Back tension in a top archer: Yi 2007 study

The study follows a single Korean recurve archer, then ranked second in the FITA world rankings (185 cm, 82 kg, aged 21, with 12 years of practice), over 60 arrows at 70 metres recorded with a motion analysis system. The aim was to describe his back tension, taken as the movement of the shoulder blade and shoulder from anchor to the end of the follow-through, and to see which measurements went with the score of each arrow (10, 9 or 8).

From release until the arrow left the bow, the marker on the bow did not move detectably (0.00 mm both forwards and vertically), while in a university archer used for comparison it moved by fractions of a millimetre. The top archer's drawing hand moved back by 1.24 mm and vertically by 0.90 mm. Mean shoulder and trunk angles were almost identical for arrows scoring 10, 9 and 8: the horizontal shoulder angle in the follow-through, for example, was 32.36, 30.45 and 31.43 degrees. Arrows scoring 10 did, however, show the lowest variability in the follow-through on most measures, and the angular velocity of the shoulder after release followed the same pattern for 10s and 9s and a different one for 8s.

In the regression, the horizontal angular velocity of the shoulder after release and the vertical shoulder angle in the same phase together explained 15.6% of the variation in score; with four variables this rose to 16.3%. The authors conclude that this archer's shot is very stable, that the horizontal angular velocity of the shoulder affects the score and that, beyond kinematics, unmeasured factors are at play.

Recording took place outdoors, at 70 metres on a 122 cm FITA target face, with a mean wind of 2.0 m/s and the archer's own equipment: five series of 12 arrows, each in four ends of three. Eleven reflective markers (ten on the trunk and upper limbs, one on the bow next to the pressure button) were tracked by seven Qualisys infrared cameras at 200 frames per second; two high-speed cameras at 1,000 frames per second filmed the bow and the hand. The data were filtered at 30 Hz.

The shot is split into two phases, from anchor to release and from release to the end of the follow-through. For each phase the authors calculated the horizontal and vertical angles of the drawing shoulder, their angular velocities and the rotation of the trunk in the horizontal plane, and compared them across arrows scoring 10, 9 and 8. With expert coaches they then chose eight variables for a multiple regression on score, using both stepwise and forced entry, and also tried curvilinear models. For bow and hand displacement there is a comparison with a university archer.

This is a study of one archer, and the authors say so: the results cannot be generalised, and psychological and physiological factors are left out. On a day with 2 m/s of wind, an arrow's score also depends on things that shoulder movement does not capture, and the number of arrows scoring 10, 9 and 8 is not given. The comparison with the university archer also rests on a single archer, about whom nothing else is reported.

Displacements to a hundredth of a millimetre are at the limit of what two cameras can measure, and the paper gives no measurement error, so the 0.00 mm should be read as movement that was not detected. The sections are not always consistent: the summary describes a five-variable model explaining 17.9%, while the results report 15.6% and 16.3%; one of the conclusions gives vertical angular velocity a curvilinear model, while the results say the linear one fits better; and some displacement values differ between text and table. Finally, the relationships found are weak: taken on its own, no variable explains more than 10% of the variation in score (the horizontal angular velocity of the shoulder, the only one significant in both models, 7.7%).

The study shows what a world-class shot looks like up close. From anchor to release the drawing shoulder moves by only a few degrees and always in the same way, and the mean angles barely change between arrows scoring 10 and 8. What set the better arrows slightly apart was the continuity of movement after release: shoulder angular velocity followed the same pattern for 10s and 9s, and on most measures the 10s were the least variable. This fits the view, common among Korean coaches, that the follow-through continues the back tension and shows its quality.

For any other archer, though, these figures are not a model to copy: they apply to one archer whose technique was already very stable, and they explain less than a fifth of the differences in score. The clearest result, if anything, is how little of the score depended on what shoulder kinematics could detect.