Three groups. One box.

These three sets of readings have the same minimum, the same quartiles, the same median and the same maximum, so a box plot draws them identically. It is not being careless: five numbers is all a box has. The samples behind them are not remotely alike, and the only way to see that is to look at the readings themselves. Press the button and each box comes apart into the 60 observations it was summarising.

The boxes are derived, not supplied: each datum carries its raw points and the library computes the five-number summary. That is also what makes the explode possible, because chart.rowSeries() can hand back the observations behind every mark. The dots land in a beeswarm (plotOptions.unit.layout: 'scatter'), which spreads equal values off the centre line instead of stacking them on top of each other.