One arrow comes back from the boss with a chipped nock, another has taken a direct hit from the one beside it, a third has ended up in the gravel behind the target. Picking them up, you always ask the same question: can I still shoot this one? Alongside it comes another, less common but more serious, which surfaces when you hear of a shaft snapping at the moment of release: how can that happen, when a moment earlier the arrow was sitting whole on the string?
Research has never studied damaged arrows the way an archer sees them, with cracked carbon or scarred nocks. It has, however, measured the things the answer depends on: the loads an arrow takes, the point at which a shaft gives way, and how much a crooked shaft, years of vibration and a manufacturing defect matter. The answers differ widely in how firm they are, and each one has its own status in the registry.
Where an arrow takes its knocks
Archers take it for granted that an arrow takes its hardest knocks when it leaves the bow and when it hits the target, and hardly any in the air. Only one direct measurement has followed the arrow all the way: an electronic point with an accelerometer, screwed onto compound arrows. The instrument read up to 16 times the acceleration of gravity, and along the shaft the signal went beyond that range only at launch and at impact, while in free flight it stayed within it. By filming seven shots from a compound set at 64 lb with two cameras at 5,000 frames per second, with a light 23.2 gram arrow chosen on purpose, the authors estimated the two peaks: about 1,100 g at launch and about 3,700 g of deceleration when stopping in a bag target, more than three times as much.
For the launch the figures recur elsewhere, since models, calculations and a second measurement on another compound place it between about 600 and 1,250 times gravity, over 12 to 19 milliseconds. Flight and impact, on the other hand, have been measured by one group only, and only along the shaft, so the claim that the arrow is stressed hardest only at launch and at impact, while in flight almost nothing happens to it is reported. There are no figures for flight and impact with the recurve, or for hard foam bosses, and nobody has measured what an arrow takes when it strikes another arrow, the stand or a stone.
Why a shaft buckles
During the power stroke the string pushes the arrow from the nock, while the point, which carries a good share of the mass, has to be accelerated along with everything else: the shaft is squeezed between its two ends, like a column loaded from above. A long, slender column carries that load up to a threshold, the critical load that mechanics derives from stiffness and length with Euler's formula; beyond it the column does not shorten but bows out sideways, further and further. This is buckling, and you hear that an arrow too weak for the bow breaks in exactly this way, during the launch.
In the calculations available, though, the arrow stays below the threshold in a regular shot. One model of the bow and arrow system places the loss of stability at a dimensionless force of 14.79 and the peak thrust of an average target bow and arrow at 12.3. The only estimate on real equipment, a compound with a spine 500 carbon arrow, gives a peak axial load of about 60 newtons against a critical load of about 100, a margin the author considers acceptable. That is why the claim that during the release the arrow buckles under the load on its point is contradicted: the evidence goes against it, but it is scant and indirect, and nobody has instrumented an arrow during a shot close to the limit.
The arrow trapped between string and bow
The case in which a shaft breaks on release is a different one, and the only laboratory test comes from kyudo. There, at full draw, the point protrudes a little beyond the bow; if the archer draws further than they should, or uses an arrow that is too short, the point slips behind the bow, and on release the arrow is trapped between the string and the bow with the bow's full force on the nock. The authors, Japanese mechanical engineers, write that the accident is very rare and happens mostly to archers whose draw length is not yet settled or who shoot an arrow that is too short, but that it can also happen to intermediate and advanced archers, because in kyudo archers prefer the point to protrude only a little.
To reproduce it they took two types of shaft in duralumin, a 7000-series aluminium alloy, one meant for bows of 98 newtons and up and the other for bows of 147 and up, cut to 850 millimetres. They compressed them end-on in two ways, slowly in a testing machine and suddenly with a pneumatic cylinder filmed at 800 frames per second. The two shafts gave way at about 50 and 65 newtons, values that Euler's formula reproduces to within a few percentage points and that sit far below the force of the bows they are meant for. Once they had buckled they bent until they snapped, after about a quarter of a metre of travel in the slow test and in about two tenths of a second in the fast one, and the break always came exactly in the middle, with sharp metal edges.
The figure that explains the mechanism is the stress. In the fast test the shaft reaches about 30 megapascals, against the roughly 500 at which the alloy begins to deform permanently: the material is not overloaded, and it is the instability of the column that bends the shaft until it snaps. The authors warn that the fragments fly right in front of the archer's face. The claim that a shaft that buckles on release breaks in the middle and leaves sharp edges is therefore reported: the test is clear-cut, but it is a single one, on two aluminium kyudo shafts, and carbon shafts and target arrows have not been tested.
The same authors note that in target archery the drawing hand settles more easily at the chin and the draw is shorter, so the point is less likely to slip behind the bow. If it does, the mechanics suggest a similar situation, with the arrow squeezed between string and bow, but that is reasoning rather than measurement: how often it happens in Olympic archery, and how a carbon shaft breaks in that position, nobody has measured.
Spine and the margin
The margin between a regular launch and the threshold depends on the equipment. The compound estimate lists what narrows it: more draw weight, more point weight, a longer or weaker shaft; and the author, with the figures for his bow, expects arrows with a spine above 800 to fail. It is an estimate, in a preprint, and it describes no observed breakage, but it shows which way the margin shrinks, towards a shaft too weak for the bow shooting it.
This is the same rule that guides the choice of spine, which exists to let the arrow leave the bow cleanly. That arrow spine has to be matched to draw weight and arrow length is reported: every model that has looked at it agrees, while nobody has compared the groups of archers shooting a suitable spine and an unsuitable one. Those models also give two figures that matter here: moving to the next spine calls for about eleven newtons of draw weight, and ten millimetres of shaft are worth four and a half newtons, so a shorter arrow behaves like a stiffer one. Cut it too short, though, and the point moves towards the rest, closer to the case the kyudo test reconstructs.
