Recurve arrow in high-speed video: Park 2012 study

Using a camera running at 4,000 frames per second, the authors filmed the first 15 milliseconds of the shot of 16 recurve archers from national teams. On average the string stays on the fingers for 1.9 ms. Carried by the string, which moves sideways as it leaves the fingers, the nock reaches its furthest point towards the archer's body at 5.4 ms, 13.1 mm off the bow's centre line (from 9 to 20 mm depending on the archer). The arrow, bent by this movement, is straight again at 8.6 ms, bends the other way with a peak at 12.4 ms and leaves the string at 14.6 ms. Contact with the pressure button had always ended by 8 ms. In all 16 archers the bow moved sideways by no more than 1 mm and rotated by less than 0.1 degrees while the arrow was on the string: the bow's movements, including the flexing of the sight and stabilisers, come afterwards.

Technique varied a good deal from one archer to another. In 11 the string left the middle finger last, in 4 the index and middle fingers together, and in one the index finger; in three the fingers were not perpendicular to the string. Seven of the 16 appeared to twist the bow about the grip, and four touched their face with the string, lightly (three on the chin, one on the chin and lips). Only in 4 did the nock leave the string on or very close to the bow's centre line: the separation point varied widely between archers (standard deviation 4.9 mm), and when it was well off the centre line the string oscillated strongly sideways afterwards. The men, with heavier draw weights, had slightly shorter times than the women: 14.0 ms against 15.2 ms for the nock leaving the string.

The stated aim was to provide data for checking mathematical models of arrow flex. The authors conclude that comparing a model with a single archer, however good, risks blaming the model for differences that come from technique, and that a shooting machine is acceptable for this kind of test, provided it reproduces the release adequately, precisely because the bow does not move before the arrow has gone. They also note that habits usually regarded as faults, such as twisting the bow or touching the face with the string, can go with very high scores as long as the archer repeats them closely on every arrow.

The footage was taken at the test event for the London 2012 Olympic Games, at Lord's Cricket Ground in October 2011, from volunteer archers shooting at 70 metres. The camera (an NAC Memrecam GX-1 at 4,000 frames per second, one image every 0.25 ms, with 800 by 800 pixels) sat above and in front of the arrow's path and looked down on it to about 2 metres in front of the bow. The videos were analysed frame by frame, counting pixel displacement and using the known diameter of the shaft as a scale; the estimated errors are about 1 mm laterally and one or two frames on the times.

The archers were 8 men and 8 women, all right-handed, using a mix of bow models. Draw weights and draw lengths are as stated by the archers: from 204 to 227 newtons for the men (mean 214) and from 146 to 203 for the women (mean 171), where 212 newtons is 48 pounds; arrows were 699 to 775 mm long for the men and 660 to 711 mm for the women. Fifteen used Easton X10 arrows and one Easton ACE, with spine in line with the manufacturer's charts. Pressure button settings and point weights were not available. The order in which the fingers left the string, bow torque and face contact were judged by eye from the video, not measured with instruments. The same footage is also used in a second paper by Park, published the same year, to check a model of the arrow's lateral flex.

The number of arrows filmed per archer is not given, and the data come from a single event under that day's conditions. Bow torque is inferred from the shape of the arrow, the clearance of the nock past the pressure button and the position of the bow hand, not measured on the bow; finger order and face contact are visual judgements too. Draw weights and draw lengths are self-reported, and the pressure button settings, which matter for exactly this kind of arrow behaviour, are missing.

Nothing links what is seen in the videos to where each arrow landed or what it scored. The idea that imperfect but repeatable technique still produces excellent scores therefore remains the authors' observation, not a finding of the study. Some mean times differ by a tenth of a millisecond between the text and the table (8.7 and 8.6 ms for the straight arrow, 14.7 and 14.6 ms for separation): this record uses the table.

The study shows how short the shot is from the arrow's point of view: on average the string stays on the fingers for 1.9 milliseconds (2.25 at most) and the arrow stays on the string for 14.6. In that time, even for the best archers in the world, the bow does not measurably move sideways or rotate, so the bow dropping and the accessories flexing after the shot do not change how the arrow leaves. What does matter, according to the authors, is the bow hand and the way the string leaves the fingers: twisting the bow, or fingers coming off the string differently from one arrow to the next, changes the sideways push the string gives the arrow.

For tuning, the most concrete finding concerns where the nock leaves the string: only four archers out of sixteen had it on the bow's centre line, and for the others the string vibrated sideways after the shot. The authors think that once this is set correctly an archer can feel the difference, as a quieter bow with a different feel in the hand; it is an indication, not a measurement. That seven world-class archers seemed to twist the bow does not make twisting advisable: it shows that different techniques coexist at that level, and that one champion's shot, seen up close, comes with its own quirks.

Anyone filming their own shooting needs a very fast camera to see any of this: the authors found 4,000 frames per second sufficient. At 240 frames per second there are more than 4 milliseconds between one image and the next, so the whole of the string leaving the fingers fits inside a single gap between frames.