Snow? 90 degrees? On a historic weather day, an early look at winter ahead.

Temperatures on a raw and windy Wednesday afternoon were a good 40 degrees lower than they were just one week ago, when they hit a record 95. The nontechnical term for such meteorological lunacy would be “October.”

This is a notorious transitional month, when the daily loss of daylight reaches a seasonal peak around here — better than 2½ minutes a day through the 17th — and the mid-latitudes are caught in the crossfire of the annual epic battle between the advancing winter and the reluctantly retreating remnants of summer.

Further evidence of the sometimes chaotic results: On Oct. 10, 1939, the region had its latest-in-the-year 90-degree reading in the period of record, dating to 1874; on Oct. 10, 1979, Philadelphia had its earliest measurable snow on record, 2.1 outrageous inches.

This also is the season for long-range forecasters to engage in an epic battle of their own, trying to solve the enigmas of the atmosphere and take their best shots at predicting just how triumphant winter ultimately will be.

Among the forecasts publicly released so far, the consensus is that winter will get off to a late start with a generally mild December, with any cold and snow more likely at the back end of the season. Given recent trends, a warmer-than-normal winter is a reasonable bet, the government says.

But winters never replicate, and the last one was a particularly rough one — for the forecasts. And in the early going, says Paul Pastelok, the veteran seasonal forecaster for AccuWeather Inc., the commercial service in State College, Pa., the atmosphere has been stingy with clues about the 2019-20 season.

“This is one of the toughest winter forecasts I’ve ever known,” he said.

For now, AccuWeather is calling for temperatures and snowfall to finish near their normals after a mild December. Average winter temperatures are in the mid-30s, and snowfall is just over 22 inches.

WeatherBell Analytics, a commercial service based in New York, was on the same page (should we say website these days?) with AccuWeather on temperature and early warmth, but was more generous with the snow, calling for about 28 inches in Philadelphia.

The government’s Climate Prediction Center has odds favoring above-normal temperatures in much of the nation, with the best chances in an area from Delaware to Maine, including Philadelphia, and the West.

(We would add that all rumors aside, the Old Farmer’s Almanac is predicting quite a mild winter for Philadelphia and the state, with below-normal snowfall; but feel free to ignore it. First off, with the almanac’s publishing deadlines, that winter forecast could well be a year old. And it is worth noting that Philadelphia shares a zone with both Richmond, Va., and Boston. The almanac’s claim of 80% accuracy is “simply an estimated average,” says spokesperson Ginger Vaughan.)

So where would one look for clues if they were to be found? One obvious place is the temperature record. The climate center’s long-range outlooks rely heavily on trends, and in general, recent winters have been warmer than they were in the 20th century, parallel with worldwide warming trends.

In Philadelphia, the winter average temperature in the last 30 years, 36.4, is about 2 degrees higher than the 100-year 20th century average. The 1990s were particularly warm.

As other forecasters, AccuWeather’s Pastelok looks to the tropical Pacific for help. Weather moves west to east, and when sea surface temperatures out that way are particularly warm or cold, they agitate the overlying air, and that can exert a strong influence on U.S. winters.

Right now, the ocean temperatures are no great shakes either way, although Pastelok said that one model, the European, sees a potential warming pattern in the Pacific that would favor milder conditions and mostly rainstorms for the East. Overall, however, the models haven’t been much help, he said.

Another place to look would be Siberia, where a robust snow cover in October could foreshadow a cold winter in the East, a hypothesis championed by Judah Cohen, a scientist at Atmospheric & Environmental Research in Massachusetts. Cohen said the Siberian snow got off to a quick start at the beginning of the month, but went into retreat during the weekend.

Expect more outlooks to be issued in the next few weeks, but use with caution. Seasonal forecasting remains a work in progress, if not frustration.

Last year the consensus called for quite a snowy winter, including forecasts by the local television meteorologists, WeatherBell, and Steve DiMartino at WeatherConcierge.com. The final total was 17 inches.

» READ MORE: Winter outlook sees cold February, above-normal snow.

What was the problem?

A little thing called chaos. It is worth remembering that nothing happens in a vacuum in the atmosphere, a 10-mile deep gas that misbehaves like a liquid attached to a sphere spinning at about 1,000 mph while hurtling through space.

Said DiMartino in his forecast post-mortem, “There is so much we still don’t understand.”