A bent shaft
After an impact the most common question is whether the shaft is still straight. The registry rates the claim that a straighter shaft gives tighter impacts as contested, because the direct tests, all by machine, point in different directions depending on how different the shafts being compared are. The one closest to a shaft bent by a knock comes from an international group of archers coordinated from Australia: Easton X7 aluminium shafts straight to within a thousandth of an inch, bent deliberately at the centre with a straightener to three and then six thousandths, and shot from a compound mounted on a machine at 25 m. With the vanes angled at 0.9 degrees the group diameter went from 27 millimetres to 30 and then to 48.
In 2025, by contrast, on fifty-nine Easton X10 shafts shot by a machine at 70 m, straightness entered no model of the point of impact, because those shafts were far straighter than the tolerance required: 0.007 millimetres on average, with a maximum of 0.015 against the 0.025 declared. The two tests fit together if, within a modern competition set, straightness varies too little to matter, while a shaft well outside tolerance can open up the group. The authors of the bent-shaft test add that aluminium can lose straightness when shot into hard targets, and in the same test the nock changed the group more than straightness did: on the same straight shafts, changing the nock took the group from 44 to 27 millimetres. It is on this single test that the registry reports that the type of nock changes the size of the group.
Wear from vibration
It is also said that arrows do not tire, and that as long as the shaft is intact it stays the arrow it was on the first day. The only test is a 2019 preprint in which one arrow per material, carbon, aluminium and cedar, was put through about 600,000 vibration cycles and then measured again. The authors conclude that nothing significant changes, but in their tables the estimated stiffness after the cycles is higher in every measurement, by 52% in the carbon arrow and by between 14% and 29% in the others; the estimated mass of the carbon arrow also drops, though, from 10.3 to 9.0 grams, which an intact shaft should not do, so part of those differences is noise in the method. The claim that hundreds of thousands of vibration cycles do not appreciably change an arrow's characteristics is therefore contradicted, by evidence that goes against it but carries little weight. The test also covers vibration only: knocks on the boss and between arrows are left out.
Spotting the arrow that leaves the group
Many archers notice that an arrow has a problem because it lands outside the group more often than the others. A machine sees this better, because it takes the archer out of the picture. In a small 2020 test a machine with a 40 lb recurve shot six arrows at 70 m on a day with little wind; two were defective, one of them missing a vane. The four sound arrows stayed within a circle with a radius of twice the standard deviation, the two defective ones landed outside it, and each always on the same side, the one without a vane low and the other to the right. That an arrow-launching machine at 70 m picks out defective arrows is reported, because nobody has checked that the arrows the machine rejects are the ones that lose points in competition.
Without a machine, the problem is the noise in your own shooting. In the only comparison made at the same facility, at 55 m, an archer spread 107 and 108 millimetres vertically, against 34 for a robot and 9 for its improved version; there was a single archer shooting different arrows, and on this comparison the registry reports that a well-built shooting machine groups tighter than an archer. Picking out a defective arrow by hand therefore takes numbered arrows and many ends, because the defect has to stand out against a much larger error in the shot.
What I take from it on the shooting line
No study has tested a rule for deciding when to retire an arrow, and what follows is how I read these measurements on the shooting line, not a verified procedure.
The first point concerns length. The case the kyudo test reconstructs comes from a draw longer than expected or from an arrow that is too short, so the arrow should be cut so that the point stays in front of the rest even when, at full draw, you pull a few centimetres further than usual. This matters most for beginners whose draw length is not yet settled, and the length should be checked again whenever the draw length or the clicker position changes. An arrow that drops off the rest at full draw is not released: let down slowly and start again.
The second concerns spine. The margin between the load of the launch and the threshold narrows with more draw weight, more point weight and a longer or weaker shaft, so when I go up in draw weight I check against the manufacturer's chart that the spine still suits the bow. It is the same choice that tuning depends on, and the article on the archer's paradox explains why.
The third concerns inspection. Since the highest loads come at launch and when the arrow stops, I look the arrow over every time it has hit something other than the boss, above all at the nock and around the point. The check usually taught for carbon, flexing the shaft by hand and listening for a crackle, has never been tested by any study; for aluminium there is the spin test on the fingernails, and the authors of the bent-shaft test write that a shaft straight to within a thousandth of an inch spins smoothly, one at three thousandths a little less so and one at six bounces noticeably. An arrow with a crack, a splinter or a split nock I put aside without shooting it, because nobody has measured how a cracked carbon shaft copes with the launch.
Arrows that look intact but fly off line call for patience and numbers. An arrow that lands outside the group on the same side over many ends deserves watching: before blaming the shaft I change its nock, which in the bent-shaft test mattered more to the group than straightness, and if it keeps straying I put it aside. For a shaft that is still straight and unmarked after a knock, on the other hand, the data give neither a reason to throw it away nor any assurance that it is the shaft it was before, and there is no shot count for retirement with a measured basis. Separating a crooked arrow from the errors in your own shot is the work described in the article on arrow impact patterns.
Where the research stops
Almost everything known about how an arrow deteriorates comes from test benches and machines. Breakage through buckling has been tested on two aluminium kyudo shafts, and nobody has compressed a carbon target shaft in the same way; acceleration in flight and at impact has been measured by one group only, with a compound and a bag target; fatigue has been studied on one arrow per material and for vibration alone. Nobody has followed a set of arrows through a season to see when and how they crack. For now the decision to retire an arrow rests on well-understood mechanics, a few laboratory tests and a common sense that nobody has put to the test.