Bow grip in kyudo: Koike 2007 study

The authors built an instrumented Japanese bow to measure what the left hand actually does during a kyudo shot, and tested it on 19 members of a university club. The action in question is called tenouchi and serves two purposes: twisting the bow anticlockwise, seen from above, because the string passes through the middle of the wood while the arrow sits on the right side, and without the twist the arrow would scrape the bow and fly to the right of the target; and pushing the upper part of the bow forward (uwaoshi), because the grip sits a third of the way up and the shorter lower limb recovers faster, which would send the arrow high.

The field measurements give three numbers. The upward push moment lies between 1 and 5 Nm, with repeatability that varies a great deal from archer to archer. The torsion about the bow's long axis is about 0.57 to 0.61 Nm in most of the subjects, and it is the quantity that repeats least well. A tenth of a second before the arrow leaves the right hand, every subject applies a torsion proportional to the force exerted on the bow.

The difference the authors read between those who hit the target more often and those who hit it less is not in the size of the forces but in how repeatable they are from shot to shot. In subjects with a high hit rate the force on the bow and the push moment overlap almost exactly from one trial to the next; in those with a low hit rate the traces scatter. In the better subjects the torsion grows over the last five hundredths of a second before the arrow leaves the right hand; in the middling ones it falls; in the weakest it wanders. Subject 10, with 67 per cent hits in practice and 91.6 per cent in competition, has the most closely overlapping traces; subject 13, at 51 and 50 per cent, has the most undulating torsion.

The instrumented bow carries four pairs of strain gauges for the bending moments and two pairs of torsion gauges, bonded above and below the grip, on the inner and outer faces of the limb, with the leads routed along the inner face. From the two pairs above the grip and the two below, through the moment equilibrium equations, the force applied to the handle, the upward push moment and the torsion are derived.

For calibration the authors built a rig: the grip clamped in a vice, a wire hooked to the string and pulled by an adjustable load up to the maximum real draw length, with a load cell measuring the tension. Because in a Japanese bow the part below the grip is shorter and the attachment point slides upwards as the load grows, the wire was adjusted so that the direction of the tension always passed through the grip, so as not to generate moments about it. The relation between sensor voltage and applied moment was approximated with a fourth-order polynomial; for the torsion sensors, a couple applied to a plate fixed symmetrically on the limb gave a linear relation. The final check compares the force estimated from the sensors with the one read by the load cell: the line relating them is y equals 0.99x plus 0.82.

In the experiment, 19 members of the university kyudo club took eight shots each in competition form at 28 metres, with an instrumented bow of a weight close to their own. Bows were built at 13, 14 and 16 kg in the standard length (221 cm) and at 16 kg in the long length (227 cm), all of fibreglass, with a synthetic string and with each participant's own arrows. At each shot the shooter's own comment on how it had gone, the point of impact and the arrow's speed were recorded. The six strain signals were converted to voltage by a dynamic amplifier and digitised. The instant of release was identified from the abrupt change in the signal of the sensor fitted below the grip, where the bow recovers faster; time zero in the charts is the moment the arrow separates from the right hand. Six subjects appear in the charts, with three trials the shooter judged good and three judged poor. For those six the hit rate of the previous week is given, in practice and in competition, and all six used the long 16 kg bow.

This is a four-page contribution to a sports engineering symposium, without peer review in the usual sense. The bulk of the work lies in building and calibrating the instrument; the experimental part is illustrative.

Nineteen people shot, but the results shown are those of six. Who those six are and why they were chosen is not explained. The comparisons between those who hit more often and those who hit less are read by eye off the charts, subject by subject, with no statistical analysis at all: there are no correlations, no comparison between groups, no measure of how closely the traces resemble one another. Sentences of the kind subject 14 holds the bow well and therefore keeps a steady hit rate are the authors' readings of three traces.

The hit rate used as an outcome does not come from the shots measured but from the previous week, in practice and in competition, and for the same archer the two values can diverge considerably: subject 9 has 67.6 per cent in practice and 44.4 in competition, subject 10 has 67 and 91.6. How these percentages were collected is not stated. The eight shots of each participant have a recorded point of impact, but the point of impact does not enter the comparisons.

The equipment used in the test is not the participants' own: purpose-built fibreglass bows of a weight close to but not equal to theirs, and in the charts shown all six used the same long 16 kg model, whereas with their own equipment the tenouchi might be different. The hand was measured through the bow, not on the archer: the position of the fingers, the pressure at different points of the palm and the orientation of the wrist all stay outside the measurement.

The conclusion that an optimal torsion proportional to the draw force needs to be applied is not demonstrated by the data: the proportionality observed describes what the participants were doing, while the optimality is the authors' addition. Nobody varied the torsion to see where the arrow went.

The study concerns kyudo and its asymmetrical bow: it does not transfer to the Olympic recurve, where the arrow rests on a rest and a pressure button, and where bow torque is treated as a fault to be removed rather than an action to be performed.

Within kyudo, the contribution is twofold. The first is the instrument: a bow that records force, upward push and torsion while the archer shoots, and the authors' reason for building it is one of method. Between the release and the arrow's separation two or three hundredths of a second pass, too little for the archer to feel what happens; feedback on the period before the release, they argue, is easier to read and easier to act on than feedback on the instant of release.

The second is the figure that stays with you: among the participants, those who hit more often were not applying larger forces, they were applying the same forces every time. It is repeatability, not magnitude, that accompanies the hit rate. Said with caution, because these are six traces read by eye and because the hit rate comes from the previous week rather than from the shots measured.

For anyone teaching kyudo, what remains is the order of magnitude of what the hand does: 1 to 5 Nm of upward push and about six tenths of a newton metre of torsion, applied in proportion to the force of the bow. These are measurements on nineteen students with test bows, and they give a scale rather than a target.