Henri Lioret’s “Marvelous” Cylinders

In late 1888, Thomas Edison came out with a “talking doll”—a doll with a tin torso that housed a mechanism to play wax cylinder recordings. Unfortunately, Edison’s “Dollphones” (as they were sometimes called) suffered major mechanical problems and their fragile cylinders (made of a type of soap) tended not to survive repeated playing or the rough and tumble of children's handling. Some suspect that children may even have been frightened of the dolls.

A renowned 19th century French dollmaker, Emile Jumeau, had heard about the Edison dolls and thought he could try something similar. Aware of the supreme clockmaking skills of fellow Parisian, Henri Lioret—showcased, for example, when Lioret designed a wristwatch with an alarm that imitated crickets—as well as of the relevance of clockmaking to automated toys, Jumeau asked Lioret if he would be interested in designing a cylinder player to fit inside one of his Bébé Jumeau dolls.

Henri Lioret (1848-1938) agreed to the task, thereby becoming a major figure in the history of recorded sound. He came up with a playback mechanism distinct from Edison’s, and innovated as well by using molded celluloid (an early form of plastic) as opposed to Edison’s wax-like soap. The cylinders also featured a brass “hub” which added further support to the celluloid. Today, Lioret's use of celluloid is considered a major innovation in the recorded sound field, since Edison did not begin using it until 1912.

The talking Bébé Jumeau doll was for sale in time for Christmas 1893, with cylinders tailored to French, English, Spanish, and Russian markets. For the French market, there were 17 different 30-second-long cylinders from which to choose, each containing a different song. The product was apparently enough of a success to encourage Lioret to stick with the phonograph business for two more decades. In 1895, Lioret released a new phonograph, also designed for children called "le Merveilleux" (meaning "marvelous" or "wonderful" in English). In the Merveilleux, the cylinder playback mechanism fit inside a cardboard box that also functioned as storage for two cylinders. Advertisements for the Merveilleux described it as an “amusing and scientific toy.”

With the release of the Merveilleux, the catalog of 30-second cylinders expanded beyond the original 17 to include recordings of vocal duets and instrumental solos. According to Lioret’s color-coding system, song recordings were denoted with blue labels, while instrumental recordings were denoted with orange labels.

Cylinders for the dolls and the Merveilleux were mechanically interchangeable (and for Lioret, all fell into a category he referred to as “No. 1” cylinders)—though, given that some of the Merveilleux cylinders contained male and female voices singing in duet, it seems unlikely that they would have been used in one of the dolls.

Catalogs advertising available Lioret cylinders from 1896-1899 feature a juvenile section, where both the Jumeau and the Merveilleux machines were advertised. Along with descriptions of the doll and Merveilleux machines, were lists of available recordings.

Incidentally, by 1899, Lioret had devised an additional use for the Merveilleux mechanism, incorporating it into portable advertising displays for the Menier chocolate company.

 The Rodgers and Hammerstein Archives of Recorded Sound has in its collections seven of these 30-second “No. 1” cylinders. As of this writing, NYPL is the only institution to own these particular recordings and very few other No. 1-type Lioret cylinders can be found in Worldcat (a global database containing collections from 17,900 libraries). Those from our collection all have blue labels and titles that are listed in the “Chansons enfantes” sections of Lioret catalogs from 1896-1899. (As of this writing, no catalogs from before 1896 could be located). Since our cylinders are not labeled with “Bébé Jumeau Phonographe,” they were most likely intended to be played on a Merveilleux machine as opposed to a doll. (According to an article on the Association des Amis des Instruments et de la Musique Mécanique website, these cylinders can be identified because they have the words “Bébé Jumeau Phonographe” printed on the label).

We had digitized one of these cylinders back in 2012 to accompany an exhibit. Entitled “Il pleut bergère” (or, “it’s raining, shepherdess”); the cylinder contains a 30-second arrangement of an aria from a 1780 opera Laure et Pétrarque, by Fabre d'Églantine, later popularized as a song in favor of the French Revolution.

You can hear the recording in the Digital Collections here.

Primary source documentation of Lioret’s operation is difficult to find. We found some catalogs uploaded to personal websites by private collectors in France, in French periodicals available through HaithiTrust, and in a handful of newspaper advertisements found in NYPL databases. 

The rarity of the cylinders, their marked difference from our “run-of-the-mill” cylinders from Edison and Columbia Records, along with the scarcity of information about them, for us made them valuable primary source documents in and of themselves. Thus, we were especially eager to digitize the remaining six Liorets.

