Comment by leereeves

> The appearance of two 3.5-sigma events suggests that the data is no longer being randomly generated around that mean, but is instead drifting away from that mean.

Not if the two 3.5-sigma events are in opposite directions, as in this case.