"It's not the things we don't know that causes the trouble, it's the things we know that ain't so." - Artemus Ward.

In 1928 Louis Dublin, who later became Chief Actuary for Metropolitan Life, used mortality tables to estimate how low the death rates for humans in different age groups could possibly go. From this, and as he put it, "without intervention of radical innovations or fantastic evolutionary change in our physiological make-up," he concluded that the maximum possible life expectancy of humans for both males and females was 64.75 years. At the time, the life expectancy in the US was about 57 years, so his conclusion was not ridiculous.

Unfortunately, Louis Dublin did not have access to mortality data from all countries. Given such data, he might have shown hesitation in publishing his results. A survey in New Zealand, performed seven years earlier, indicated that women in the non-Maori population there already enjoyed a life expectancy of 65.93 years, beyond what Dublin had calculated at an absolute maximum.

In 1936, with the New Zealand data in hand, Dublin tried again. This time he and fellow-worker Alfred Lotka came up with a life-expectancy upper limit of 69.93 years. Dublin and Lotka were highly intelligent men, two of the creators of modern mathematical demography. However, their prediction of an ultimate life-expectancy lasted only five years. Icelandic women surpassed the calculated maximum in 1941, with an actual life expectancy of 70.30 years.

That was more than sixty years ago. It is tempting to say that we are wiser now, and no one today would repeat Dublin and Lotka's mistake. Right?

Not quite. In 1990, and again in 2001, studies published in Science magazine assessed the life spans that might be achieved if we do not make "implausible" and "overly optimistic" assumptions. The results showed that at age 50 life expectancy would not exceed 35 more years, "unless major breakthroughs occur in controlling the fundamental rate of aging." There has been no such major breakthrough, but Japanese women as a whole already enjoy a life expectancy of close to 85 years.

What is it that has consistently led researchers into human longevity to underestimate human staying power? The most plausible answer I can offer is that people have been too swayed by their prior convictions. Rather than believing the data, we say, surely we can't live that long, and modify our estimates.

What the data say, with astonishing consistency and in records that go all the way back to 1840, is that the life expectancy in the women of the longest-lived nation at any particular time goes up by three months each year. Thus, in 1840 the women of Sweden lived an average of 45.71 years, longer than anywhere else. By 1900, the women of New Zealand were the record holders, with a life expectancy of 59.95 years. By 1960 the place to be was Norway, where female life expectancy was 75.83. In 2000, the women of Japan held the record, with a life expectancy of 84.62 years. And it is not necessary to use large intervals of forty or sixty years. If you plot all the data, year by year, the trend is always very close to a straight line. If you are then willing to extrapolate, you will conclude that by 2060, the average life expectancy of women in the longest-lived nation will hit the century mark. Also, if the future in any way resembles the past, that longest-lived nation will not be the United States. Not once, in 160 years of records, have the women (or men) of the United States topped the longevity charts.

And what about men? For whatever reasons - smoking, stress, life-style - human males have not in the past lived as long as females. In 1840, Danish men lived longest, at 43.11 years, but they died on average more than two-and-a-half years younger than Swedish women. Male longevity has gone up, but the difference between men and women has increased over time. Today, the males of the longest-lived nations (Iceland and Japan) live close to 78 years, but this is nearly seven years less than their women-folk. Again, if we are willing to extrapolate from history, by 2060 the life expectancy of men in the longest-lived nation will be 91 years.

You may say, so what? Is it important that forecasts of life expectancy be accurate?

It is, if you want to impose standards of honesty on our public officials. Increases in life expectancy imply that a larger fraction of the population will consist of the old and the very old. This will affect everything from health care to pension benefits, and only a government with its head in the sand can ignore a nation's longevity trends.

Am I being cynical, if I suggest that our government may well prefer to believe that life expectancy will not go much beyond its present value? Otherwise, it will be necessary to make radical changes to our social security system and to our health care system. When in 1890 the world's first national pension system was introduced, in Germany by Otto von Bismarck, benefits went only to those over seventy - of which there was a negligible number. At the time the men of the longest-lived nation (New Zealand) had a life expectancy of 54 years, women of 57 years.

There is another factor to be considered here, one that ultimately may imply social changes more profound than those required by increases in longevity. The references that I have available to me tell how long people lived in, say, 1900, but nothing about their physical condition. A couple of hundred years ago, an "old" person was anyone over fifty. Today I see at my local health club energetic and active eighty-year-olds. People in the developed nations are not merely living longer, they are living in much better health. If energy and well-being go hand in hand with longevity, something which today's medicine seems to promise, then a surfeit of centenarians is nothing to fear.

Maybe the way of the future will be to lower the age of "first retirement" to forty-five, send everyone back to school for a few years, and allow people after completion of a second career to move to final retirement at age eighty-five or ninety.

A crazy idea? Perhaps. But less crazy than you'd have been judged had you suggested, back in 1840, that people might one day live for an average of eighty-five years.