Tremor, posture and recurve archery score: Rose study

Fifteen male recurve archers, from sub-elite to elite, aged 26.6 ± 9.3, completed two sessions: an indoor shooting session, 150 arrows on an 18-metre FITA track, with the mean score per arrow as the performance measure; and a laboratory session, in which the tremor transmitted to the bow, postural sway, shoulder girdle steadiness and maximal pull strength were measured.

Three measures correlate negatively with the score. Shoulder girdle steadiness is the strongest (r squared 0.45, p < 0.01), followed by the power of vertical tremor in the 3 to 8 Hz band (r squared 0.30, p < 0.05) and the amplitude of vertical tremor (r squared 0.26, p < 0.05). Linearly combining the two vertical tremor measures gives an r squared of 0.54 (p < 0.01).

Two measures correlate with nothing: horizontal tremor transmitted to the bow, and the path length of the centre of pressure. On horizontal tremor the authors note that the power concentrates in the band below 1 Hz and suggest that archers manage to compensate for a tremor that slow during the release phase. On postural sway their conclusion is clear: it is not a predictor of the score.

There is then a finding about relative draw weight. Mean maximal pull strength was 532 ± 117 newtons and bow force 196 ± 16, that is about 37% of the maximum (the percentage is my own calculation on the two means). When bow force exceeded 40% of the maximum, performance tended to decline, but with r squared 0.43 and p = 0.054, hence without reaching the threshold.

The score comes from 150 arrows shot on a FITA-approved indoor track at 18 metres. The authors explain the choice of indoor shooting: it removes the weather from the confounders. Performance is the mean score per arrow.

In the laboratory the archer stood on a force platform and aimed at a target placed 2.5 metres away. After warm-up and familiarisation they shot six arrows with the aiming phase extended to 10 seconds, and the data come from the aiming phase of the last four. Postural sway is the path length of the centre of pressure recorded by the platform. The tremor transmitted to the bow was measured with an accelerometer attached to the bow: the amplitude is the standard deviation of vertical and horizontal acceleration, and the power was computed in three bands, 0 to 1 Hz, 3 to 8 and 8 to 12, normalised to the total power between 0 and 30 Hz.

To isolate shoulder girdle tremor, a separate task was used in the same aiming position: the archer had to hold a force equal to their own draw weight steady, with real-time visual feedback of the force. Steadiness is the ratio of standard deviation to mean force, that is a coefficient of variation: the lower it is, the steadier the force. Maximal pull strength was measured in the same setup.

Force data were sampled at 1 kHz and accelerometer data at 0.25 kHz. The analysis consists of correlations between each measure and the score, plus a linear combination of the two vertical tremor measures.

This is a two-page conference abstract, peer reviewed. A bibliographic note: the University of Copenhagen research portal lists three authors, while this PDF carries four, with Bente R. Jensen and their affiliation at the head of the article. The primary document confirms the four names.

The predictors and the performance come from two different tasks. The score is from 150 arrows shot at 18 metres; the laboratory measures are taken while aiming at 2.5 metres, with the aiming phase extended to 10 seconds, and shoulder girdle steadiness is measured not by shooting but by holding a force steady while watching visual feedback. These are conditions chosen to isolate the tremor, and for that they work, but the link to shooting runs through an inference: none of these measures was taken during the arrows that produced the score.

The sample is fifteen people and the analyses are correlations. With fifteen points, the threshold for p < 0.01 sits around an r squared of 0.41, so steadiness's 0.45 clears it only narrowly: it is a hint with a wide margin of uncertainty, not a stable estimate. The authors put it in measured terms, calling steadiness «a fair single predictor».

The combination reaching r squared 0.54 should be read carefully for two reasons. It is a two-predictor model estimated on fifteen observations, with no r squared adjusted for the number of predictors and no validation, and in those conditions the raw value overstates what would be obtained on new data. And the two predictors combined, the amplitude of vertical tremor and its power between 3 and 8 Hz, are derived from the same signal from the same accelerometer, so they carry no independent information.

The comparisons are many and uncorrected. Between the two directions of tremor, the three frequency bands for each, the amplitude, the steadiness, the maximal strength and the postural sway, the possible correlations against the score number more than ten, and the ones reported are those that came out. On fifteen archers, with that many comparisons, a few results at 5% are what chance produces.

The finding about 40% of maximum, which is the most quotable, does not reach the threshold: p = 0.054, and the authors describe it as a trend. In the conclusions, though, it becomes guidance on how to choose draw weight. The same holds for the advice to raise maximal strength in the aiming position: it rests on an external study about the relationship between relative load and steadiness, not on an intervention tested here. Nothing was trained in this work and no change over time was measured.

The format and the sample remain. Two pages of proceedings, with no descriptive tables for the predictors, no confidence intervals and no effect sizes beyond r squared. Men only. And the level runs from sub-elite to elite, with a score distributed in the figure between 7.5 and 10 points per arrow: a range that wide helps a correlation emerge, and part of what is seen may be the difference between levels rather than a relationship within each archer.

The most useful result is the one that falls away, not the one that holds. Body sway on the platform does not predict the score, in any form: the authors are explicit. What does predict, as far as a study on fifteen people can predict, is the steadiness with which the shoulder girdle holds a constant force, and the vertical tremor transmitted to the bow. Of the two tremor directions only the vertical counts: the horizontal one, which concentrates below one hertz, that is in slow oscillations, does not link to the score, and the authors think archers manage to compensate for it.

A frequency band is indicated: between 3 and 8 Hz in the vertical plane. That is a tremor too fast to be corrected consciously, because a voluntary correction takes longer than an eighth of a second, and the authors trace it back to oscillations of central origin. Translated into practice: the tremor that counts is not the one seen wandering in the sight, it is the faster one that is not seen.

The most tractable number concerns draw weight. In these archers the bow force averaged 196 newtons against a maximal pull of 532, that is about 37%; converted, around 44 pounds against a maximum of about 120 (the conversions and the percentage are mine, not the paper's). Beyond 40% of maximum, performance tended to worsen, at p = 0.054: it is not a significant result, and it is exactly the kind of threshold best not treated as a rule. It does remain a sensible way of framing the question about your own draw weight, that is as a percentage of what you can pull at most rather than in absolute pounds.

One last point for anyone wanting to measure these things at a club. This study's steadiness test does not require shooting: you hold a force equal to your own draw weight in the aiming position, watching visual feedback, and look at how much it varies. It is the strongest predictor the study finds, and it is built from a dynamometer and a screen.