The final pull: what happens in the last millimetres before the clicker

With the sight already on the centre, the arrow keeps sliding back under the clicker for a few more millimetres. In a group of international junior archers, measured with a sensor glued to the clicker blade, the best of them drew the arrow back about 5 millimetres over the last three seconds, while another needed only about 2.5 in the final second; in the study carried out thirteen years earlier with a different instrument, the draw hand moved back about 5 millimetres over the final five seconds, slowing to roughly a millimetre per second. This is the final pull, and in the registry the claim that it lasts two or three seconds and adds only a few millimetres of draw is reported: two studies have followed it millimetre by millimetre on the Olympic recurve, and both are small.

Coaches talk about it a great deal, usually through a single rule: once you are aiming you do not stop, you keep expanding until the clicker drops. The measurements point the same way as the rule, but they say more precisely what "not stopping" means, and how much the pull can differ from one archer to the next without the score suffering. The picture rests on small samples, and I will flag it whenever the number of archers changes how much a result can carry.

What the final pull is

When a recurve archer reaches anchor, the clicker blade rests on the arrow point, a few millimetres from its tip. The most delicate phase of the shot starts there: the archer aims while still pushing the bow and pulling the string, until the point comes out from under the blade and the clicker snaps against the riser. The research literature calls it the final pull, or the push-pull, to stress that both arms are working together; coaches more often call it expansion.

Mechanically it is close to isometric work, with muscles under tension that change length only very slightly. The difficulty lies elsewhere. In those same seconds the archer has to hold the sight on the centre, control the arrow as it slides back and keep the bow still. The German researchers who devoted two papers to the final pull set this out explicitly, and none of the three jobs can wait, because the clicker drops when the point clears it, not when the archer is ready.

How to measure a few millimetres

A movement of a millimetre per second cannot be seen with the naked eye, and barely shows on ordinary video. The first study to measure it, published in the early 1990s by a biomechanics laboratory in Paris, fitted a displacement sensor to the handle, with a small iron rod attached to the lower limb: the further the archer draws, the more the limb bends and the further the rod travels inside the sensor's coils, and with a scale factor worked out for each bow, limb movement becomes movement of the draw hand. The signal was sampled seventy times a second and analysed over the second and a half before the clicker dropped, a moment marked by an electrical contact that the clicker closed as it fell.

The device was awkward to fit, and according to the researchers who returned to the question thirteen years later, that may be why it had remained the only study of draw length in the final pull. Their own solution was simpler: a strain gauge glued to the clicker blade. The point tapers, so every time the arrow moves back or creeps forward the blade bends differently and the gauge records the change; knowing the shape of the point, the bend gives the change in draw length. The trace is ready as soon as the arrow has gone, and the authors report that the system was used frequently in the training of the German junior national team. The other ways of instrumenting the clicker, with the limits of each, are set out in the "How it is measured" section at the foot of the page.

How long it lasts and how far it goes

The two instruments describe the same shape. The hand comes into position in a fast stretch, at around half a metre per second, and then slows abruptly. In the first study it moved back about 5 millimetres over the last five seconds, at close to a millimetre per second, and over the second and a half before the clicker a straight line fitted its path well on every shot. Among the German juniors, on the three shots reported, the archer with the highest average was always about 5 millimetres from the click three seconds before the shot, while the second-best archer needed only about 2.5 millimetres in the final second to make the blade drop.

Pinning down when the final pull begins is harder. The German authors say so plainly: on three shots by their best archer it started between 2.1 and 1.9 seconds before the shot on the first, between 3.0 and 2.8 on the second and between 2.6 and 2.4 on the third, and on some arrows the small fluctuations in draw length during aiming blurred into the start of the pull, with an uncertainty of up to 0.6 seconds. That is why they did not calculate an average duration. The scale holds all the same: two or three seconds in the best archer of the German group, between 2.0 and 2.5 seconds in ten experienced traditional Turkish archers, who do not use a clicker, while in the first study the slow phase took up almost the whole of the aim, with the hand already slowing five seconds before the clicker. The archer governs these few millimetres almost entirely through the information coming from their own muscles and joints, as the authors of the first study had already noted.

