

Eight Qualisys infrared cameras at 200 hertz tracked eighteen reflective markers placed according to Rab's upper limb model (top of the head, seventh cervical vertebra, sternum, sacrum, ears, acromions, coracoid processes, radial and ulnar styloids, third metacarpophalangeal joints and anterior superior iliac spines) plus three markers on the bow. A side camera at 100 hertz established the moment the arrow leaves the string. Surface electromyography (Delsys) at 2,000 hertz recorded eight muscles on the drawing side.

The recordings were made in an ordinary training hall, with the target at 30 metres and five minutes of warm-up. Each archer provided twelve valid trials, 336 in all: valid meant that the coach judged the movement acceptable and that all the parameters had been captured.

The draw runs from the bow's highest point to the anchoring moment, defined as the first frame in which the lateral point of the right elbow moves less than 0.5 millimetres fore and aft for five consecutive frames; the aim runs from there to the arrow leaving the string. Angles were computed in three dimensions with Euler angles in Visual3D, and the coefficient of variation was taken on the peaks and troughs of the shoulder and elbow angles. The muscle signal was filtered between 5 and 450 hertz with a fourth-order Butterworth, rectified into an envelope, normalised to each muscle's maximum over the twelve arrows, integrated per phase, and the coefficient of variation of that integral was taken across the twelve arrows. Comparisons by one-way analysis of variance and the LSD test.

Whom the paper calls a beginner has 2.2 years of practice and shoots a 42-pound bow in a provincial squad. That is not a beginner in any ordinary sense, and the elite archers average 4.9 years of practice: the comparison runs between athletes who are close together, not between extremes.

There are two problems with the numbers. The first is an inconsistency: the methods say the coefficients of variation were computed for the drawing arm, Table 2 is headed for the drawing arm, and then the results and the discussion both speak of shoulder abduction of the bow arm. The two cannot both hold, and the paper does not make clear which is meant. The second concerns significance: the peak abduction values marked as significant are 1.1% in the elite group against 1.2% in the beginners, with standard deviations of 0.7 and 0.5 in groups of eight people. A tenth of a percentage point does not separate those distributions, and the paper gives neither F values nor exact p-values for any of the comparisons. They can be recomputed, though, from the published means and standard deviations with the stated procedure, and the result is this: for peak abduction the analysis of variance gives F = 1.59 on 2 and 25 degrees of freedom, that is p = 0.23, so there would have been nothing to compare post hoc, and elite against beginners comes out at p = 0.71; for trough abduction the omnibus test sits just inside the threshold (F = 3.41, p = 0.049) and of the two marked comparisons only the one between intermediates and beginners holds, at p = 0.016, while elite against beginners comes out at p = 0.10. Of the four comparisons marked in the angle table, then, only one stands. The muscle activation table, by contrast, holds up: for all five marked muscles both comparisons come out below 0.05, with the single caveat that for the finger flexor the omnibus test stops at p = 0.055, so those two comparisons follow a non-significant overall test.

No score was measured: performance is group membership. And twenty-two comparisons were made across three groups with no correction for the number of tests.

The abstract says the differences were most pronounced in the flexor muscle groups, while the table shows the deltoids and the middle trapezius separating the groups just as much as the finger flexor and the biceps.

One last technical note: the paper describes all twenty-eight archers as left-handed, and then explains that they held the bow in the left hand and drew with the right, which is what is usually called a right-handed archer. It is a slip in translation, but it can confuse anyone reading only the abstract.

The sample (eight, twelve and eight archers, the Chinese national team and provincial squads, twelve arrows at 30 metres, the same electromyography system) matches the one in another paper by the same author on muscle synergies, where the beginners are described as coming from city clubs rather than provincial squads. The registry links this record to a third paper on the same data. The journal, finally: the text was received on 20 November 2024, revised on 19 December, accepted on that same 19 December and published the following day.

What this study describes is consistency, not quality. It does not say how a strong archer moves, it says how closely the movement repeats from one arrow to the next, and the difference between the two ideas is the whole scope of the paper.

The honest reading is a narrow one: among these twenty-eight, the higher-level archers repeated the muscle activation of the draw more closely, in five muscles out of eight, and the muscles in question cover the whole chain, from the fingers to the elbow to the shoulder to the shoulder blade. It is not one muscle in particular: it is the whole draw, repeated more closely.

The half of the message nobody comments on sits in the aiming phase. There nothing separates anyone, and everybody's coefficients run between 41 and 63%, four or five times the draw-phase figures. Measured this way, over twelve arrows, full draw is the variable part of the shot for everyone, the national team included. That is worth knowing before going looking for consistency in the hold.

Since no score was measured, this record cannot say that repeating yourself more closely brings points: it can say that in this sample the archers with more years behind them did repeat themselves more closely. The causal order was not touched here, and an archer might repeat themselves more closely because they shoot better, rather than shooting better because they repeat themselves.

The other useful part is the method rather than the result: the coefficient of variation of muscle activity over a dozen arrows is a simple measure, obtained from a single sensor, and here it separates the levels better than the joint angles do. Twelve arrows, after all, are two ends.