Showing beginner archers their form: Miyauchi 2012

Two engineers at a Japanese college of technology built four devices to show beginners what a coach would normally point out in words, and arranged them into three stages of practice. The starting problem is a practical one: in Japanese school clubs it is rare to find a coach who knows archery, and a beginner cannot recognise on his own how his bow hand moves.

The first two devices are training sights. The tilt and bearing of the bow are recorded on shots that hit the centre, that position is taken as the reference, and the device signals when the bow comes back into it, within a tight tolerance such as one degree: one does so with a matrix of lights mounted beside the sight, where the lit point moves towards the middle as the bow approaches the reference, the other with sound, a scale rising by an octave as the bow closes in and a cue at the right point. The third records the tilt, twist and bearing of the bow arm every tenth of a second for ten seconds and plays them back as a graph and an animation, with the six classic release faults: the arm swinging up, dropping, coming back, falling forward, and twisting either way. The fourth widens the measurement to six joints, the shoulder, elbow and wrist of both arms, and stacks on a single chart the deviation from the reference position over thirty-six shots, that is six ends of six.

The test of effectiveness covers only the two sights. Beginners with less than a year of practice shot twelve ends of three arrows with and without the device, and grouping was measured as the geometric mean of the three distances between the three holes. With the light sight at 30 metres the geometric mean fell from 17.3 centimetres, standard deviation 5.2, to 10.7 with standard deviation 1.7. With the sound sight at 18 metres it fell from 11.9 centimetres, standard deviation 3.6, to 5.3 with standard deviation 1.3. Without the device the mean distance moved a great deal from one end to the next; with it, the figure dropped and settled. The authors also report that a first-year beginner trained with the method scored 563 out of 720 at the prefectural autumn competition, over 72 arrows at 30 metres, third of twelve and first among the newcomers.

Tilt and twist were picked up with a three-axis accelerometer, bearing with a digital compass; in the six-joint device the compass was replaced with a module combining compass and accelerometer with tilt compensation, because in the first version the bearing was skewed by the inclination of the sensor. The sensors connect to a small Arduino board, the data go to a microSD card through a logger or over usb to a computer, and are drawn there with Processing. The stated resolution is one degree and the sampling interval one tenth of a second. The sensors of the full device were strapped to the six joints with elastic bands.

The reference, which the authors call the teacher angle, is set with a few trial shots: the angles of the bow, or of the six joints, are taken from a shot that landed in the ten. In the cumulative chart the following thirty-six shots are plotted as deviations from that reference, one degree of deviation drawn as two units on the vertical axis, on a scale running to twenty either way; were the position always the same, the points would stay stacked at the origin, whereas they stretch into a bar as the position falls apart.

The comparison on the two sights was run on beginners with less than a year of practice, over twelve series of three arrows with and without the device, at 30 metres with the light sight and at 18 metres with the sound one. The grouping index is the geometric mean of the three distances between the holes of a three-arrow end, that is the cube root of their product, in centimetres: the closer to zero, the closer the arrows lie to each other. The method as a whole was then applied to a first-year beginner in a school club, and the result reported is his competition score.

The test does not say how many archers took part. The paper speaks of beginners with less than a year of practice and reports twelve ends per condition, but not the number of people, their age, the draw weight or the bow, and there is no statistical test: only means and standard deviations. Nor is the order of the two conditions given, so the difference between with and without the device may carry the effect of warming up or of settling into the series.

The effect is measured while the device is in use. Nobody shot again without the sight after working with it, so the study shows an aid that works while it is there rather than learning that stays. The distinction matters, because the sound and light sights tell the archer when to let go, and an aid of that kind can stand in for the feeling instead of building it. The authors themselves note that only one sight may be used in competition.

The grouping index is crude. The geometric mean of the three distances between three arrows depends heavily on where the third one lands, and over three arrows it cannot separate a tight group away from the centre from a centred one. The two distances, 30 and 18 metres, were tested with different devices, so the two results do not compare with each other.

The competition result is a single case. One beginner, one competition, no comparison with archers who did not use the method: 563 out of 720 is a figure, not evidence for the three-stage programme, which as a whole was never tested. Part of the stated purpose also stays outside the results, namely engineering education: the paper appears in an electrical engineering journal and describes a teaching exercise, with equipment built by students. The text is in Japanese, with an English title and abstract.

What this work shows is that a beginner given a live view of how his bow is tilting groups better within the same afternoon. At 30 metres the arrows of an end go from a mean spacing of 17.3 to 10.7 centimetres, at 18 metres from 11.9 to 5.3, and the variation between ends falls. That is a large effect, measured on a simple index and with the device fitted to the bow.

What it does not show is that the beginner learned anything. Nobody had him shoot again without the device, so there is no telling whether the hand stays still once the light goes out. Nor does it show that the three-stage programme works, since the test covers only the first two devices.

For a coach there is one idea that transfers without any electronics: the reference here is not an ideal position from a manual, it is the position that this archer held on the shot that went to the centre. The device compares every shot with that one and shows the deviation stacked over thirty-six arrows. It is also a reminder of why a beginner without an outside eye struggles so much: the wrist twist that tilts the bow and shifts the string picture is hard to feel and easy to see, once something makes it visible.