Better arrows come with a steadier pull

To find out whether the shape of the pull matters, the first study used a simple measure. For each shot it fitted the straight line that best described the hand's movement and calculated how far, on average, the recorded points strayed from it: the smaller the number, the more evenly the hand moves back, without jerks, stops or reversals. There were eight archers, all men, five of them elite and aged 20 to 25 and three at university level, and each shot 36 arrows at 90 m in ends of three, on competition timing.

The first result concerns skill: the elite archers had a steadier final pull than the university archers. The second is more useful to a coach, because it looks inside the same archer. The authors split each archer's arrows into two groups, those scoring more than one standard deviation above his own average and those more than one standard deviation below; for the first archer, whose average was 7.25, that meant comparing the 9s and 10s with the 5s or worse. In all eight archers the better arrows had a steadier pull than the worse ones, and the difference was statistically significant for every one of them.

Scroll the figure→

The limits are those of an experiment of its time. There are eight archers, the arrows are compared only at the extremes, leaving out the many in between, and the measure is indirect, taken from the limb rather than the hand. The study also shows an association: it cannot tell whether an uneven pull moves the arrow or whether the unevenness simply comes with a shot that was already going astray for other reasons. The authors suggest two possible routes, since an uneven pull can change where the hand is at the release and can also disturb the aim.

Pauses and creeping on the worst arrows

The study with the instrumented clicker measured seven juniors, two young men and five young women aged 16 to 19, who had competed at the Junior World Championships and the German national championships. Each shot 66 arrows at 30 m indoors, in ends of six with no time limit, using their own bow. The gap in level between them was small: average scores per arrow ran from 9.31 to 8.72.

The table carries an inconsistency that comes from the originals: in both papers the scores range from 6 to 10 for S6 and from 5 to 10 for S7, yet the captions of the traces give S6 a 5 among the three worst arrows and S7 three 6s. I have left it as published, because the data in the papers do not show which of the two figures is the right one.

Among archers this close, the authors note, the clear-cut differences seen in the first study did not reappear, and their analysis of the final pull is purely descriptive: no statistical test, just a reading of the traces from a few selected arrows. Even the two weakest archers, on the 9s and 10s the authors show, had irregularities that they describe as nothing out of the ordinary. Within those limits the picture holds together. The archer with the highest average drew the arrow through the clicker evenly, and the traces of that archer's arrows resembled one another even when they were not identical; the two with the lowest averages were the least steady, and two specific behaviours turned up on precisely their worst arrows.

The first is creeping. On archer S6's three worst arrows the pull broke off when the point had almost come out from under the blade: the archer stopped pulling, tension dropped and draw length shrank, so the arrow slid slightly forward, until after a moment the pull started again. The authors describe it as a phenomenon well known in archery, which they call shrinking, and they saw it this clearly only in that archer. The second is the pause: on two of archer S7's three worst arrows the pull stopped just before the clicker and draw length stayed still for a moment, without shrinking, before the pull resumed.

The second-best archer in the group shows, though, that steady does not mean uniform. On all 66 arrows this archer split the final pull into two phases, a fast stretch followed by a slow and continuous one, which usually lasted under two seconds and on at least one arrow more than three. From the outside the slow phase might look like a pause; the authors read it as a personal, trained way of reaching the click, and this archer's average was the second highest in the group. In these data, then, a slow or two-phase pull was not enough to make a worse arrow: what turned up on the low-scoring arrows of the two weakest archers was draw length that stopped or shrank just before the clicker.

Scroll the figure→

With two small studies pointing the same way, the registry reports the claim that a continuous final pull through the clicker leads to better shots. These are associations, and in the study of the juniors the part on the pull is a description of selected traces, not a statistical comparison.

While the arrow moves, the aim tightens

The juniors' bows carried a second instrument: an optical system designed for rifle shooting, fitted to the riser in place of the lower stabiliser, which records a hundred times a second where the bow is pointing on the target. From that trace the authors calculated the hold box, a rectangle measuring how far the aim moves horizontally and vertically, expressed in square score points. They worked it out over three windows: from three to two seconds before the shot, from two to one, and in the last second.

