It’s History in the Playing

Say “player piano” and you are probably thinking of a pizza parlor and the tinny plinka-plinka of mechanical ragtime. Say “reproducing piano” and the response is probably, “What’s that?”

Yet the reproducing piano was the high-tech, high-fidelity wonder of early-20th century music recording. Electronic phonograph recording was in its infancy, limited in time and sonic dimensions, while reproducing piano mechanisms faithfully captured the styles of artists ranging from Victor Herbert and George Gershwin to Claude Debussy, Enrique Granados, Gustav Mahler and Richard Strauss.

As recording technology developed, sporadic efforts have been made to transfer this legacy from piano rolls to other media, with scant expressive success. Now a new series of CDs is mining this artistic trove seriously. The Pierian Recording Society-Pieria was the ancient Macedonian site where the Muses were worshiped-has released discs of Debussy and Granados playing their own music, and a double-CD set of the legendary Austro-American pianist Fannie Bloomfield Zeisler.

“It has been a labor of love, 43 years of dogged work and listening,” says Kenneth Caswell, the recently retired general manager of the Austin Symphony in Texas. It is thanks to his collection, and his persistence and perfectionism, that these historic performances are now available in astonishingly clean and detailed recordings. “I always loved music, and I started collecting old 78s for 5 cents and 10 cents each back in the 1950s. Then I got an old pump-operated player piano and then got interested in reproducing pianos.”

Caswell’s performances come from piano rolls made with the Welte-Mignon system, invented by Edwin Welte and Karl Bockisch in Freiburg, Germany, and first sold publicly in 1904. Many other systems followed, principally Duo-Art and Ampico in the United States, and were available either as separate cabinets that could be attached to the keyboard of a standard piano or as factory-installed units in instruments from hundreds of makers.

The market was huge and voracious. Mechanical pianos of various types frequently outsold conventional pianos, with more than 100,000 produced annually in this country alone for almost 10 years, until the stock market crash of 1929 and the following Depression virtually wiped out the industry.

The standard player piano used a single paper roll, punched to indicate only note and rhythm. The less-common reproducing systems used two rolls, with many more punches, including notches along the sides of the rolls, indicating accents, dynamics, tempo changes, pedaling and other nuances of phrasing and articulation.

Photographs of Welte’s studios show a genteel Edwardian salon with a piano-Steinway or Feurich by Welte’s preference. The artist would play the piano, with the mechanism automatically marking the paper roll. Each of the competing systems had its own way of measuring and reproducing dynamics and accents. However, until Ampico brought out a new system in 1929, only the Welte-Mignon system directlY recorded dynamics-varying degrees of loudness and softness-the others added dynamics later.

Some current efforts to reproduce this music are computer-aided, but Caswell prefers a naturalistic approach, as close to the original situation as possible. He records in his Austin home, using vintage Neumann microphones and a painstakingly restored, 1923 Feurich upright that has the string-length of a grand piano and is equipped with the Welte-Mignon system.

“I do it the old-fashioned way,” he says. “I do not do any computer stuff. I try to do it as Welte himself might have, using the same model piano Welte used to demonstrate his rolls.” Though elegantly simple in concept, this process has proved dauntingly complex in practice. Caswell needed 57 takes from seven copies of the roll to get Debussy’s “La plus que lente” just right. There are myriad subtle adjustments that have to be made to account for variances between the recording and the reproducing instruments and different issues of the rolls.

The core of Caswell’s personal piano-roll collection comes from the late Richard Simonton, a Toluca Lake collector. Simonton and his wife, Helena, went to Freiburg in 1948 and 1952 to make LP recordings with Welte and Bockisch, using Bockisch’s piano rolls. Issued on Columbia, these recordings revived international interest in the material, but were plagued by adverse postwar conditions, including damaged and poorly maintained equipment, fluctuating power and background noise.

“I learned so much from Dick Simonton,” says Caswell, 70. “He got all kinds of tips and tricks directly from Welte that were never in the factory manuals, and under his tutelage I got well ahead of the game. I became convinced that these performances could be reproduced accurately.”

The results are sonically vivid and interpretively mesmerizing, allowing listeners to hear Debussy and Granados play clearly with great personal distinction. Indeed, critic Bernard Holland found it almost too clear, preferring the low-fidelity pops and hisses of early electronic recordings precisely for the sense of technological distance they provide. “Mr. Caswell’s eerie cleanliness is beautiful,” Holland wrote in the New York Times, “but it tries to tell us that those 87 years [since Granados’ 1913 Welte recording session] never passed, that by unfreezing a long-ago present it becomes the present all over again.”

On the other hand, Martin Anderson wrote in the International Piano Quarterly that “anyone with half an interest in the piano will find these two issues [the Debussy and Granados discs] completely riveting-they are mandatory listening and they (re)make musical history.”

Many current ideas about how to play this music, or even exactly what this music really should be, may come under review because of the discs. Debussy, for example, confirms a famous discrepancy when he plays the half-note sections of “La cathedrale engloutie” at twice the published tempo. The variants Granados introduces in his music have been studied from Caswell’s material and incorporated as footnotes into a new edition of Granados’ piano scores in Barcelona from pianist/editor Alicia de Larrocha and scholar Douglas Riva (who contributed a booklet essay to the Pierian Granados disc).

Piano rolls are easily edited, which means that the authenticity of the performances they capture is suspect in some quarters. Caswell says that Welte never permitted post-session editing, however, except for correcting blatantly wrong notes. Caswell has used multiple editions of the rolls from collections around the world in his research and made comparisons with electronic recordings.

The Granados disc, for example, has the pianist playing his transcription of a Scarlatti sonata from Welte-Mignon rolls and from an Oden 78 disc recorded a few months later. The latest releases from Caswell’s collection on Pierian are the Zeisler discs, which feature large Chopin and Beethoven pieces and a collection of mostly Romantic miniatures. Forthcoming are Ravel and Scriabin discs, and a collection of performances by Teresa Carreno. The Pierian discs are distributed in the United States by Albany Music Distribution, and are widely available from the usual retail and Internet sources.

Considering that there are more than 8,000 listings in the Welte catalog alone, the reproducing piano seems to offer almost unlimited projects for enterprising musical archeologists with a technological bent.

“It takes a lot of work, about six months to a year for each,” Caswell says in an understatement. “I think I’ll need another 20 years to get through the best stuff.”