How Can Anybody Return a Ball at Two Hundred Kilometres an Hour? They Cannot React

A serve struck at around two hundred kilometres an hour covers the court in roughly half a second, and it slows on the bounce, so the returner has perhaps six tenths of a second from the moment of contact. A simple reaction to a visual signal takes around two tenths of a second before any movement begins, and moving the body and the racket into position takes far longer than the remainder. Waiting to see where the ball goes and then responding is therefore impossible at the top level, and any account of returning based on reaction speed is wrong. The returner must already be moving before they could possibly know where the ball is going.

Anticipation draws on information available before the ball is struck:

• •The ball toss position, which differs between serve directions
• •Shoulder and hip rotation during the wind-up
• •The angle of the racket face in the final instant before contact
• •Where the server is standing relative to the centre mark
• •Patterns from earlier in the match and from previous matches
• •The score, since particular serves are favoured at particular points

Experiments comparing expert and novice returners established the mechanism clearly. Showing video of a serve that is cut off at the moment of contact, and asking where the ball will go, produces accurate answers from experts and chance answers from novices, which demonstrates that experts are reading the body rather than the ball. Cutting the video earlier still, before contact, degrades expert performance progressively, which identifies which parts of the movement carry the information. Obscuring particular body regions with occlusion goggles isolates what is being used. Eye tracking shows experts looking at the trunk and the racket arm rather than at the ball. The skill is therefore perceptual learning rather than faster reflexes, and expert reaction times to simple signals are ordinary.

Published record speeds are less comparable than they appear and several factors inflate them. Measurement is taken at the racket rather than at the returner, and the ball loses a substantial proportion of that speed crossing the court and more on the bounce, so the speed a returner faces is considerably lower than the headline. Radar equipment differs between tournaments and has been shown to read differently at the same event in different years. Altitude matters, since thinner air slows the ball less, which is why records set at high-altitude venues are treated cautiously. And speed alone predicts effectiveness poorly, since a slower serve with heavy spin that kicks away is frequently harder to return than a faster flat one.

Because the returner is reading the delivery, the server's task is to make the delivery unreadable, and the techniques for that are deliberate. Using an identical toss and an identical body action for every serve direction removes the cues, which is a technical goal pursued explicitly by coaches and is what distinguishes the best servers as much as speed does. Disguising until the last possible instant compresses the time available further. Varying placement unpredictably prevents pattern reading, and analysis of professional matches finds that servers are measurably less predictable than chance would require in some patterns and more predictable in others. Speed matters and disguise matters at least as much, which is why the fastest server is not the most effective one.

A fast serve allows around six tenths of a second, which is too little to see the ball and respond, so returning depends on reading the toss, the body rotation and the racket face before contact. Video cut off at contact produces accurate predictions from experts and chance from novices, which shows the skill is perceptual rather than faster reflexes. Servers respond by making every delivery look identical.