You might expect the aim to wander most in that last second, when the archer has to aim, complete the pull and keep the bow still all at the same time. The opposite happened: in all seven archers the aim moved significantly less in the last second than in the two seconds before. Within each archer, the last-second aim was also steadier on 10s than on arrows scoring 8 or less. Between archers, however, the average hold box said nothing about the average score, and the steadiest archer in the group was not the best. This study is what the claim that while aiming, the bow moves less and less as the release gets closer rests on, and in the registry it is reported.

Scroll the figure→

The authors did not calculate the hold box over the final pull alone, because they could not place its start precisely enough; they do point out that on many arrows the pull began less than two seconds before the shot and the aim still tightened. The most direct link shows up in the archer who crept. The three arrows on which the pull broke off were also the ones with the most aiming movement in the last second, and leaving out the arrows with an uneven pull brought that archer's mean hold box down from about 0.59 to about 0.45. For the archer who paused, the numbers did not work out the same way. The authors conclude that a steady pull helps keep the aim on the centre; resting on a single archer, that is a pointer rather than proof.

One last detail applies to all seven: in the last second the aim moved more from side to side than up and down. Two other groups find the same, so the registry also reports the claim that while aiming, the bow sways more horizontally than vertically, with the caveat that the balance between the two directions varies from one archer to another.

What the muscles do around the click

At the click, muscle activity rises almost all at once. In ten recurve archers from the Brazilian national team, 72 arrows each at 70 m with electromyography on twelve arm and trunk muscles, eleven of the twelve were more active between fifty milliseconds before and fifty after the click than in any window of the preceding half second, and their traces moved almost in unison, with correlations between 0.72 and 0.99, the great majority above 0.90. The exception was the lumbar multifidus, which stayed between 2% and 4% of maximal contraction. It is a master's thesis, with measurements from a single laboratory, and it compares the click only with the moments before it: in some muscles, such as latissimus dorsi, upper trapezius and posterior deltoid, the average was higher still in the tenth of a second that followed. The claim that at the clicker drop almost all the arm and trunk muscles are more active than in the half second before, and they move together is reported in the registry, and it does not say that the click coincides with each muscle's peak.

How hard the drawing arm works is less clear. In four archers from the Turkish national squad at 18 m, with electromyography on the drawing arm, a force plate, an aim-tracking system and a timer linked to the clicker, in the second before the click the posterior deltoid and the finger flexors and extensors were working at between 82% and 85% of maximal contraction and the trapezius at 72%; in the second after, the deltoid and flexors fell to around 30% and the extensors to 58%, while the trapezius stayed where it was. A Japanese study from 1987, on two world-class archers, had instead seen the deltoid, recorded on a different head, fall silent only for an instant at the release and then go back to working hard together with the back. With one experiment for and one against, both small, the claim that while aiming, the drawing arm's muscles work at over 80% of their maximum, and at the release almost all of them switch off except the trapezius is contested in the registry, not least because percentages this high depend heavily on how maximal contraction is measured.

The same Turkish study compared 10s with 8s and found two differences that concern the pull. In the last second before the clicker the finger flexors, the muscles that hold the string, were working slightly less, at 80.0% of maximal contraction against 85.6%; and the total draw time, from the first activity in the trapezius to the clicker, was shorter, at 4.8 seconds against 5.6. There are only four archers, though, the arrows are all analysed together, and the same experiment appeared in two papers that describe the sample once as female and once as male: these are data to handle with care.

Neither result stands up on its own. On the draw hand, in eight high-level French archers at 70 m, the finger flexors were on the contrary more active on 10s than on 8s or lower, and for that reason the claim that on better arrows the finger flexors work less before the clicker goes off is contested in the registry. On timing, the Turkish figure compares arrows, pooled across the four archers, rather than archers with one another, and it covers the whole draw, not just the final pull. When archers are compared with one another, the difference some studies find between experts and beginners disappears among competent competitors: in sixteen young Malaysian recurve archers the best eight took 1.46 seconds to reach anchor and the others 1.36, with no significant difference. That is why the registry lists as contested the claim that higher-level archers complete the draw more quickly.

