Lake George Ice Records: Reading a Century of Freeze Dates

Freeze-over and ice-out describe different events

Freeze-over

The point at which the lake is recorded as completely ice covered. The Lake George Association's February 2026 account describes coverage from the south end to the north end.

Ice-out

The recorded end of the lake's ice cover, a separate observation from the start of freezing.

A season without full freeze-over

A winter when complete coverage is not recorded. Ice can still form in bays or other parts of the lake.

An empty cell

A missing entry in the document. It cannot be converted into a no-freeze observation.

The difference between the last two categories is visible in the archived PDF. It explicitly marks some seasons as not freezing, while other rows lack dates. Combining both into a count of open-water winters would turn missing information into an event. [4] None of these terms measures the thickness of ice at a particular location. A historical freeze date is not permission to walk, drive or fish on the ice. Conditions for a visit need current local information rather than a date from this collection.

What the different records actually cover

The archive also has a shorter list of seasons without freeze-over. It is a selected historical page, not a list maintained through 2026. An apparent mismatch with a later article can reflect an older endpoint or incomplete selection rather than a disagreement about the same fully defined period. [2]

A date printed at the top of a document gives useful context, but it does not fill the rows beneath it. The PDF's season labels extend farther than its complete observations. A chart must show the actual populated record rather than assuming every labeled season has a usable pair of dates.

The short ice season described in 2017

The Lake George Association's historical account reports full ice cover from February 13 to February 26, 2017, a span it describes as 13 days. This is a specific interval for complete coverage, not a statement that every piece of lake ice existed for only those days. [3]

The same account provides averages and historical extremes. Those figures retain the date boundary of that account. Repeating its average as a freshly calculated 2026 statistic would imply that the intervening winters had been included when they had not been checked here.

On February 4, 2026, LGA announced another full freeze-over. That later observation can be added as a separately sourced event. It does not, by itself, supply all the intervening observations needed to recalculate an average or a long-term trend. [4]

Two checks before drawing a continuous timeline

First, preserve the winter-season label. A season crosses a calendar-year boundary, so a December date and the following spring's ice-out date belong to different calendar years. Sorting the dates only by month and day can reverse their actual order.

Second, flag contradictions instead of silently repairing them. In the PDF's extracted text, the 1980-81 row presents a December freeze date whose year falls after its April ice-out date. That needs comparison with the original observation record before it becomes a duration. An earlier row also carries a year that does not match its season. These are reasons to inspect the underlying entries, not reasons to discard the entire archive.

A reproducible transcription should retain the printed value, record any correction separately and identify the supporting source. Its totals should distinguish confirmed no-freeze seasons from missing observations. The two-page DPW summary remains available for that work, including its unfilled cells and inconsistent years.