In 2010, a Greek scientist named Antonios Travlos set out to find the best way to teach a group of high school students to serve a volleyball.

He didn’t put it quite like that—he said he wanted to study “specificity and variability of practice, and contextual interference in acquisition and transfer of an underhand volleyball serve.” But that’s an awful lot of complicated words for what amounts to something pretty simple.

What was this guy trying to do?

What Travlos really aimed to get at was this. He wanted a group of kids to learn a particular skill, in this case, serving a volleyball to a specific spot on the court. And he wanted to find out the best type of training that would help them learn that skill.

First he tested a group of students, none of which had any prior volleyball experience, to see how accurate they were at hitting the volleyball to a target spot. Then he divided them into six equal  groups; five of which would practice Monday, Wednesday, and Friday for forty-five minutes each, and a control group which would not practice at all.

The first group, the “target” group, only practiced serving the volleyball to that exact target spot, and the second, “different” group, only practiced serving the volleyball to a different spot. The last three groups all practiced serving to three different spots (none of which was the target). The only difference in these three groups was the order they practiced serving those spots: The “blocked” group practiced their serves in “blocks”—they served a bunch of shots to the first spot, then a bunch to the second spot, then a bunch to the third spot. The “serial” group served in the same sequence every time: 1-2-3-1-2-3-1-2-3. And the “random” group served to those three spots in a random order called out by their PE teacher—they might go 1-3-2, 2-1-3, 3-1-2, and so on.

Then, after their week of practice, they all rested and were tested the following Monday (72 hours later) on how accurately they could serve to the target spot.

Who did the best?

It probably won’t surprise you that the “target” group, which had spent all their time practicing their serves to that one pre-determined spot, was the most accurate. What might surprise you, however, is how the other groups did. The “random” group was a very close second, and did much better than both the third-place “serial” group and the fourth-place “blocked” group. And the fifth group, the “different” group, did far worse than any of the other four, almost as badly as the control group which hadn’t touched a volleyball over the previous week!

The other major finding was how well each group retained its skills. The “target” group did about as well during the Monday test as they had the previous Friday at their last practice. So did the “random” group, which was almost as accurate hitting the target spot as they had been to the three different spots they had practiced serving to. The “serial” group was a little worse on Monday than they had been on Friday—but the “blocked” group, which had had some of the best results during practice, tested drastically worse. Remember, every group had the exact same number of repetitions, and the “random”, “blocked”, and “serial” groups had served to the same three targets the exact same number of times. The only difference was the order they served in.

Remarkably, , the “random” group not only retained their skills after seventy-two hours, but also had the best results serving to a totally new spot—almost as well as the group that had practiced serving exclusively to the target, and almost as well as they had done serving to the spots that they had practiced. And, of course, they did far better than the “blocked” and “serial” groups!