Sagarin realized that contest officials might have kept spring ice breakup records for the past 85 years, down to the exact minute. Their records, he recognized with excitement, could help reveal whether global warming has affected the Arctic by showing if spring has been coming earlier. As soon as Sagarin returned home, he called the Nenana Ice Classic contest headquarters. Officials there gladly mailed him copies of their records–and the documents were everything he had hoped for.

Sagarin's use of obscure historical records to answer a current research question is a prime example of a little-known science called phenology. Long a neglected backwater, phenology is the study of recurring natural events such as flowering, breeding, and migration–or, in Sagarin's case, spring ice breakup. Relying upon old diaries or logs that tracked seasonally repeating phenomena has its own peculiar strengths and weaknesses. But today, such records are taking center stage as a surprisingly powerful tool in the study of climate change. For all their high-tech satellite studies of the planet, scientists still need the recollections of the long dead to understand global warming. Such insights are critical because global warming could disrupt weather patterns and ecosystems across the planet.

The Arctic especially fascinates global warming phenology researchers, because that's where the world's largest temperature increases have occurred in recent decades. Studies indicate that the Arctic has just gone through its warmest century in 400 years, with plant activity in the far north jumping 11 percent during the final decades of the twentieth century. One climatology researcher predicts that global temperature increases in this current century will be double that of the last, and the Arctic will be hardest hit.

IN THE WORLD OF PHENOLOGY, discovering historical data to study requires more than a modicum of serendipity. After all, how do you know where to look to find records in the first place? Recently for instance, John Magnuson, a lake ecology researcher at the University of Wisconsin in Madison, learned of a document listing 100 years of freeze and thaw dates for a lake in Maine. The record was hanging in a restaurant foyer, next to a board filled with business cards. "It's like treasure hunting," Magnuson explains. "Sometimes you find the records by accident."

That was certainly the case with Sagarin and the Nenana Ice Classic. The beauty of the annual contest is that it relies on the gambling compulsions of more than 100,000 Alaskans. To keep the contest fair and accurate, the rules haven't changed in 83 years. The yearly tradition is so popular that it was written into state law in 1959.

The Ice Classic traces its beginnings to a group of engineers who overwintered in 1917 in the town of Nenana, 55 miles southwest of Fairbanks. They were waiting to build a railroad bridge across the Tanana River. Until the river melted, they couldn't finish. Pooling their money, $800 in all, they placed bets on when the river's three-foot layer of ice would break under the pressure of upstream waters. Being engineers with too much spare time on their hands, they built a wood contraption that was cabled to an onshore clock to mark exactly when the ice broke up. Though they called a 'tripod,' the 30-foot device actually has nine legs rooted in the river's ice.

The spring thaw officially arrives when the black-and-white-striped tripod collapses through the ice or drifts far enough to move its cable a hundred feet, yanking the onshore clock to a stop. Sometimes, the tripod slowly sinks in rotting ice; other times upstream debris knocks it down. Either way, the lucky winner is the one who guesses the day, hour, and minute when the clock halts. Then Nenana waits another year to cut another tripod from the woods. "Nobody has no idea when the breakup is gonna come," says Perci Dike, a Nenana local. "If I did, if I had some idea, I would have win the dang thing there years ago."

As a safeguard against cheating, a 24-hour watch is stationed at the river during prime ice break days, usually April 25 to May 10. Along with that precaution, the tripod design and contest rules all make for a precise scientific experiment, says Sagarin. In fact, the Ice Classic has provided some of the most trustworthy data available yet in the field of phenology.

Every phenological experiment starts with a scientist trusting the stories of people who may not be the most accurate observers of nature, and that raises some potential problems. How does the researcher know that an observer's records are true? Take the example of an Arctic island explorer named Joseph Dewey Soper, who recorded when he sighted caribou throughout the year for the Canadian government. According to Soper's journal, he didn't see any caribou for seven months in 1931. But his documents neglected to mention that he couldn't travel during that time because of a knee injury from slipping on sea ice, Canadian researchers learned.

Another difficulty is how recordkeepers define when a noteworthy event has occurred. For example, a tree could be said to have new leaves when there are visible buds, or when the first leaf is fully grown. New observers who take over a recordkeeping tradition may do things differently if the rules aren't clearly defined when the old observers die.

By contrast, the records from the Ice Classic sounded too good to miss out on. Within two hours of receiving the documents, Sagarin plotted a graph with the ice breakup data--and found he had hit the jackpot. The results showed that spring melts in Nenana today come on average five and a half days earlier than in 1917.

Sagarin wondered whether other data existed that could back up his results. He attempted to find snowfall, rainfall, or air temperature records from weather stations in Fairbanks and Nenana, but ran into the sorts of problems that kill many phenological experiments. Nenana had not taken any measurements during many of the last 85 years. Records from Fairbanks, on the other hand, were not very helpful, in part because the station had moved once. For Sagarin's desired level of precision, snow or rainfall records taken in more than one location were useless.

