Flip 10 Coins at Once

Ten is large enough to feel like a proper sample but small enough that a genuinely fair coin still produces a lopsided-looking split — a 7-3 or even 8-2 heads-to-tails result is well within normal statistical variation at this sample size, not a sign that anything is wrong. The standard deviation of a fair coin over 10 flips is only about 1.6, which means results a couple of flips away from an even 5-5 split are common rather than exceptional; the split only reliably tightens toward 50/50 as the batch size grows into the hundreds or thousands, which is exactly why a single batch of 10 is useful for a quick game but not for actually testing whether a coin is fair.

Ten flips at a time is a convenient unit for demonstrating the difference between theoretical probability and observed results in a statistics or probability lesson — students can flip a batch, record the split, flip again, and see firsthand how much a small sample can vary even though the underlying 50/50 probability never changes. Repeating the batch several times and tracking the running total across all of them is what makes the convergence toward an even split visible, since any single batch of 10 is too small on its own to reliably demonstrate it.