In March 2019, we had the opportunity to tackle this project in NYPL’s Audio and Moving Image Preservation Studios, where staff audio engineer Jeff Willens executed the transfers. Because it is not every day that someone transfers a Lioret cylinder, I asked Jeff to recall his experience for posterity.

Tell us about the machine that you used to digitize these cylinders.

Willens: The playback machine used for this project is also coincidentally French. The Archeophone was invented by Henri Chamoux in Lyon in the 1990s and is a more modern cylinder playback machine. It can be adapted to accommodate any size cylinder along with its various speed, warpage, height, or groove anomalies. Except for the original vintage machines, it is currently the only type of machine in the world that can physically play these Liorets. That said, it must be in perfect physical alignment, which can take even an experienced user upwards of half an hour to achieve. 

Was there anything special you had to do in order to make it work, besides having an Archeophone?

Willens: The No. 1 cylinders consist of a brass hub, with a celluloid coating for the grooves. On one side, the core has four spokes with a small center hole. For this side to stay firmly attached to the Archeophone, I took a small plastic hub from inside a VHS tape and used it as a clamp to prevent the cylinder from wobbling. The other side has a flat brass flange (also with a small center hole plus an outer label) that extends about ¼” beyond the edge of the celluloid all the way around. This posed a problem for playing the Lioret because the start of the groove was right against the side of the flange. Most phono cartridges are too big to allow the stylus to get that far over. Consequently, with a common cartridge, some of the recording couldn’t be played.

 In searching for the narrowest cartridge possible, I found not only the perfect one, but also the best-sounding stylus size for this project. The solution was an Ortofon OM78 cartridge with a 2.5 mil spherical stylus. 

To get the stylus as far over on the cylinder as possible, right up to the flange, the forked end of this stylus housing was snipped off on the right-hand side. Not only did this allow the capture of those beginning grooves the other cartridges couldn’t reach, but also the OM78 happened to have the perfect size needle to fit down into the groove and get some really good results.

What else makes these cylinders unusual? 

Willens: Usually, most early cylinders ran at a speed of 120 RPM. But Henri Lioret specified 100 RPM, or thereabout, which seemed to be the most natural-sounding speed and pitch for the voices on these cylinders. Had there been instruments on these recordings, I would have tuned their pitch to A435, the French concert standard at that time. Instead, I tried to find a normalized pitch close to 100 RPM (they were never exact) and tune the cylinders to each other, as the same performers appeared on all of them. In some cases, we had different performances of the same song, which made this process easier. 

Did you clean the cylinders beforehand?

Willens: A note included by Henri Lioret in each of the cylinders states that, for best results, the grooves should be lightly coated with “refined or olive oil to preserve them for a long time.” I consulted with other cylinder experts and chemists around the country, who agreed that olive oil is an unstable compound and rather acidic; it would actually degrade the celluloid. Therefore, oil should never be used on Liorets. A simple dry brushing was done to pre-clean them. 

For the audiophiles out there, did you do any digital manipulation to the files?

Willens: The Liorets were transferred flat, with no signal processing, into Wavelab, our Digital Audio Workstation, at 96 kHz/24-bit and archived as Preservation Master files. From there, they were phase-corrected for vertical-cut grooves, and digitally cleaned using both CEDAR and IZotope RX7. I attempted to retain as much of the sound on these recordings as possible and reduce unwanted noise (but not eliminate it).  Once approved, they were archived as Edit Master files.

What would you say overall about the experience?

Willens: Probably the most challenging thing about this project was the utter lack of information anywhere, including online, technical or otherwise, about Lioret cylinders. I literally had a single obscure book from France on Henri Lioret, which was my bible. [Editor's note: this book, entitled Henri Lioret, Clockmaker and Phonograph Pioneer, is available in the Rodgers & Hammerstein Archives of Recorded Sound as *LE 17-3797]. Everything else came from collaborating with truly outstanding colleagues from both NYPL and around the world, and my own experimentation. The surprising part of it all was discovering how well-recorded these were for 1895. They were nice and loud and much of it was remarkably clear and intelligible. And even a bit creepy, which is always fun. 

Cylinders can be fragile in many ways, and sometimes you only get one shot at doing a transfer right before a source is potentially destroyed. So this was a bit nerve-wracking too. It’s not for the faint of heart. But all these Liorets survived the process quite well. I think we did all right. 

Stay tuned in the coming months when these six additional cylinder recordings will be added to Digital Collections!