Shoulder position and archery accuracy: Ivanov 2025

Eighteen Bulgarian archers, described as high level, each shot six arrows at 18 metres indoors with inertial sensors on their shoulders and arms. The paper reports, arrow by arrow, the mean, minimum, maximum, standard deviation and coefficient of variation of eight shoulder angles, along with the mean score of each end.

Some angles repeat well from one arrow to the next. Total flexion of the left shoulder, which on a right-handed archer is the bow shoulder, runs between 74.3 and 75.5 degrees, with a coefficient of variation between 6.2 and 7.7 per cent; abduction of the same shoulder runs between 72.5 and 74.5 degrees, varying between 5.8 and 7.6 per cent; total flexion of the right shoulder stays around 119 degrees, varying between 8.9 and 10.5 per cent. Other angles are far more scattered: flexion of the left shoulder, with means between 53.0 and 55.9 degrees, has a coefficient of variation between 29.0 and 33.3 per cent, and external rotation of the right shoulder, between minus 45.7 and minus 49.4 degrees, reaches between 52.8 and 65.8 per cent.

The extreme case is external rotation of the left shoulder. Its mean shifts from 3.4 to 18.1 degrees across arrows, with a standard deviation of about 50 degrees, minima falling to minus 129 degrees and maxima above 160. Over the same series the mean score per arrow falls from 8.3 on the first end to 7.6 on the sixth, by way of 7.5 on the second and 8.5 on the third, and arrows off the target face go from none on the first end to two from the fourth onwards. The author concludes that external rotation of the bow shoulder at the moment of release is the parameter with the strongest bearing on shots away from the centre, and that holding that shoulder in place until the string goes accompanies the more accurate shots.

The measurements were taken in an 18 metre hall, chosen to remove wind, rain and uneven ground from the picture. The target face was 40 centimetres across, with three spots set vertically. The bows were assumed to be well balanced, meaning the author took the equipment not to affect the stability of the archer and bow together, and does not describe them: draw weight, bow type and discipline do not appear.

Kinematics were recorded with the Noraxon MyoMotion inertial system, in its 2014 version, over six shots per archer. Eight parameters were collected: total flexion of the left and right shoulder, flexion of the left and the right, abduction of the left and the right, and external rotation of the left and the right. The paper does not say where the individual sensors were fixed, how calibration was done, or how the zero of each angle was defined.

The analysis is descriptive statistics only: for each of the eight angles and each of the six arrows the paper gives the mean, minimum, maximum, standard deviation and coefficient of variation, and discusses every parameter with two charts, one of means and one of the coefficient of variation. The abstract also announces a correlation analysis, but no correlation coefficient appears anywhere in the text. Scores sit in a separate table, giving the mean score of the end and the number of arrows off the face, without figures for individual archers.

The weakest point is the jump from the data to the conclusion. Angles and scores sit in two separate tables, one per arrow across the eighteen archers and the other per end across the group: there is no comparison between an individual archer's shoulder rotation and the score of that archer's shot. The sentence about bow shoulder rotation spoiling accuracy is the author's reading rather than a measured relationship, and this design cannot separate a cause from a coincidence.

Two inconsistencies sit in the numbers, visible in the paper itself. The coefficient of variation column for external rotation of the left shoulder reads 0.0 for all six arrows, which cannot follow from a standard deviation of about 50 degrees on a mean of a few degrees. And the comment on right shoulder abduction says the coefficient of variation is the highest of all at 35 per cent, then quotes values above 100 per cent that are the angle means copied in place of percentages.

A good deal is left out. Eighteen archers and six arrows each make a short series for a claim about repeatability, which at competition distance is judged over dozens of ends. The level is not defined beyond the adjective, and nothing says whether the archers were right or left handed, so reading left and right as bow and drawing side is our own inference. There is no statistical test, no comparison between more and less accurate archers, and no measurement of the bow. The author gives as his own limitation that the trials were held indoors, and that outdoor results might differ. The venue is a supplement of a regional university journal.

For someone who shoots, the firmer part of this work is not the conclusion but the table. Across six consecutive arrows, the angles describing how high and how open the bow arm sits repeat within a few percentage points, while rotation of the shoulder about its own axis moves by tens of degrees within the same archer. That is a description of what repeats by itself over a short series and what does not, and it holds even though the explanation the author offers remains unsupported.

The study does not show that correcting that rotation raises a score, since nobody compared an archer's angles with that archer's points, and it does not say which position is preferable, since no reference value is given. For a coach, the reminder in it is that a stable mean angle does not guarantee a still shoulder: the mean of a rotation swinging between minus 129 and plus 163 degrees describes almost nothing of what actually happened.