Bow grip pressure: Martin 1992 study

Archery instructors have always said that the way the hand presses on the bow grip decides where the arrow goes. This study tried to measure it, and did not find it. Fifteen Olympic-style recurve archers shot 72 arrows at 30 metres with fifteen thin force sensors stuck to the grip, and from those readings the moments the hand applies to the bow in the 150 milliseconds before release were estimated: one about the horizontal axis, which would tip the bow up or down, and one about the vertical axis, which would turn the arrow left or right.

The archers fell into two clearly distinct groups. Eight with a twelve-month FITA average and a most recent score above 1,250 points out of 1,440, among them four of the six members of the 1988 United States Olympic team and four more from the national team; seven with an average and most recent score below 1,150. During the test the difference showed: the mean distance of the holes from the centre was 4.3 centimetres in the first group and 8.6 in the second.

On the distribution of force, the general picture is the one coaching describes: almost all the load sat on the posterior aspect of the grip, the part against the palm, while the sensors on the thumb and fingers read much lower forces, as expected from a hand that pushes rather than grasps. Within that area the force sat higher rather than lower and more towards the fingers than towards the thumb.

The result is that none of this was tied to where the arrow landed. The correlations between estimated moments and the arrow's vertical or horizontal deviation were small and often not even in the expected direction: of fifteen archers, only one showed a significant link with vertical deviation and three with horizontal deviation. Between the two levels there were no statistically significant differences in force distribution, although the better archers had moments about the horizontal axis roughly twice those of the less skilled, with very high standard deviations and large differences within each group. Fatigue produced nothing systematic, even though the 72 arrows were shot in a session of about two hours at a faster pace than usual precisely in order to tire the archers; the only effect over time is a moment about the vertical axis that grew in the later part of the series.

One thing alone separated the two groups: consistency. The stronger archers were steadier from arrow to arrow in the vertical distribution of force, at a probability of 0.045, and that agrees with their mean vertical deviation, 2.7 centimetres against 5.7. On the horizontal distribution, the one coaching usually gives all its attention to, there was no difference at all.

The sensors are force sensing resistors, strips of 15 by 28 millimetres and less than a millimetre thick with a circular sensing area of 12 millimetres, attached with double-sided tape and covered with surgical tape that shaped them to the grip and protected the leads. Seven were placed on the posterior aspect, the part against the palm, two on the thumb side and six on the finger side, with a fixed spacing between centres of about 20 millimetres vertically and 15 horizontally; since the hands were of different sizes, the areas of palm and fingers that ended up over each sensor varied from archer to archer. The archers reported feeling the grip was bulkier, but not that their hold or technique was disrupted.

Each sensor was calibrated before data collection by applying and removing nine loads from 0 to 37.4 newtons, with a quadratic curve fitted to the points: of the 225 equations obtained, 79% had a coefficient above 0.96 and only 2% below 0.90. After six practice arrows, each archer shot 72 arrows at 30 metres in a single session of about two hours, at a faster pace than in competition, chosen precisely to bring out fatigue, with the instruction to aim for the centre on every arrow.

The sensors were sampled at 40 Hz for six seconds around each release, but only for three blocks of six arrows each: arrows 1 to 6, 34 to 39 and 67 to 72, so eighteen recorded arrows per archer out of the seventy-two shot. To synchronise the recording with the shot, two electrical signals marked the key instants: the closing of the clicker, obtained with a copper strip fixed to the bow, which happens about 180 milliseconds before the release, and the release itself, detected by an inertia switch mounted on the horizontal stabiliser, two strips brought into contact by the bow's abrupt acceleration. The performance of each recorded arrow is the position of the hole on the target in cartesian coordinates, giving vertical, horizontal and resultant deviation from the centre.

The moments are estimated, not measured. For the horizontal axis the authors used the seven posterior sensors, weighting each force by a factor approximating its vertical distance from the axis, positive above and negative below. For the vertical axis they used every sensor except the one centred on the axis, with a positive sign on the finger side and a negative one on the thumb side, but with no distance weighting, because the curvature of the grip makes an actual lever arm hard to estimate for each sensor. From these they also derive two estimated moment arms, unitless numbers running from minus 3 to plus 3 for the horizontal axis and from minus 1 to plus 1 for the vertical one, which say where the force tends to be concentrated.

