Fingers on the string and release: Kim 2017 study

The study uses no bows and no archers. Eight right-handed men of about 30 pressed with their fingers on force sensors, as if holding a string, for 5 seconds, and then let go as quickly as possible, in three conditions: with index and middle fingers, with index, middle and ring fingers, and with all four. A model used the force at the moment of release to work out where a virtual projectile would have landed.

With more fingers, the force during the hold strayed less from the required value: the error, measured as root mean square error, was higher with two fingers than with three in all three directions (forwards and back, up and down, side to side), and higher than with four forwards and back and up and down. On the virtual target, precision, meaning the spread of hits from one trial to the next, was better with three fingers than with two. Mean distance from the centre did not change between conditions, and nor did the time taken to bring the force to zero or its variability.

In the maximal force test the index finger pushed a mean of 29.75 newtons, the middle finger 16.46, the ring finger 11.09 and the little finger 14.93. During the hold, adding fingers increased the sideways component of the force towards the back of the hand, with the middle and ring fingers carrying more of it than the index.

The participants sat with the right shoulder at about 45 degrees of flexion and 45 of abduction and pressed on four six-axis sensors mounted on a frame, oriented so that the push was horizontal. The force to be held was 50% of the maximal force of all four fingers together, measured beforehand, with zero newtons required vertically and sideways; a monitor showed the forces and where each shot landed. Each condition had 25 trials, with 5 minutes' rest between conditions and 10 seconds between trials, and the participant chose the moment of release without any external cue.

The aiming phase analysed runs from 1.5 to 0.5 seconds before release; the release starts when the rate at which force falls reaches 5% of its peak and ends when the force reaches zero. The virtual projectile had a mass of 1 kg, the virtual string a stiffness of 0.7 N/m, and launch speed fell the longer force stayed on the fingers during the release. Conditions were compared with repeated-measures analysis of variance and paired t-tests.

The task resembles a shot only in its structure: holding a force steady and then letting it go. The fingers push on rigid sensors rather than pulling a string that slides off and opens them at the release; there is no bow, no draw length and no clicker, and the load is a fraction of the fingers' maximal force, quite unlike the draw weight of a bow held at full draw. The target is computed from a simplified model in which a slower release costs the projectile energy, not from real arrows.

Eight participants is a small number, the paper does not say whether any of them had ever shot a bow, and the values for error, release time and precision appear only in the figures, with no numbers in the text. The four-finger condition does not match any grip used in finger shooting, where three fingers are the norm and two are sometimes used. Finally, the significant difference in precision concerns only the comparison between two and three fingers.

For finger shooters, the finding closest to practice concerns the number of fingers on the string: in this laboratory task, with three fingers the force stayed steadier and the virtual hits were less scattered than with two, without the release becoming slower. This is consistent with the three-finger grip, the most common in recurve and barebow, but it does not show that archers who use two fingers, or a different grip, lose precision, because nobody shot an arrow.

The idea that several fingers compensate for each other's errors comes from research on motor control of the hand, and here it finds support in the holding force. How much it matters with a string that slides, a bow at full draw and a clicker remains to be measured.