Anyone who has shot for a few years knows the scene: same distance, same bow, and thirty points fewer than yesterday. Or an average that has not moved for months, while the group, looked at closely, is no longer what it was. Both come down to the same question, how repeatable your shot really is, and research has approached it from several sides: it has compared archers of different standards, measured muscles and joint angles arrow by arrow, counted the arrows of head-to-head matches and asked archers how their good days and bad days go.
The answers are tidier than you might expect, although none of them is settled. As level rises, arrows land in tighter groups, but what becomes more consistent is not the whole movement: it is the part that decides where the arrow goes. Measurements of the movement, even taken together, account for only part of a single arrow's score, while the score of a whole round, in the only study that sets it beside laboratory measures, varies far less than muscle activity. From one day to the next accuracy changes a good deal, and in the only study that has measured it, archers with national placings stand out for the accuracy they report on their bad days, not their good ones. And after two arrows outside the 10, the third goes worse for everyone, in an archive of almost twenty thousand arrows from head-to-head matches.
That is why consistency is measured on the target, without a laboratory, as long as you keep the right numbers over enough arrows. Further down I explain how I do it with the archers I coach.
What consistency means
Consistency means shooting the same way every time, which is not the same as shooting well. An archer who puts every arrow in the 8, in a tight group, is more consistent than one who alternates 10s and 6s with the same average: the first has a shot to build on, and a sight adjustment brings the group into the middle; the second has a shot that changes from one arrow to the next without warning, and is far harder to correct because there is no telling which arrow to start from.
The technical term is dispersion, how widely the arrows are spread around their mean point of impact. It should be kept apart from centring, which is how far that mean point sits from the middle, because the score adds the two together and does not tell you which one is costing points. On paper a tight group off centre and a wide group well centred can give the same total, and in the only study that measured the score and the position of every arrow on the same shots, twelve young archers using an electronic bow on a projected target, two boys one point apart had groups of which one was almost twice the size of the other. That is why the registry states that two archers with the same score can group very differently, a claim with the status reported. Centring is fixed with the sight, as I explain in the article on reading your group; dispersion is not, and it is the subject of this article.
Consistency and level
Every study that has compared archers of different standards finds the group tightening as level rises, and none finds the opposite. The easiest comparison to read is also one of the oldest: in 1990 a Belgian group measured, at 18 m, the contour of three-arrow groups shot by three Olympic archers and six national-level archers. In the Olympians it was between 7.8 and 10 cm, in the national archers between 11.9 and 26.7, about twice as large on average.
Other samples say the same with different measures. In 29 Turkish archers, over 72 arrows each at 18 m, the spread of the beginners' impacts was five to seven times that of the high-level archers; at a 2013 World Cup stage the top twenty men's compound archers lost 0.30 points per 36 arrows to off-centre groups, against 1.35 for the average of their division. That is why the registry's claim that the higher the archer's level, the tighter and better centred the group is reported: there are seven independent samples and all point the same way, but they compare different people measured once, and the better centring of the top archers shows up clearly only in men's compound.
The same holds for scores. In 188 American recurve archers, split into four groups by their annual average, the standard deviation ran from 21 points in the world-class group to 38 in the weakest, with one exception, the national and collegiate group, the tightest of all at 14; in the Greek national team the standard deviations of competition scores were between 27 and 40 points, against 69 to 77 in unselected archers with three to five years of competition; in the Chinese national team the average per arrow at 70 m was 8.7 with a standard deviation of 0.5, in the Beijing squad 6.8 with 1.4. These figures need careful reading, because almost all of them measure how close the archers in a group are to one another, not how closely each one repeats themselves. The only sample that looks inside the individual archer is Swedish: in twenty young archers, score and the standard deviation of their ends moved in opposite directions, with a correlation of minus 0.65. The registry keeps the two apart, stronger archers are also the most consistent ones and as the level rises, scores within a group of archers sit closer together, and both are reported.