From the click to the release

A German study suggests another way of thinking about the click. Nine archers from the German national teams shot 35 normal arrows and five extended shots, in which they had to wait more than a second after the click before releasing, with electromyography on eight muscles. The activation patterns of the two kinds of shot were similar. The interval between click and release was also shorter than the archers' simple reaction time to a sound in seven of the nine, and shorter than their reaction time measured while holding a laser dot on the gold in all nine. The authors infer that the motor programme for the release starts before the click, and that the sound serves more as confirmation of a consistent draw length, to be used on later shots, than as a starting signal; their comparison is the sound of the ball on the racket, which a tennis player hears once the stroke has already been played. In all nine, activity in the pectoralis major on the bow side rose just before the release.

If that reading holds, the final pull and the release form one movement that the clicker does not trigger but confirms, which fits the draw-length measurements, where on the worst arrows of the two weakest juniors the movement broke off before the clicker dropped. But there is only one study, a short conference paper, and on the simpler reaction task two of the nine archers were not faster: the claim that in expert archers the release is not a reaction to the clicker going off is reported in the registry and no more than that.

On how long that interval lasts, the laboratories agree. Since 1987, six independent groups using different instruments at distances from 8 to 90 m have put it at around 180 milliseconds in the measurements up to the 1990s and between 140 and 165 in those after 2010, and the claim that between one and a half and two tenths of a second pass between the clicker sounding and the arrow leaving is corroborated in the registry. That time is not fixed, though, even in elite archers. In seven archers from the German senior and junior teams, thirty arrows at 18 m recorded with a microphone, the interval changed from one arrow to the next with a coefficient of variation between 3.94% and 7.25%, about 6 to 12 milliseconds either way, and none of the seven repeated the same time on every arrow. The study is small, and the registry reports the claim that even in elite archers the time between the clicker and the release changes from arrow to arrow; in the same group it was that variation, not the average time, that went with the average score, as the archers who varied least scored most.

As for speed, its weight on the score is small. In the four Turkish archers the time was the same on 10s, 9s and 8s, between 142 and 143 milliseconds, even though the paper's abstract claims it was shorter on the better shots; in 39 elite archers measured before a World Cup stage a shorter time went with slightly better arrows, for an effect worth about half a point from the quickest to the slowest in the group. The claim that the quicker the reaction to the clicker, the better the shot is reported in the registry with that small weight, and the authors of the World Cup study warn that even shorter times might mean the archer is anticipating the click rather than reacting to it. The clicker, its timings and its problems have an article of their own.

That leaves what happens to draw length when there is no clicker at all. In nine elite Korean recurve archers, all women, twenty arrows with the clicker and twenty without, in random order at 8 m, draw length measured just before the release did not change; without the clicker the draw was completed sooner and the aim was tighter. It is a single session, indoors and at short range, with archers who have used a clicker for years, but for now the claim that without the clicker you lose draw length is contradicted in the registry.

Holding costs: tremor and fatigue

The French study also contained a second experiment, which helps explain why the final pull cannot go on for long. Seven university-level archers had to hold the push-pull position until exhaustion, at an effort the authors put at 40-50% of their maximal strength, and lasted 52 seconds on average. In the posterior deltoid the electrical signal shifted towards lower frequencies, by an average of 0.72% per second, the classic sign of fatigue, and a tremor appeared at 8 to 12 hertz. On this one sample, the registry reports the claim that holding the shooting position demands 40-50% of maximal force, and at that level the muscle fails in under a minute; and a hold to exhaustion shows how much reserve there is, not what happens on every arrow.

The same tremor showed up in the movement of the limb, and the two signals were closely linked, with coherence values between 0.81 and 0.96 in that band: the muscle's tremor was reaching the bow. A year later, in twelve Italian archers, a protocol of repeated holds increased the sideways sway of the bow in all twelve, and the claim that muscle fatigue increases bow sway while aiming is reported in the registry. The limb's movement also had a second, slower component below 5 hertz, which fatigue did not change and which the French authors attribute to the control of the pull itself. Hence their four suggestions, none of which the study tested: a less powerful bow, which runs up against the distance to be shot; more maximal strength, so that the same bow takes a smaller share of the archer's capacity; a damper designed to counter tremor; and a draw in a single movement, as the elite archers of the time used, which shortens the time under tension.

