In the middle decades of the 20th century, Edwin D. Goldfield might have known more about his fellow Americans than anyone else in the nation.

A brilliant statistician with the U.S. Census Bureau from 1940 until his retirement in 1975, he helped make possible many of the information-gathering innovations that transformed the bureau from a counting agency into a sophisticated operation that today compiles and interprets huge amounts of general statistical information about how Americans live and work.

Not that he would ever have divulged what he knew. It's against the law to reveal census information that could identify a person, household or business. Goldfield not only adhered to the law, he was a true believer in privacy.

In 1950, before renovations temporarily closed the White House, Secret Service agents visited the Census Bureau. They told Goldfield they needed to collect information about people who lived near the house where they were planning to move President Harry S. Truman. It was a matter of national security, they insisted.

Can't do it, Goldfield told the Secret Service. The privacy of census respondents was paramount.

Goldfield, who died Sept. 27 of cardiovascular disease at age 87, was himself a private man. A lifelong bachelor, he lived alone in a modest apartment in a working-class neighborhood of Temple Hills. No one knew for at least a day that he had died until he failed to show up for a Cosmos Club luncheon he had arranged for longtime friends from the Census Bureau.

Goldfield, who was born in Brooklyn, N.Y., was brilliant from the beginning, said Claire Afromsky, his cousin. She recalled how he would read stories to another cousin who was three years older, even before the two were school-age.

Another cousin, Gloria Alpert, grew up in the same household. She remembers hearing him pace his room at night while he studied. He was valedictorian of James Madison High School in Brooklyn in 1935 and graduated from City College of New York four years later, majoring in statistics, mathematics and economics.

His nephew Mark Goldfield said his father told him there was a young woman in Ed Goldfield's life many years ago, but his uncle's parents objected to a marriage because she wasn't Jewish. Goldfield acceded to their wishes. Work became his life.

He became interested in a career in statistics while working as an intern in the office of New York Mayor Fiorello H. LaGuardia. Goldfield took a federal examination to qualify as a junior statistician and scored the second-highest grade in the country. He took the test a second time and scored even better.

In 1940, a faculty adviser at Columbia University, where he was doing graduate work in statistics, advised him to go to Washington and take a temporary job at the Census Bureau to gain practical experience. The temporary assignment evolved into a career in national statistics that lasted nearly a half-century. He later worked for the National Academy of Sciences, where he headed the Committee on National Statistics.

He arrived at the Census Bureau at an enormously innovative time. Among the changes he helped implement was probability sampling, first used the year Goldfield arrived.

Current Census Bureau Director C. Louis Kincannon, who met Goldfield in 1963, said Goldfield performed "an integrative function" over his long career. He was able to work across categories in the complex process of compiling statistical abstracts and other compendia of information.

"Ed was always asking imaginative questions," Kincannon said. "He could be a bit of a pest sometimes, but he was a sharp guy who went around and talked to people and stayed in touch with leaders of the profession. That gave him an advantage in knowledge and preparation."

He also took part in the changeover in the 1940s and 1950s from tabulating machines, first used in 1890, to punch-card machines invented by former Census Bureau employee -- and IBM founder -- Herman Hollerith. The punch-card machines evolved into computers. As program coordinator for the 1950 Census, Goldfield had at his disposal the pioneering computer known as UNIVAC 1.

For his personal use, he had pencil and paper, Mark Goldfield discovered last week. In going though his uncle's apartment, he found that Ed Goldfield did not own a home computer and had tabulated his taxes over the years by hand.

Growing up, Mark Goldfield would see his uncle occasionally. He remembered him as a quiet, somewhat formal man who always brought small gifts for him and his sister. Ed Goldfield quietly helped pay for their college educations.

Mark Goldfield also discovered that his uncle the statistician was "a very, very well-organized pack rat."

Neatly arranged in the small apartment were World War II ration books, tax returns dating to 1940, books and articles about statistics and grade-school drawings that his nephew and niece had sent him more than 40 years ago. He also had held on to a collection of bowling trophies he had won.

Goldfield believed the American people were getting their money's worth from the federal agency he helped shape. "I have the feeling that virtually everything that the Census Bureau produces has benefits that are greater than its cost," he said in a 1991 oral history.

For a frugal man, who until the day he died drove a 1981 Oldsmobile, that was worth noting.

Ed Goldfield was a man of paradoxes. At work, he had at his disposal pioneering technology. At home, he hand-tabulated his taxes and didn't own a computer.