There is also an almost arithmetical reason. An archer shooting close to the 10 has little room above, so their ends cannot swing as much as those of an archer sitting on the 7, and consistency read off the score is largely the same thing as level, seen from another angle. Nobody, moreover, has followed the same archers as they improved: that the group tightens first and the score rises afterwards is what I see on the range, not a measurement.
What becomes more consistent in the shot
If the group tightens, something in the shot must repeat more closely. Between experts and beginners the direction is clear. In the 1990 Belgian study, comparing the traces of ten muscles from one shot to the next, 92% were identical in the Olympians, 41% in the national archers and none in the beginners; among high-school kyudo practitioners the position of the elbow changed between two shots by seven tenths of a millimetre in a third-year practitioner and by more than thirty millimetres in a beginner; in eight elite French archers, compared with seven intermediate archers over 36 arrows at 70 m, the arrow-to-arrow standard deviation of the score was 0.91 against 2.02. The claim that expert archers repeat the shot with less variability than beginners is reported, and no more than that, because these are all comparisons between different groups, often tiny ones, and nobody has followed the same archers while they learnt.
Among archers who are already skilled, though, the picture changes. In eight experienced archers, split into an elite group and a good county-standard group, the consistency of posture at release did not separate the two, and the elite archers often came out more variable; in the world's number two archer, over 60 arrows, the vertical shoulder angle before release was more consistent on arrows scoring 8 than on those scoring 10. The most persuasive reading comes from a study of six elite archers from a national squad, 90 arrows each at 18 m. On the arrows scoring 9 and 10 the shoulders, elbows and wrists varied in such a way that the variations cancelled out and the arrow stayed on line, with a compensation index at release of around 0.4 to 0.6, against 0.1 to 0.3 on the arrows scoring 6 to 8. It is a single study, so the claim that on better arrows the joints of the shot compensate for each other more is reported, but it suggests a different way of thinking about consistency: in those six archers the best arrows did not come from holding everything still, but from letting whatever does not move the arrow vary.
This also helps explain why everyone shoots in their own way. As early as 1990, in fifteen highly skilled archers, nearly all from the US national team, the activation of the two muscles that close and open the fingers on the string varied by 15 to 16% between shots by the same archer and by more than 50% from one archer to another. In 2025 a neural network, working on almost six thousand shots by four Korean national-level archers, identified who had taken the shot in 98.7% of cases from the paths of the shoulders, elbows and wrists alone. The choice was between four people, and the network may also have learnt build and camera framing, so the claim that body movement alone tells you which archer took the shot is reported. For a coach the point is a different one: an archer's consistency is measured against themselves, not as a distance from a model.
Two kinds of consistency in particular go with the score. The first concerns time: in seven archers from the German national squads, over 30 arrows at 18 m, the coefficient of variation of the time between the clicker snap and the release ran from 3.94% in the archer ranked second to 7.25%, and accounted for 66% of the differences in average score. The second concerns force: in fifteen recurve archers ranging from sub-elite to elite, over 150 arrows at 18 m, the more the shoulder force in the aiming position varied from one measurement to the next, the lower the score, and that single variable accounted for 45% of the differences, while maximal strength did not predict the score. These are associations between archers, in seven and in fifteen people, the second presented as a conference abstract, and nobody has trained those kinds of consistency to see whether the score rises. In the registry among elite archers, those with a more consistent release time score higher and holding the force steadier over time gives better scores are both reported.
How much of the score the measurements explain
So far it might seem that measuring the movement well is enough to know where the arrow will go. It is not, and this is the limit the whole registry runs into. Every study that has tried to put a figure on it finds that measurements of movement, even taken together, account for a minority of a single arrow's score, or for a share that varies widely from one archer to the next. The strongest sample, 39 elite recurve archers from four nations, reaches 42% with three variables, namely clicker reaction time, draw weight and sway speed after release; in the world's number two archer five shoulder and trunk parameters account for 17.9%; in three Malaysian national archers muscle activity does not exceed a third. In four young female archers aged 14 to 16, with a model built for each, the share runs from 28 to 72%, but with up to eleven variables over thirty or so shots, which is how high shares are obtained.
