Muscle timing in archery: Machado 2018 study

This master's thesis from the University of São Paulo recorded the electrical activity of twelve muscles in 10 archers from the Brazilian national recurve team (4 men and 6 women, aged 15 to 32, with between a year and a half and 21 years of shooting) while they shot 72 arrows at 70 metres. It does not look at score: it looks at how muscle activity changes in the half second before the clicker falls and the half second after.

Before the clicker the picture is fairly clear. From 500 to 50 milliseconds before the fall, activity stays almost constant; in the interval from 50 ms before to 50 ms after, eleven of the twelve muscles are more active than in any earlier interval. Triceps, for example, goes from about 10% of maximal contraction to 73.5%, and the finger flexors from 3.8% to 44.1%. Biceps and pectoralis major rise already between 50 and 150 ms before. The only muscle that does not change is lumbar multifidus.

After the clicker activity falls, though not everywhere: finger extensor, biceps, triceps, pectoralis, serratus anterior on the string side and upper trapezius decrease between 50 and 350 ms, while the finger flexors, rectus abdominis, multifidus, latissimus dorsi, serratus on the bow side and posterior deltoid stay at the same level. Across the twelve ends of six arrows, some muscles change their activity over the session in some intervals (multifidus and bow-side serratus in all of them) and others do not. Finally, the cross-correlations between signals are almost all high, between 0.72 and 0.99.

Data were collected outdoors, at the field where the team normally trained, with each archer's own equipment. After maximal voluntary contractions of each muscle and a warm-up of three ends of six arrows, every archer shot two rounds of six ends of six arrows at 70 metres, with a 20-minute break halfway; someone else collected the arrows. To be in the national team, and in the study, an archer needed a FITA score above 1,000 points; anyone with an arm or trunk injury in the previous 12 months was excluded.

Surface electromyography (EMG System Brasil, 16 channels, 2,000 samples per second) recorded the superficial finger flexor and the finger extensor, the long head of biceps, triceps, the clavicular head of pectoralis major, rectus abdominis, serratus anterior on both sides, lumbar multifidus, latissimus dorsi, upper trapezius and posterior deltoid. An accelerometer on the wrist was used to find the reference moment. The signal, filtered at 20 Hz and rectified, was expressed as a percentage of maximal contraction and divided into windows around the clicker fall: five of 100 ms before (labelled anticipatory adjustment), three of 100 ms after (compensatory adjustment) and three modulation windows between 40 and 180 ms after.

The windows were compared with one-way analysis of variance and Tukey's test, and the ends with a repeated-measures analysis. A correlation above 0.9 counted as strong.

The statistics pool the arrows of all the archers as if they were independent data: the degrees of freedom (4 and 3,199, then 11 and 639) correspond to about 640 shots per window, not to 10 people. With so many observations even small differences come out significant, and dozens of comparisons made muscle by muscle and window by window, with no correction, make it likely that some of the effects across ends are due to chance. Standard deviations are often larger than the means (73.5% with a standard deviation of 169.25 for triceps), a sign of a very skewed distribution and perhaps of artefacts from the bow's shock at the moment of the shot: the text mentions crosstalk and movement artefacts in general, but does not connect them to these values.

It is unclear how the instant of the clicker fall was identified: the text says the accelerometer picked up movement after the string was released, so the reference may be the shot rather than the clicker. The labels anticipatory and compensatory adjustment are interpretations: the window called anticipatory also includes the 50 ms after the clicker, and the modulation windows overlap the compensatory ones. Nor does the text always say which arm each muscle was on.

The section on correlations does not match its own table: the text says the finger flexors and extensors have an inverse, not strong, relationship, while the table gives 0.95. Score and arrow position are missing altogether, so no result can be linked to the quality of the shot. Lastly, this is a thesis, not a paper that went through external peer review, and its English is at times hard to interpret.

For anyone who shoots, the thesis describes something familiar: in national-level archers the hold at full draw is steady for most of the last half second, and almost every muscle recorded, from the hand to the trunk, becomes much more active only in the 50 milliseconds around the clicker fall. After the clicker some muscles relax, while the trunk and the bow side stay active at the same level, as if holding the frame still while the arrow leaves.

This does not say what the right way to expand or release is, because no measure is linked to score and there is no comparison with less experienced archers. The author herself concludes that archers use different muscular strategies to reach the same outcome, which makes it of little use to look in these data for a model to copy.

The part on fatigue is weak too: some muscles change their activity over the twelve ends, but with arrows treated as independent data and many comparisons it is not possible to say whether this is fatigue or chance variation.