Ice amount gave no clear linear relation to peak date spread across four Great Lakes in 1973-2026

We tested whether winters with less ice had more widely separated peak dates across Superior, Michigan, Huron and Erie. We used 54 winters in a fixed NOAA GLERL table for 1973-2026. Ice amount was the mean of four lake-specific annual maxima. Date spread was the standard deviation of their peak dates. A linear fit with year adjustment gave -0.71 days of spread per 10 percentage points of cover. A 95% paired five-year block interval was -1.46 to +0.24 days. All nine planned checks also had intervals that include zero. The result gives no clear linear association. It does not establish equal timing spread in low- and high-ice winters. Chart frequency changes, unknown peak tie rules, a short daily-chart period, one summary table and block assumptions limit inference. Source records remain external. The package contains code, a plan, exact source provenance, aggregate results, slope draws, a figure and an independent audit.

Great Lakes · ice cover · peak dates · date spread · null result · block bootstrap