In eleven Japanese university archers, bow tremor while aiming tracked each archer's total score closely, with a correlation of minus 0.74, but had no link at all with the score of the single arrow. That is the difference between saying who is better and saying where the next arrow will land, and the authors themselves list what is left out: control of the forearm at release, the passage through the clicker, concentration in the follow-through, the straightness of the arrow, the wind. The registry sums it up like this: the movement parameters measured during the shot account for only part of an arrow's score. The claim is reported, because these are associations from a single session, in small samples and often in regional journals.
If laboratory measurements explain little, one might hope for the coach's eye or for a simple test, but here too the data are weak. Two studies find a link between rated technique and score, in 62 Indonesian archers and in 31 Thai archers; the first does not describe how technique was rated, and the second appears in a journal with questionable editorial standards and finds a moderate correlation, around 0.4. The claim that an expert's judgement of an archer's technique predicts their score is reported, and neither study describes a rating made without knowing the scores. In beginners, a test of throwing a tennis ball at a target on a wall was linked to the score with the bow, with r = 0.51, in 17 secondary school pupils at the end of a basic course; it was, however, the only one of nine measures to show a link, and with nine correlations one may cross the threshold by chance, so the claim that in beginners a target-throwing test is linked to the score with the bow is also reported. As for years of practice, the studies point in opposite directions: in the Chinese national team the official technical grades follow the years, though without a statistical test, while among 62 high-level archers years did not clearly separate European from national level (15.7 against 11.8, p = 0.061) and arrows shot per week did (696.5 against 273.7). The claim that years of practice tell how good an archer is is contested.
That leaves the target. Only one study sets the stability of the score beside that of laboratory measures in the same archers: fourteen university recurve archers, measured before and after twelve weeks. Over a simulated round of 72 arrows at 70 m the score varied between the two sessions by 3%, with a typical error of 4 points, while the electrical activity of the muscles in the second before release varied by 8 to 55% depending on the muscle. Two measures, however, challenge the score's lead: strength, with higher agreement between sessions than the score, and two draw-side shoulder angles, which also varied by 3%. That is why the claim that the score is the most repeatable measure we have of an archer is contested: on the word "most" the same study points both ways, and between the two sessions there was a training programme that raised the score. What can be said is that the score of a whole round moves far less than an electromyography trace, and that as far back as 1969, in a course for beginner schoolgirls, two rounds shot on consecutive days correlated at 0.78 to 0.80.
How much it changes from day to day
Anyone who has shot thirty points fewer than yesterday wants to know whether that is normal. The only direct measurement comes from kyudo: fifteen Japanese university archers shot twenty arrows at 28 m in the morning, on two or three days of the same week. None of them hit the same percentage in every session, and between the best day and the worst the average went from 83.7% to 61.7%, twenty-two points. Part of that gap is produced by chance, because twenty arrows are few and the best and worst days are picked afterwards, out of two or three; the study does not separate it out, and the claim that an archer's accuracy varies a great deal from one day to the next is reported, on a single study and in kyudo, where a hit or a miss is counted rather than a score out of ten.
The same group provides the figure most useful to anyone who competes, although it comes from the archers' own estimates rather than from a count. Those with national placings did not report higher accuracy on their good days, 89.3% against 85.0%, but did report higher accuracy on their bad ones, 64.3% against 48.8%. It points the same way as what we saw about level: in those estimates, what sets the better archers apart is not so much their best day as their worst.
How much chance weighs on a score is the subject of another article, on how many points it takes to say one archer is better: two identical archers, over the same 72 arrows, end up a few points apart by chance alone. Before looking for a cause for those thirty lost points, then, I check how many arrows lie behind the comparison.