Where to start from: a short final pull

How long the final pull should be has been asked directly by only one study, and not on the Olympic recurve. Ten experienced traditional Turkish archers shot from about a metre and a half, with accelerometers on both hands and on the bow, first stopping at set points, at 98, 96, 94 and 92% of full draw, and then completing the shot with the final pull; they also shot with no final pull, aligning directly at full draw. With the stop at 98 or 96%, which meant average final pulls of about 1.6 and 3.2 centimetres, jerk in the bow and in the bow hand was lowest; holding at full draw produced more in the bow hand, and the longer final pulls, up to about 6 centimetres, produced more in both hands. The claim that pausing at 96 to 98% of full draw before the final pull reduces the small jerks while aiming is reported in the registry.

The limits weigh heavily: traditional bows with no clicker, three shots per condition, the distances always tried in the same order, a regional journal and no record of where the arrows went. A traditional bow at a metre and a half is a long way from an Olympic recurve at 70 m, and the stop imposed by the protocol is a different thing from the involuntary pauses that turned up inside the final pull on the German juniors' worst arrows.

The only attempt to train it

The French study also tried to correct the pull. A computer followed the draw-length signal and, whenever the hand slipped forward past a threshold, played a sound until draw length returned to where it had been; the threshold was set for each archer so that only the larger slips were flagged. Six archers shot two series of 36 arrows on different days, one under normal conditions and one with the sound. With the signal the pull became steadier in five of them, while in the sixth it became markedly worse. Scores did not improve significantly, and the authors suggest that the extra information to process interfered with other important parts of the shot.

There is also a design problem: the threshold was set from the first series, so the series with the sound always came second, and part of the improvement may simply be practice. That is why the claim that a sound cue that flags collapses in the draw makes the draw steadier is contested in the registry. Whether training a steadier final pull raises scores has not been put to the test against a control group.

What I take from it for the range

Taken together, these studies show that on better arrows the movement under the clicker is steadier, and that on worse ones it can stop or shrink. For work on the range I take three things to watch from them, knowing that none has been put to the test in an experiment on scores.

The first is how much travel the clicker asks for at anchor. The German authors put the blade a few millimetres from the end of the point, and in the two archers whose figures they report about 5 millimetres were enough over the last three seconds, or 2.5 in the last one. A clicker that asks for more travel lengthens the pull and the time under tension; the traditional-archery study found less jerk with a final pull of one and a half to three centimetres than with a longer one, and the fatigue measurements find more tremor as the hold gets longer, although the first figure comes from traditional bows shot at close range and the second from holds taken to exhaustion. So before working on technique, it makes sense to check where the point sits at anchor and, if needed, to move the clicker.

The second is the shape of the pull rather than its speed. The second-best archer in the German group reached the click in two phases on all 66 arrows and had the second-highest average; what appeared on the worst arrows were stops and creeping. Without an instrumented clicker you can look for them by watching the point at close range or filming it from the side with a phone: a pause, or a slip forward of a few millimetres just before the clicker, is easier to catch in slow motion than live.

The third concerns the click. Only one study says that the release starts before the clicker drops, the idea that a faster draw sets the better archers apart is contested in the registry, and among strong archers the time from click to release seems to weigh on the score more through its consistency than through its brevity. On what these data show, then, the continuity of the movement that leads to the sound deserves more attention than the speed of the reaction to it.

@misc{campagna2026_trazionefinaletiroconarco,
  author = {Campagna, Matteo},
  title = {The final pull: what happens in the last millimetres before the clicker},
  year = {2026},
  howpublished = {matteocampagna.com},
  url = {https://matteocampagna.com/en/articoli/the-final-pull-in-archery},
  note = {accessed 2026-10-16}
}