[WayneS]

The study by the World Weather Attribution analyzed weather records dating back to 1880 and found the cold weather that hit a swath of the U.S. from Maine to Minnesota tends to happen once every 250 years. In the early 1900s, it happened about once every 17 years.

Let me get this straight. Using 137 years of data they determined that a certain event occurs once every 250 years. But they also know that a little over 100 years ago (in the early 1900s), the same event occurred once every 17 years.

I'm going to have to call bullsh!t on that.

[Myrddin]

Typical ignorance in sampling data. The Nyquist limit requires sampling at twice the highest frequency expected in the data. The corresponding inverse is sampling period. You must sample long enough to capture a full repetition of the lowest frequency events or you miss them. Those rules are more obvious to people doing digital signal processing, but they apply equally to other kinds of sampling.