Trunk muscles in kyudo: Matsunaga 2017 study

Nine men who practise kyudo shot with surface electromyography on eight trunk muscles, four on each side: rectus abdominis, external oblique, internal oblique together with transversus abdominis, and erector spinae. The signal from each shot was decomposed by non-negative matrix factorisation, which looks for the smallest number of muscle groups firing together that is enough to account for what was recorded.

In all nine, two groups suffice, and the authors call them modules. The first is the same at both levels of practice: the scalar product measuring how closely the muscle weightings match comes to 0.99, and the module is driven by the erector spinae on both sides during the first half of the shooting sequence, where the arms rise and the bow opens. In the second half, that same module stays more active in the higher-grade group than in the lower-grade one.

The second module is where the two groups part: the scalar product falls to 0.44. In the four higher-grade archers the module is dominated by internal oblique with transversus abdominis, in the five lower-grade archers by external oblique. The moment the module switches on is similar in both groups, in the second half of the shot: what changes is which muscles make it up, not when it comes in. The authors read the difference as a different strategy for holding the trunk still, and note that in the lower-grade group individual differences in external oblique were large, a sign of several ways of coping. The title calls it a pilot study, and the conclusion is written as a hypothesis: internal oblique and transversus activity during the shot may bear on performance.

The nine participants averaged 22 years of age, 1.70 metres in height, 65.2 kilograms and seven years of practice, and trained once or twice a week. The split into groups follows the grade in the kyudo proficiency examinations: four held level 4 or above and were called the elite group, five held level 3 or below and were called the novice group.

After some practice shots, each archer shot four arrows at a 36 centimetre target set 28 metres away, and the two best shots, those that hit the centre, were analysed. The signal was collected at 1,000 samples a second with a wireless telemetry system, using three by two centimetre adhesive electrodes two centimetres apart, placed at the usual landmarks: three centimetres lateral to the navel for rectus abdominis, midway between the costal margin and the iliac crest for external oblique, two centimetres below and medial to the anterior superior iliac spine for internal oblique and transversus, and three centimetres lateral to the fourth lumbar vertebra for erector spinae. The skin was abraded and cleaned with alcohol to bring impedance below two thousand ohms.

Before the trial each muscle was taken to maximal voluntary contraction against manual resistance, held for three seconds, to provide the reference for normalisation. A camera in front of the archer, running at 240 frames per second and synchronised with the electromyography, marked the start and end of each shot. The signal was filtered between 20 and 450 hertz, rectified, normalised to the root mean square of the maximal contraction and interpolated to two hundred time points, so that every shot had the same length. The factorisation was repeated asking for one to eight modules, keeping the lowest number that accounted for more than 90 per cent of the variance; two modules were enough. Two modules counted as the same when the scalar product of their weightings exceeded 0.75.

It is a pilot study, and the title says so. Nine men, four against five, two shots analysed each: with numbers like these there is no statistical test between groups, and the comparison rests on a single similarity index computed on each group's mean weightings. A low index says the two mean patterns differ, not how likely it is that the difference would recur in other archers.

The muscle at the centre of the hypothesis was not measured on its own. Surface electromyography at that site picks up internal oblique and transversus abdominis together, and the authors indeed write of a single channel: the whole argument about transversus, the deep muscle the hypothesis starts from, rests on a signal that also contains the muscle lying over it.

Nobody measured performance. Only shots that hit the centre were selected for analysis, so there is no comparison between good and poor shots, and group membership was defined by examination grade rather than by score. The sentence about a possible bearing on performance is a hypothesis of the authors that their data do not test. Trunk motion was not analysed, because the camera stood in front rather than to the side, and the authors give this as a limitation: some participants shot with the lumbar spine extended, which is another route to a stiff trunk and one that stays outside the measurement.

The action is the kyudo shot, on a 36 centimetre target at 28 metres, with a sequence codified in eight stages and the trunk and legs held still. Transfer to an Olympic bow, where the distance, the clicker and the posture are all different, is not studied here. The archers called elite trained once or twice a week: the label reports a grade, not a training load.

The contribution of this work is a way of looking rather than a result. Instead of asking how hard a muscle fires, it asks how many activation patterns are needed to describe a shot: two are enough, and the first, the one from the back during the draw, is identical at both levels of practice. The difference between the more and the less advanced archers does not sit in the first half of the shot but in how the trunk is held still once the bow is open.

For someone who shoots, the honest thing to say is that this study points to no exercise. Nine kyudo practitioners, two shots each and no scores: there is not the material to argue that training the deep abdominal muscles raises points, nor to say which of the two patterns is preferable. The authors themselves offer the link with performance as a hypothesis to be tested.

What remains is a hint about where to look. If two archers hold the same position with different muscles, a check that stops at the eye or at a photograph will not separate the two cases: the position can be identical while the way it is held up is not. It is also a reminder of why, in a group of beginners, the same apparent fault may have different origins from one person to the next.