On the shooting line you hear it all the time: hold your follow-through, don't drop the bow, let the string hand finish its movement. Sooner or later almost every archer I coach asks me the same thing, and it is a fair question: what should I do after the release, and why does it matter if the arrow has already gone?
The measurements side with the doubt on one point and complicate everything else: the arrow leaves the string within a few thousandths of a second, but afterwards the body moves a great deal, and not in the same way on good arrows and worse ones. The studies are almost all small, and here I set out what they saw and how firmly it can be said.
How long the arrow stays on the string
The place to start is a camera running at 4,000 frames per second, trained on sixteen world-class recurve archers at the London Olympic test event in 2011, shooting at 70 m. From the moment the fingers begin to open, the string leaves them after 1.9 milliseconds on average; the nock reaches its closest point to the body at 5.4, the arrow straightens at 8.6 and separates from the string at 14.6, with differences between men and women under a millisecond and a half. It is a single systematic measurement of the sequence, and it is clear-cut; a 1987 measurement on a single archer gives twice as much, though, perhaps starting from a different instant, so the registry lists the precise figure as contested, while the order of magnitude is the same in the models of the bow too.
The same footage holds a second finding: in all sixteen archers the bow did not move sideways by more than a millimetre, or rotate by more than a tenth of a degree, while the arrow was on the string. The limit is the camera's resolution, and a Korean study of just two archers found the same in the then world number two and a movement of one or two tenths of a millimetre in the university archer. So the bow staying still while the arrow is on the string is also a reported claim, measured mostly at the highest level.
Then the arrow flies. At 70 m, in thirty-nine elite recurve archers filmed with high-speed cameras, the flight took 1.24 seconds, and an independent measurement, taken with a chronometer mounted on the bow, gives the same order of magnitude; with only two studies, though, that flight time remains a reported claim. Put side by side, these figures answer the first half of the question: the string's push is over within the first few hundredths of a second, when the bow, at least in the top archers measured, has not yet moved by any appreciable amount. The follow-through starts after that.
What the body does after the release
If the arrow is gone that quickly, the follow-through is everything that happens afterwards, and there the body moves far more than while aiming. Three independent samples of archers, measured in different ways, point the same way, but none of the three compares the two phases with a statistical test and the measurement windows differ in length, so in the registry the claim remains reported. In the strongest, 39 elite recurve archers measured on a force plate before a World Cup stage, the centre of pressure under the feet travelled 20 millimetres on average in the second before the release and 48 in the half second after, and peak sway velocity always came after the arrow had gone.
A more recent study of 37 pre-elite archers in Taiwan shows where the movement is concentrated. In the 19 recurve archers, the string-hand wrist went from 7.5 degrees per second in the half second before the release to 271.4 in the tenth of a second after, and the bow-hand wrist from 12.4 to 132.6, while the ankle barely moved, from 1.8 to 3.9. The jump is mostly in the arms and the bow.
The shoulder tells the same story in degrees. In the archer who was then world number two, filmed over sixty arrows at 70 m, the shoulder moved 2.5 degrees horizontally between anchor and release and 31.4 after it; in eight experienced Korean university archers, horizontal rotation fell between 25 and 35 degrees in about three shots out of four, with a mean of 31.2. On range the two studies agree, on speed much less, and the typical rotation values remain a reported reference: about thirty degrees is a fairly stable figure, though measured on nine archers from a single research group, while speed, between 94 degrees per second in the single case and a mean of 135 in the second study, does not yet give a reference.
The string hand does not stay still either. In 1976 a Japanese group saw it slide back at the release in experienced archers and stay where it was in beginners, without measuring it; in 2007, in the few milliseconds between the release and the arrow leaving, the top archer's hand moved 1.24 millimetres front to back against 0.70 in the university archer. These are very few people, and the claim remains reported.
So it is normal for the body to move after the release. Part of that movement may simply be the bow unloading, but how much is passive and how much is the archer's own doing has not yet been measured, and for anyone who shoots that is the gap that matters most.
