By Stephen Budiansky

Stephen Budiansky is a correspondent for The Atlantic Monthly, where the full version of this story first appeared in March 2000. It is reprinted with permission.

Paul Machlin's volume of 17 transcriptions of Fats Waller's organ, piano, and vocal performances (Thomas Wright "Fats" Waller: Performances in Transcription 1927-1943), published as part of the American Musicological Society's Music of the United States of America (MUSA) series, may go a long way toward establishing Waller as an important, even great, American composer.

A few of Waller's more popular piano pieces were published as sheet music during his lifetime--but, Machlin says, these are "deeply simplified" versions cranked out by "some Tin Pan Alley hack" who listened to Waller's playing and came up with at best a rough approximation. The recordings of Waller's performances are thus the only authentic source for producing a written score.

Transcribing these musical sounds onto paper decades after they were recorded was a surprisingly difficult and exacting task. Machlin was motivated to try it, he says, in part because he was frustrated at how often Waller's work has been misunderstood by critics and historians who simply do not appreciate his formidable keyboard technique and his inventive genius. "You only really understand that when you write it down and look at it," Machlin says.

Like many pianists, Machlin realized that Waller's keyboard works were something exceptional. He was dismayed by critics who said that Waller had "wasted his talent" or wasn't serious about his music. "There's a particular kind of white jazz historian who sees jazz as an expression of oppressed people," Machlin says. "And so when they see a commercially successful African-American, it somehow 'lessens the authenticity.'"

Computer programs exist that can automatically transcribe musical sound into musical notation, but Machlin found they were of little use with Waller's music, mainly because they are not sophisticated enough to capture his rhythmic subtleties. So Machlin works the old-fashioned way. First he listens to a song until he has a basic picture firmly in his mind of its harmonic shape and of what the hands are doing in each phrase. Then he listens to one bar--or sometimes half a bar--at a time, writes down what he thinks he hears, and then tries it out on the keyboard to see if it sounds right. Then he goes back and fills in what he's missed. Sometimes he plays a tape at half speed to try to pick out what Waller was doing. Next comes an extensive process of editing and checking, and seeking the opinions of others. "You have someone look over it and they question just about everything," he says. The low fidelity of many of the early recordings doesn't help. Because musical sounds contain natural overtones, it can be extremely difficult to tell whether all or only some of the notes of a chord were actually played. In a few passages, where "despite a lot of listening and agony" Machlin is still not absolutely sure, he has marked in brackets on the transcriptions what he thinks Waller was doing. It can take a full day's work to produce just a starting draft of twenty-four bars; transcribing a first draft of one complete piece can take weeks.

Machlin says he hopes that pianists and organists will play the pieces, but he admits that some funny questions do arise. Dan Morgenstern, the director of the Rutgers Institute of Jazz Studies, has observed that Waller must have played a piece like "Honeysuckle Rose" every working day of his life, and he never played it the same way twice. "The recording becomes a frozen performance," Machlin says, "a benchmark--willy-nilly, in spite of itself." But it's all we have to go on, and "whatever else you may speculate about, you know that that happened."

"Some people say to me, 'Why should someone play these pieces when Fats Waller has already done it?'" Machlin says. "My answer is 'For the same reason you'd play a Chopin étude when Chopin has already done it.'"