The second question is whether accuracy drops in competition. On the range it is taken for granted, but the few studies that compare the two do not find it. In five university kyudo archers, followed in the days before a competition and at the competition itself, the probability of a hit did not fall; the only result in the other direction concerns three Korean women archers in contention for the national team, who shot below their training level at the first competition of the season and above it in the next two tournaments. The claim that accuracy does not drop in competition compared with practice is reported, on samples of three to six people. Tension too, in the only study that has tracked it within a competition, ten Brazilian archers in a simulated 60-arrow indoor round, stayed low and flat, 3.1 at the start and 2.2 at the end, while reported fatigue rose; and the two halves of the round gave similar averages, 218.8 and 216.0 points. In the registry the tension archers report does not build up over the course of a competition and scores do not fall in the second half of a competition, even though fatigue builds up are both reported: almost all the competitions are simulated, the samples are small, and the two halves are nearly always compared without a test.
After an arrow outside the 10
Within a competition the question changes, and becomes what happens after a poor arrow. The largest data set comes from elite recurve head-to-head matches: 19,122 arrows, in 6,374 ends of three, from the London, Rio and Tokyo Olympics, the 2021 World Championships and a European Grand Prix. After two arrows outside the 10 the third was worth 12.2% less than after two 10s, and a single arrow outside the 10 among the first two was enough to pull it down a little. The same archive shows that in that situation the men's third arrow was worth 3.5% more than the women's, while in general the last arrow of the end does not differ between the sexes, and that the gap widens by 2.4% when the end must be won to avoid elimination. The claim that after two arrows outside the 10, elite women archers recover less than men on the next arrow is reported: it is an observational analysis that has never been replicated, and it says that this happens in competition records, not why.
Whether one good end leads to another is a different question, and here the two data sets disagree. Counting the ends of a whole season in three top Brazilian archers, an end in the best third of their own was not followed by another as good in 67 to 74% of cases, roughly what you would expect if the previous end counted for nothing; across the arrows of head-to-head matches, on the other hand, there is a streak, but it shows mainly on the side of mistakes. The first finding rests on three people, the second sits within a set of three arrows in which the archer knows how the match is going, and the claim that after a good end it is easier to shoot another good one is contested.
On what to do after a mistake, research describes more than it measures. In an old study of almost two hundred American archers, dwelling on past mistakes was the mental item in the questionnaire most closely linked to the season's average score; it was, however, a snapshot taken once, and an archer who shoots badly simply has more mistakes to dwell on. In fifty young Malaysian archers, moreover, worry about making mistakes was the only mental skill that did not separate the better archers, so the claim that archers who dwell on past mistakes shoot lower scores is contested. As for how archers respond, a French study from 2001 followed a coach and four national team archers through two competitions: after a disappointing end one of the typical sequences was to look for the fix on their own, trying to relax, rehearsing the shot mentally and talking to themselves, while after a good end they changed nothing. In other sequences the same archers went to the coach, and four archers described without measurements cannot say which route works better: the claim that after a poor end, one of the ways elite archers respond is to look for the fix on their own is reported. In practice, between ends I work with archers on the pre-shot routine, which is the part of the shot that can be repeated unchanged even after a 7.
What makes consistency drop
When dispersion grows over a session, the first suspect is fatigue, and here there are measurements. In seven university archers held in the shooting position until exhaustion, the tremor arising in the posterior deltoid showed up in the oscillation of the limb; in twelve Italian archers, after six sets of ten 20-second holds at full draw, lateral bow sway had increased in all of them, by 6 to 39%, and less in the stronger archers, with a correlation of minus 0.65 between the increase and FITA score. In that same study the score did not change in anyone. In the registry muscle fatigue increases bow sway while aiming and higher-level archers cope better with the fatigue of shooting are both reported; the second also draws on a comparison over 72 arrows at 70 m, in which a beginner lost 10.8% of grip strength while an archer from the Brazilian national team stayed stable, but that is two people.