The analysis crosses arrow position with the estimated moments archer by archer, with one-tailed tests because the direction of the effect was expected, and compares the two levels and the three blocks of arrows with a two-factor analysis of variance, applied both to the mean values and to their variability from shot to shot.

What the study measures is not the moment acting on the bow during the shot but an estimate built from fifteen pressure readings in the 150 milliseconds before release. The authors assume that estimate adequately represents the moments over the very brief stretch in which the arrow is still on the string, the only moment in which they would matter, and they state it as an assumption. The weights used to compute the moments are approximations of distances rather than measured distances, and for the vertical axis no distance weighting was applied at all; the results are therefore unitless numbers, comparable between archers only crudely.

The sensors are few: fifteen in all and seven on the part that counts. The authors acknowledge that this resolution may not be enough to catch subtle changes in pressure distribution. In some archers two of the posterior sensors exceeded the calibration range, and the means for those two are declared to be read with caution. The sensors also add thickness to the grip, which the archers noticed.

The most delicate point concerns the measure of performance. Deviation is computed from the centre of the target, assuming every archer was aiming at the centre; the authors write that it would have been more appropriate to measure deviation from the point the arrow was actually pointing at an instant before release, but that point could not be determined. Beyond that, nothing else in the technique was controlled: the string release, which the authors themselves cite as the factor archers consider decisive for arrow flight, was not measured, and its variability ends up inside the deviation attributed to the bow hand.

The sample is fifteen archers, eight and seven per group, and the standard deviations of the estimated moments are large: the twofold difference between the two levels on the horizontal axis fails to reach significance for exactly that reason. Some effects are declared as close to the threshold, with probabilities of 0.053 and 0.10. Of the 72 arrows, 18 were recorded, in three blocks, so fatigue is assessed from three snapshots rather than a continuous curve. The distance is 30 metres, while the belief the study tests refers explicitly to effects magnified over 30 to 90 metres.

The authors also flag what they chose not to do: they did not ask the archers to torque the bow deliberately while holding the aim steady, a test that would have said how much the force distribution must change to move the arrow. Without it, two readings of the result remain possible, and they say so: either pressure on the grip matters less than is believed, or the normal variations of an experienced archer are too small for this system to see. The work is from 1992, with the bows and grips of the time, and the same force distribution on a modern grip is not necessarily the same thing.

The study tests a specific belief in coaching, that torquing the bow with the hand sends the arrow off, and fails to confirm it. Across fifteen archers, with eighteen instrumented arrows each, the moments estimated from the hand on the grip were not consistently associated with where the arrow landed: one archer in fifteen for vertical deviation and three in fifteen for horizontal. That does not mean the bow hand does not matter: it means that, in the normal shooting of experienced archers, the pressure variations this system can see do not explain where the arrow ends up.

Two figures describe how a trained archer's hand pushes. Almost all the force sits on the posterior aspect of the grip, the part against the palm, and almost none on the thumb and fingers: the hand pushes, it does not squeeze. Within that area the force sits higher rather than lower, and the better archers concentrated it closer to the top of the grip, with moments about the horizontal axis roughly twice those of the others, although the difference was not significant and the differences between people inside each group were large.

What separated the levels is not where the hand presses but how consistently it presses the same way: the better archers were steadier from arrow to arrow in the vertical distribution of force, and they were so alongside a mean vertical deviation half as large, 2.7 centimetres against 5.7. On the horizontal distribution, the one people mean when they talk about torquing the bow, there was no difference in consistency between the groups.

The third finding concerns fatigue. The idea that a tired archer changes hand or wrist position in order to reach full draw was tested in the most favourable way possible, with 72 arrows in two hours at a brisk pace, and nothing systematic emerged in the distribution of force. The only thing that changed over the series was a moment about the vertical axis, which grew in the later part, and grew more in the better archers than in the others.