Finger motion at the release: Horsak 2009 study

Eight infrared cameras filmed the fingers of a recurve archer from the Austrian B national team, aged 49, with an indoor best of 1131 points out of 1200, as he released the string. Of the 90 shots at 18 metres, 56 proved usable. The question was whether the speed at which the middle finger opens separates the better arrows from the worse ones, and the answer is no: the maximum and minimum angular velocity of the proximal interphalangeal joint of the middle finger does not separate arrows in the innermost ring from arrows scoring 8 or less (p = 0.870 and p = 0.221), and the correlations computed in addition are not significant either.

The picture of joint ranges of motion is firmer. The finger that moves most is the index, with 81 degrees in total (standard deviation 5), followed by the middle finger with 58 and the ring finger with 48. The knuckles, that is the metacarpophalangeal joints, stay almost still, between 6 and 9 degrees, and the movement lies in the two outer joints: for the index and ring fingers the larger share belongs to the distal joint (52% and 57% of the finger's range), for the middle finger to the proximal one (55%).

There is finally an observation made by eye. Taking three arrows at random from each group and aligning the curves on the minimum of the middle finger's proximal joint, the shape before and after that minimum is almost the same in the good arrows and visibly variable in the poor ones. From this the authors suppose that what counts is repeatability, and they tie it to the idea that the motor programme of the release is already under way before the clicker falls. In the table the mean maximum velocity for the poor arrows is printed as 687 degrees per second, while the minimum and maximum for the same group are 5675 and 7895: a digit is missing, and the comparison has to be read against the figure for the good arrows, 6923.

The archer shot in three sessions of 30 arrows, ten ends of three, in the biomechanics laboratory of the University of Vienna, at 18 metres on a 40 cm indoor vertical triple face. The system is a Vicon with eight infrared cameras, six at 1.3 and two at 4.0 megapixels, recording at 500 hertz. Accuracy was checked by tracking two markers mounted on a rigid object, with differences in length under 0.2 mm. The data were interpolated and differentiated with a B-spline approximation at 2000 hertz. The markers, semicircular and 9 mm across, were placed on bony landmarks of the right hand following the Vicon hand model, which computes flexion, extension, abduction and adduction for all the fingers; the analysis covers the index, middle and ring fingers, the three used in the grip most common in competition.

Of the 90 shots, 56 remained, because in the others a marker was hidden from the cameras. The shots were split into two groups following a criterion already used in the literature: very good arrows, those in the innermost ring, and poor arrows, scoring 8 or less, while the 9s and 10s were excluded. The comparison is a Mann-Whitney test. The two variables are the maximum and minimum angular velocity of the middle finger's proximal joint during the release; separately, the range of motion of each joint was computed, together with the sum for each finger and each joint's share of that sum. The same 90 shots, and the same 56 analysed, reappear two years later in a journal article by Horsak and Heller, which adds the lateral deflection of the string.

One archer only, so nothing can be said about differences between archers, nor about how far this hand resembles another. The number of arrows in each of the two groups is not given, and without it a p of 0.87 cannot be weighed: there is no telling whether the comparison was twenty arrows against twenty or six against eight. A third of the shots were lost because a marker was covered, and the paper does not say whether the losses fell evenly among good and poor arrows.

Leaving out the 9s and 10s sets the extremes against each other, which widens any difference: finding none, under those conditions, is a result that counts, but it stays a result on one person. The outcome is the score of a single arrow, which also depends on aiming and on the tuning of the equipment, and the authors themselves concede that the absence of a difference may come from the poor accuracy of the derivatives used to obtain angular velocity. The finger kinematics come from markers on the skin of small joints gripping a string under tension, which is the hardest case for a model of this kind, and the paper reports no error figure specific to the fingers. The observation about variability, finally, rests on three arrows per group chosen as examples and is not quantified at all: it is why the same authors took it up two years later with a proper analysis. The text is a conference paper.

The most useful measurement is the one of joint ranges. The release is a movement of the two outer joints of the fingers, while the knuckles stay almost still: the string leaves the fingers as they uncurl, not as the hand opens. Anyone teaching the release in terms of fingers relaxing is describing what can be seen here, and it is worth knowing that the index finger moves more than twice as far as the ring finger.

On the rest, the study supports nothing. In this archer the speed at which the middle finger opened did not separate the arrows in the centre from those outside, and the difference that could be seen in the curves was in how alike one shot was to the next: an observation made on three arrows per group, not a measurement. What remains, then, is that the speed of the finger is not a parameter to work on, at least on this basis.

For anyone filming themselves there is a practical fact: none of this can be seen with an ordinary camera. The movement of the fingers lasts a few tens of milliseconds, here it took eight infrared cameras at 500 frames per second in a laboratory, and even then a third of the shots were lost because a marker was hidden by the hand.