The muscles that stay switched on
The idea that shoulder and back muscles should stay active through the follow-through is the oldest on this subject. In 1976 electromyography on four experienced university archers and six young people who had never shot showed that in the experienced archers the signal from the muscles that stabilise the shoulder held or grew after the release, whereas in the beginners it stopped. Eleven years later another group put a duration on it: in the two world-class archers measured, back and shoulders stayed active for 1.5 to 1.7 seconds after the release, while in beginners the activity collapsed at once.
The link with scores came in 1993, from ten Japanese club archers, and that link still rests almost entirely on it. The four with a best round above 1,200 points kept at least 79% of their aiming activity through the follow-through, and on the string side the ratio between the two correlated with the best round at 0.79 across eight archers. It is an association between people, not between arrows, and a coefficient that rises to 0.95 when the sample is narrowed to the six best should be read with caution.
The same study also shows that keeping the shoulder active does not mean keeping everything on: the ascending trapezius rose to 118% of its aiming activity on the bow side and 147% on the string side, while the other four muscles recorded fell to around 60%. Twenty-five years later, in the Brazilian national team, between 50 and 350 milliseconds after the clicker, half the muscles recorded held their activity, posterior deltoid and latissimus dorsi among them, while the other half, including biceps, triceps and upper trapezius, dropped at once.
Later measurements complicated the picture rather than settling it. In eight Korean university archers the range of shoulder rotation in the follow-through was practically the same on arrows scoring 7 to 10; in four elite Turkish women archers the deltoid, in the second after the clicker, was slightly less active on 10s than on 8s, at 29.7% against 31.6% of maximal contraction; in elite French archers, arrow by arrow, what changed in the follow-through was the trunk, not the shoulder. Between levels the direction holds: in the thesis that gathers the French work, the eight elite archers kept the middle and upper trapezius of the drawing arm more active than the seven intermediates.
That is why the entry on an active follow-through remains among the reported claims, and its summary puts it plainly: in the best archers the muscles stay switched on, but nobody has shown that arrows shot with the muscles switched on land closer to the centre.
Slow or fast
If range of movement does not separate the arrows, that leaves speed, and here the studies point in opposite directions, so much so that the claim about a quicker follow-through is contested. The first figure comes from the world number two, over sixty arrows: shoulder speed after the release was linked to the score, but with a positive sign in one table and a negative one in the equation in the abstract, and together with a second factor it explained only 15.6% of the differences, so the registry keeps it as context. A year later the same research group measured eight university archers, and on the better arrows the drawing shoulder rotated more slowly: about 158 degrees per second on 7s, 142 on 8s, 119 on 9s and 121 on 10s. On the worse arrows the velocity also had a peak, a jerk, that the better arrows lacked, although the regression is only significant at a loose threshold.
The study pointing the other way measures something else: how long the phase lasts. In sixteen young Malaysian recurve archers, 36 arrows each at 70 m, the phase from the release to the end of the follow-through lasted 2.40 seconds in the best eight and 2.72 in the weaker eight, and across 576 arrows a shorter phase went with slightly higher scores, with a very weak correlation. A shorter final phase could mean a cleaner follow-through or a bow arm that drops sooner, and the study cannot tell the two apart.
The 2007 single case contains a detail that I find the most useful. The 10s were neither the slowest nor the fastest arrows, but the ones in which shoulder speed varied least from shot to shot, with a coefficient of variation of 9.9% against 16.8% on 9s and 11.2% on 8s. It is one archer, but it fits with the jerk the later study found on low-scoring arrows, and my reading is that how repeatable the movement is matters more than how fast it is: a reading that nobody has yet tested on a group.
Elite and intermediate archers
Between one level and another the measurements do not agree. In the French thesis, eight elite and seven intermediate archers at 70 m, the elite archers had a larger area, larger ranges and higher velocities of the centre of pressure in the half second after the release, the opposite of what they did while aiming. Another comparison, six elite and seven sub-elite archers at the same distance, found no difference in sideways movement after the release, between 15 and 20 millimetres on every shot, and saw only the timing change: the elite archers started their balance adjustment before the release on 77% of arrows, the sub-elite on 59%. A third comparison, of 27 recurve archers at three levels shooting at 18 m, measured the forces under the feet rather than displacement and found them lower in the elite archers in every phase, but with no significant differences. With one comparison for and two against, all small and using different measures, the idea that elite archers move more after the release is contested.