Archers say as much about the end of a session: of 59 university kyudo archers, 58% answer that at the end of training they could still shoot, but not with good form. It is a questionnaire, and it comes from kyudo alone, so the claim that at the end of a training session archers can still shoot, but not with good form is reported. In competition, as we have seen, the score of the second half does not fall: fatigue is felt and shows up in the bow, but in the few studies that have looked for it, it does not reach the scorecard.
When dispersion grows still helps to work out where to look, and it is the reading I use with my archers. If it grows only in the last ends, the first suspect is fatigue, and I look at physical conditioning; if it is high from the first arrows, the cause more probably lies in the fundamentals, an anchor that shifts by a few millimetres or a release that does not repeat; if it appears only in competition, the mental side and a routine that breaks down under pressure come to mind. It is a rule of thumb, not a measurement, and the data above make it more cautious: on average accuracy does not drop in competition, so when the group opens up in a competition I count the arrows before blaming nerves.
How to measure it on the range
Measuring consistency takes nothing more than the habit of keeping records, always at the same distance and on the same target face. I use three methods, from the simplest to the most refined.
-
Group size
Over twelve arrows, measure the distance between the two that are farthest apart and note it down every session. It is a rough measure, because it depends on two arrows only, but if it falls over the weeks, and not just in a single session, you are becoming more consistent.
-
The first twelve and the last twelve
Compare the group from your first 12 arrows with the group from your last 12 in the same session. The difference tells you how much fatigue widens your group, which is not the same as how many points it costs you: in the laboratory fatigue makes the bow sway more, but in the few studies of competition the score of the second half does not fall.
-
The spread between ends
Record the score of every end across several sessions and look at how much it swings around the mean: the smaller the swing, the more reliable your shot. It is the figure I use most often, because it comes straight from the scorecard you fill in anyway.
Studies on measurement add two cautions. The first concerns the number of arrows: a score says a lot when it adds up many arrows, while over a few shots in quick succession almost everything looks repeatable, and a comparison between two days should be read knowing how much chance contributes. The second concerns physical tests, where you often hear that one trial is not enough. It depends on what is being measured. For draw force, as far back as 1967, three trials in the same session in 49 beginner university students gave very similar values, with correlations of 0.91 to 0.98 between the best trial and the mean of the other two, so a single trial ranked the archers almost as well as three. For the electrical activity of muscles repeating helps, but the only data on archers come from a journal with questionable editorial standards, and there five maximal trials in quick succession also made the signal drop. That is why the claim that one trial is not enough and strength and muscle activity measurements on archers need repeating is contradicted: for draw force, one trial done properly already tells you a great deal.
Training for consistency
On how to train consistency, research says less than I would like. Nobody has trained the consistency of force to see whether the score rises, and the only attempt to train the consistency of the release, in intermediate archers in 1998, reports an improvement without giving either the scores or the number of archers. What follows, then, is how I work, built on what the measurements say but not proven by an experiment.
The first point is to work on the part of the shot that moves the arrow, not on all of it. The study of the six elite archers indicates that the best arrows are not the ones with the least movement but the ones in which the movement compensates, and the release traces show that every archer has a signature of their own. So I do not ask an archer to repeat an ideal form, but their own references and their own timing: a draw that comes through the clicker the same way every time, a similar interval between the snap and the release, the same sequence in the routine.
The second is the blank bale at close range, which I use to work on the shot when the target face is a distraction. I use it knowing what it can and cannot do: the only study that has listened to archers shooting at a blank bale concerns target panic, and there the relief was real in front of the bale but for several archers vanished as soon as the target face went back up, so much so that the claim that shooting at a blank bale cures target panic is contradicted. The work done on the bale, then, I bring back to the target face gradually, and that is where I measure the group.
The third is measuring. An archer who logs their own dispersion week after week sees things that the feel of a single day hides, and with the day-to-day variation we have seen it makes sense to decide on the trend over several sessions, not on the last one.