Chronometer for archery shot phases: Ertan 2005

This is the testing of an instrument, not a study of shooting. The instrument is a field chronometer that measures four things during a shot: the time between the clicker falling and the arrow tip passing a sensor on the bow, the flight time and mean speed of the arrow, and then wind and temperature. Twenty elite Turkish archers, ten women and ten men from the national team and its candidates, shot twelve arrows in the laboratory for the validity check and twelve at 70 metres on the field, repeated a week later, for the reliability check.

Validity was assessed by comparing the device's time with the reaction time derived from electromyography of the forearm muscles. The two numbers do not coincide: the device gives a mean of 177 milliseconds against 146 from the electromyography, a significant difference (p = 0.027), and the gap is explained by what the device additionally includes, namely the first part of the fingers opening and the time the arrow tip takes to cover the centimetre and a half to the sensor. But the two numbers go together, with a correlation of 0.787. Across the 240 laboratory arrows the device's time ran from 102 to 502 milliseconds and the electromyographic one from 88 to 342.

On reliability the picture is good for almost everything. Between the two sessions a week apart, the intraclass correlation coefficient is 0.985 for clicker reaction time, 0.992 for flight time, 0.990 for mean speed and 0.938 for temperature; wind speed reaches 1.000 and wind direction trails behind at 0.742, with a confidence interval dropping as low as 0.456. The session means are very close: 180 and 179 milliseconds for clicker reaction, 1.290 and 1.287 seconds for flight time, 184.2 and 183.7 kilometres per hour for mean speed over 70 metres.

The authors conclude that the device validly measures a quantity that predicts reaction time, and reliably measures the rest, and that it can serve to assess an archer's technique and the interaction between bow and arrow.

The archers were twenty, from the Turkish national team and its candidates, recruited at a preparation camp shortly before an international competition; ten were juniors. The means: 19.8 years of age and 1,235 FITA points for the men, 20.7 years and 1,241 points for the women, with 6.5 and 7 years of training. For those under 18, consent was also signed by their parents.

The device, a prototype developed by the same authors in 1996, has four components: an LED screen, the clicker reaction time measurer, the wind speed and direction measurers and a vibration sensor. A two-headed metal conductor is fixed under the clicker, one head attached to the clicker and one to the bow handle: when the clicker falls, the two heads touch and the chronometer starts. On the handle, a centimetre and a half from the clicker, sits a 150 gram sensor sensitive to metal: when the arrow tip passes in front of it the count stops, and that interval is what the authors call the clicker reaction time. The flight time measurement starts from there and closes when a vibration detector on the target registers the impact and sends it telemetrically to the main unit; mean speed is the distance divided by that time. The wind measurers stand at the midpoint of the shooting distance, parallel to the line, and record speed and direction at the instant the clicker snaps.

For validity, in the laboratory, the electrical activity of flexor digitorum superficialis and extensor digitorum was recorded with an eight-channel telemetry system, synchronised with the chronometer and with a mechanical switch under the clicker to mark the instant of its fall on the traces. The skin was shaved, abraded and cleaned, surface electrodes placed longitudinally along each muscle about two centimetres apart, with the reference on the olecranon of the drawing arm. For each arrow five seconds were recorded, two and a half before and two and a half after the clicker, and the central two seconds were processed as absolute values. Reaction time was defined as the interval between the clicker falling and the onset of muscular activity, taken as the peak latency of the rectified signal: twelve arrows for twenty archers, 240 measurements.

For reliability, on the field, each archer shot twelve arrows at 70 metres in conditions suitable for shooting, without rain or strong wind, and a week later everything was repeated with the same archers, in the same shooting order and in similar conditions. The analysis uses the intraclass correlation coefficient in model 3,1, a paired t test between the two sessions, and Bland and Altman plots with limits of agreement at plus or minus 1.96 standard deviations.

The first thing to keep clear is that the quantity measured is not a reaction time. It includes the reaction time, part of the movement of the fingers, and the travel of the arrow tip to the sensor, and on the authors' own definition it corresponds rather to a complete response time. They say so, but the name they give it, clicker reaction time, invites confusion with the electromyographic measure: the two differ by 31 milliseconds on average.

Validity rests on a single correlation, computed across the means of twenty archers. There is no comparison with a reference instrument for arrow speed, such as a chronograph, and that is a gap, because speed and flight time are precisely the quantities the device offers for checking equipment.

On reliability there is an internal tension between two analyses. The intraclass coefficient is very high for almost everything, yet the paired t test finds significant differences between the two sessions for flight time, speed, wind and temperature. The differences are tiny, 3.7 milliseconds on flight time and half a kilometre per hour on speed, and for wind and temperature it is obvious that the weather a week later is different: in those two cases the reliability of the instrument and the stability of the weather are confounded, and wind direction, the one measure with a mediocre coefficient, is also the most exposed to this. In the table of t tests, temperature is a case of its own: the t value is 1.190 with 19 degrees of freedom, the confidence interval of the difference includes zero, and the significance is printed as 0.000, three figures that cannot sit together.

Whether the instrument changes the shot is not checked. The metal-sensitive sensor weighs 150 grams and sits on the handle, where the balance of the bow is sensitive; the clicker was modified with a two-headed conductor. No comparison with and without the device is reported. The stated precision of the screen, a sensitivity of 0.1 per cent of a second and an error of 0.01 per cent, is phrased in a way that leaves it unclear what it refers to.

Finally the study connects nothing to the target. The authors write that the link between the measured time and the score can be observed and that its effects can be assessed: those are future work. Here no score was collected and no difference in performance was measured, and the sample consists only of elite archers, half of them juniors. The paper is from 2005 in a journal without a DOI, and describes a prototype built in 1996.

The study says nothing about how to shoot: it says how much a number can be trusted. Anyone using a device that measures the time between clicker and release will find here a check on that kind of measurement, made on elite archers and repeated a week apart. The clicker reaction time repeats well from one session to the next, and flight time and arrow speed are stable too, while wind direction is the least reliable measure, worth remembering if a wind reading is set beside a score.

The second useful point concerns what that number means. What a device of this kind records is not the nervous system's reaction to the click: it is a longer interval, 177 milliseconds on average against the 146 that electromyography attributes to the reaction itself, because it also includes the start of the fingers opening and the travel of the arrow tip. The two values are nonetheless well correlated, at 0.787, so an archer who reacts sooner tends to record a shorter time on the device as well. Anyone comparing their own figure with numbers read elsewhere needs to know which of the two quantities they are looking at.

The spread of the values is worth noting. Across 240 arrows from national team archers the measured interval ran from 102 to 502 milliseconds: the mean of 177 hides shots five times slower than the quickest. Whether those slow shots score fewer points is not this study's business: no score was collected, and the authors leave the link with the target to later work.