And since Fairbanks had grown considerably over the years, the records were doubly damned. City growth creates a 'heat island,' where buildings, pavement, and cars raise local temperatures. Temperatures in town didn't reflect nature. At any rate, good temperature records–taken with the same thermometer, at exactly the same time of day, every single day–are rare to find.

Another type of seasonal data that researchers can sometimes track is the sudden greenery of spring, a global phenomenon they call the "green wave." (A classic example is the "cherry blossom front" that moves from south to north in Japan.) Investigators use satellite images to register spring's march across the globe–two swaths of green burning their way toward the Arctic and Antarctica. But since accurate images date back only 20 to 30 years, they’re of limited help in following longterm changes. That’s why scientists have fallen back on phenology records of all kinds.

On the ground, old journals recording the first buds of spring can’t offer clearcut proof of climate warming either, though. The timing of plant buddings depends on more than just ambient temperature. Buds "count" warm days and then burst open, but a mild winter can confuse them and delay their opening.

All in all, says Sagarin, compared to the other methods, a more accurate way of measuring longterm temperature change is to study the freezes and melts of lakes and rivers. It's much easier to trust someone to write down when the ice melted than to track temperatures, he says. Magnuson, the lake ecologist, agrees. "In many places in the world, we have lake and river records going back 100 to 150 years," he says. "But a single record from a single point does not convince one that the world is getting warmer. It is critical to have long-term records from around the globe." Magnuson has compiled 39 different records that list ice breakups and melts in the northern hemisphere, from Asia to Wisconsin. His studies, like Sagarin's, show that spring ice melts are occurring sooner and winter freezes are coming later.

Ironically, the researchers note, global warming may destroy some phenological sources of data, as lakes and rivers at lower latitudes fail to freeze in warmer years. For instance, one of the longest running sets of historical records has tracked the freeze dates of Lake Constance, in Europe, for 1200 years. The logs belong to two churches separated by the lake; one sits in Germany and the other lies just across the border in Switzerland. Their chapels share a single Madonna statue. When the lake freezes, whichever chapel has the Madonna carries it across to the other. Unfortunately, the lake doesn't freeze every year. In some centuries, the lake didn't freeze at all. During warm periods--which experts expect to see more of--the record can't give enough information to measure climate change.

Scientists focusing on long-range bird migration patterns will face a similar problem as global warming speeds the arrival of spring. In the Netherlands, researchers have been tracking climate change by applying phenology to 20 years of records on the egg-laying habits of the pied flycatcher. Normally, at a set time before the start of spring, the birds fly north from dry tropical forests in Africa to Europe. They lay their eggs to coincide with the blooming of spring foliage that serves up a peak of insect abundance.

But with spring arriving sooner and sooner over the last 20 years, the timing of this bug feast has been moved up as well. As a result, the flycatchers have been laying their eggs on earlier and earlier dates. However, they're can't adapt forever. Many already are migrating to Europe too late to take advantage of the insect feast. Eventually, the birds may reach a point beyond which they can't lay eggs any earlier in the year, or possibly even die out; researchers have already begun to detect a decrease in the numbers of nestlings. Like a lake refusing to freeze, the spring breeding of flycatchers could become an imprecise record of climate change.

PIED FLYCATCHERS AREN'T THE ONLY ONES having trouble with timing. Just about when Sagarin was going to publish his findings, he realized a major problem with his calendar was skewing the results. Wall calendars pretend that years are 365 days long, but the true solar year (the time it takes for the earth to go around the sun) is actually about a quarter of a day longer. Every fourth year, we add an extra leap day in February in sloppy compensation. Even so, by the end of a century, the calendar can still be off by almost a whole day.

By Sagarin's reckoning, he is the first researcher in phenology to notice this problem with records using years that start on New Year’s Day. He proposes a solution: To reduce the discrepancy, researchers could start their year at the vernal equinox, the first day of spring when day and night are the same length.

The bias, however small, is real. Other researchers in the phenology world are taking Sagarin's criticism seriously. But not everyone agrees that the vernal equinox method is the best solution. Magnuson, for one, is considering whether to reorganize his own research using the winter solstice, the moment when the sun is farthest from the equator. Beyond accounting for leap years and historical calendar shifts, he hasn't applied any calendar corrections to his results. But in the worst case, his results are probably only a day off from Sagarin's.

At any rate, Magnuson likens the impact of Sagarin's suggestion to a group of people timing a race. Even if everyone's stopwatches aren't set to the right time, they can still tell who crossed the finish line first. Regardless of the exact timing of dates, the results from phenology studies still point to a strong global warming trend.

In the end, the experts say, phenology might well be one of the best ways to truly nail down the case that global warming is real. Handscrawled journals and logs from the past have provided a wealth of evidence to choose from, whether they come from priests, naturalists, or gamblers. So far, ice breakup records like the one Sagarin stumbled upon are providing some of the clearest proof yet of worldwide climate change. And as the years go on, the Nenana Ice Classic could show the fallout of global warming even more dramatically. That is, if the river is still freezing a century from now. Just don't bet on it.