Within the same level, by contrast, the studies all point the same way. Among the 39 World Cup elite archers, arrows on which the body moved faster after the release scored fewer points; among the pre-elite recurve archers in Taiwan, centre of pressure displacement just after the release predicted a lower score; among the eight French elite archers, the worst arrows had more front-to-back movement. Moving a lot after the release is not a virtue in itself, then, and the useful comparison is with your own best arrows rather than with a champion's follow-through.
The bow hand
The other piece of advice you hear on the shooting line is that the bow hand should not grip, even as the bow jumps forward. Three independent measurements point that way. In 1992 sensors on the grip of fifteen recurve archers showed that, in the last 150 milliseconds before the release, almost all the load passed through the palm, with low forces under the thumb and fingers, in national team archers and weaker ones alike. In 2009 electromyography on ten archers from the German and Turkish national teams and ten beginners showed that in the experienced archers the finger extensor, which opens the fingers, worked harder than the flexor across almost the whole two-second window around the clicker, whereas in the beginners the flexor surged between one hundred and two hundred milliseconds after the clicker dropped. Between half a second and seven tenths of a second after the clicker, with the arrow long gone, the beginners had a more active flexor than the elite archers.
Even in the experienced archers, though, the extensor's advantage disappeared between one hundred and three hundred milliseconds after the clicker, when both muscles peaked and then began to relax. A third group, studying thirty-two young Malaysian archers, found the finger extensor of the left arm more active in those who scored higher, without saying which was the bow hand, and right on the threshold of significance. So a bow hand that does not grip is a reported claim: the three studies point the same way, but the link with the individual arrow is still missing. The only study that looked at the bow hand arrow by arrow, in the French elite archers, found the extensor less active and the flexor slightly more active on the better arrows, and the authors speak of an optimal level of activation rather than relaxation, which is why a more relaxed hand on better arrows is a contested claim.
Where the research stops
Taken together, the studies describe the shape of the follow-through well and say little about how much it matters for the score. The direct link between active muscles and results still rests on the 1993 study of ten club archers; later measurements looked at different things, range, speed, duration, activation timing, the trunk, each with its own definition of where the follow-through begins and ends. And nobody has ever tried to lengthen it or make it more consistent through training and compared the outcome with a control group.
There is also a limit that concerns coaches. In a study of one experienced archer, an inertial sensor on the bow told apart the shots the archer judged good from the bad ones only in the two seconds after the release, yet the same difference could not be seen on video, even filmed from two directions at 300 frames per second. It is a single archer, and the point applies to me as well: what a coach judges by eye does not necessarily match what an instrument measures.
What I do with this on the shooting line
Back to the opening question. The follow-through does not steer the arrow, because by the time it begins the arrow has already had everything the string could give it. If it matters, it matters for two reasons. The first is that the movement after the release is the tail end of what came before, and in several of these measurements worse arrows, or less experienced archers, can be recognised precisely there: by the jerk of the shoulder, by the body moving faster, by the bow hand closing. The second is that an archer who knows the movement has to carry on past the clicker has no reason to stop a moment before it, but that is a coach's reading which no study has put to the test.
In my work with archers I use the follow-through mainly as a mirror. I do not ask for a set duration, because there is no measured reference value, and I do not ask for a big movement, because between levels the measurements disagree and within the same level more movement goes with worse arrows. What I look at is whether the follow-through looks the same from one arrow to the next: whether the drawing shoulder carries on without a jerk, whether the bow arm stays towards the target rather than dropping, whether the bow hand stays open. In my experience, when one of these changes, something has changed earlier, in the draw or in the aim, and that is where I look, knowing that the studies do not guarantee it.
The rest will come from studies that do not yet exist: one that links activity in the first seconds after the release to the impact point of each arrow, in the same archer and over many arrows, and a training study with a control group. Until then, the follow-through remains one of the best described and least explained phases of the shot.