Bernard Roux

This page

This page presents a preliminary account of the audio activities of the French engineer and industrialist Bernard Roux (1875-1958).

It is part of the site Classical 'Society' Records by Nick Morgan.

Documented from 1905 until the late 1930s, Bernard Roux’s audio products and services were an integral part of his overall business – making and selling early plastics, notably insulators and moulded components.

A little-known pioneer of audio technology in France, Roux’s firm made private and commercial recordings for a range of clients and also produced a small number of innovative electrical audio devices.

For dates of creation and latest update, please see Page information.

NB this page is still under construction

Genesis

This page grew out of an attempt to investigate a single recording made in 1923 by Bernard Roux's firm. Initially, I planned to cover it in a simple (if lengthy) blog post, which I started researching and writing in early December 2022. I soon realised that Bernard Roux's firm was not, as I had imagined, a small business which recorded discs to order for private clients, but an industrial manufacturing concern supplying a wide range of companies with products and services including audio recordings and equipment. On 29 December 2022 I abandoned the blog post and started work on this page. The creation date shown in Page information is the date when this page, drafted as a virtual web-page on my computer, was first uploaded to the internet.

This page has been written and rewritten many times. Starting as an exploration of what seemed to be just a few known recordings made by Roux's studio, it has grown to encompass several businesses with several activities and many different products, and to cover a much longer time-span and many, many more records than anticipated. There have been delays, setbacks and occasional lapses in motivation. The page's sections were created, worked on, and rewritten at different times, resulting in a less consistent tone and presentation than I would like. And in the spring of 2024 ,just as the page seemed close to being finished, I decided to compile a spreadsheet of known and suspected Bernard Roux matrices. A fascinating and satisfying exercise, this meant yet more rewriting, and also rekindled a long-standing frustration with the problem of presenting detailed discographical information online. The ideal platform for combined record-company history and discography has yet to be devised, it seems to me (Mediawiki, as implemented by me, is not it).

Sources

I carried out research for this page in the UK, mostly at home, occasionally in the British Library at St. Pancras.

The Roux disc which prompted this project is one of six held by the Bibliothèque nationale de France (BnF) in Paris, of which four have been digitized and made accessible via its digital portal Gallica. I soon found a small number of examples held in insitutional collections elsewhere. Some years earlier, I had bought one Bernard Roux test pressing myself, and I gradually discovered that other, commercially issued discs in my collection were also produced by Roux. While working on this page, I have bought a few more Roux tests and commercial productions: I have had them all digitized and have uploaded the transfers to the US non-profit repository Internet Archive.

The labels on most known Roux test pressings give away little or nothing about their contents or intended clients. Fortunately, other sources have proved rich in information about Roux's businesses and achievements. By far the richest are the historical newspapers, periodicals, directories, official bulletins and other publications (mostly in French) which have been digitized and made freely accessible, in full text, via institutional websites, notably the BnF’s Gallica, its commercial partner RetroNews and the Internet Archive. One vexing and perplexing omission from the BnF’s print digitization programme (to date) is France's important heritage of historical audio and record magazines (Musique et radio, Machines parlantes et radio, Phono-radio-musique, L’édition musicale vivante, Le Guide Bleu des Nouveaux Disques etc.); I hope we will be able to read these one day, before it's too late. Historical patents can be freely retrieved from online databases. Some genealogical information about French citizens is accessible in France's online municipal archives, although to find it one needs to know where and when an individual was born, which often means subscribing to a commercial genealogical website such as Ancestry or Geneanet.

Discographical sources consulted for this page include the BnF’s online catalogue, internet sales listings, and collectors’ and enthusiasts’ websites. In the spring of 2023, while searching for information on small French record labels, I discovered Jerome Moncada’s website Label Gallery Française, an ongoing survey which promises to be exhaustive and indispensable. Where possible, all these and other sources are cited below (see also Acknowledgements).

I have found no reference anywhere to surviving personal papers of Bernard Roux or business papers of his companies. Extensive and informative as the above sources are, a full, detailed account of Roux's life, career and legacy will probably require travel to France and consultation of sources accessible only off line.

Translations

All translations into English are mine, except where noted. I apologise for writing in English about a French subject, drawing mainly on French sources. Fortunately, after running another page from my site through a well-known online machine translator, I can reassure anyone who doesn't read English that it produces pretty satisfactory results.

Tables

Mediawiki has many good facilities and features but its tables leave something to be desired, especially in the hands of an ageing novice. I have tried to make the tables on this page more readable and consistent than comprehensive (there is always too much discographical information for any table) or easy on the eye, for which many apologies.

Disambiguation

Bernard Roux is a common French name. The person this page is about had contemporary namesakes active in various industries and professions. For instance, another Bernard Roux was an employee of the transnational Liebig concern and, reportedly, the inventor of Viandox, a long-lived brand of meat extract launched some time after World War I by Liebig's French subsidiary. The same or perhaps a different person co-founded an animal fat-rendering company.

The subject of this page traded on two 'Roux processes' (procédés Roux); again, there was at least one other contemporary 'Roux process', although it was not exploited commercially until slightly later. Also known as the procédé Roux-Color (and variations thereof), this was a colour film-projection process, invented by two brothers not related to Bernard Roux.

Metonymy

Throughout this page, the name (Bernard) Roux is used to denote both the man and his various companies, when using the latter's full, official names would seem verbose and redundant. So, if I write that a work or artist was 'recorded by Roux', I do not literally mean that Bernard Roux himself operated the recording equipment – I very much doubt he ever did – or supervised a recording session but, as I hope the context will make clear, that one or more recordists employed by him did so, or that his company as a whole executed the recording programme or contract under discussion, and so on.

Life

Bernard Roux was born on 30 June 1875 in his maternal grandfather's home in Fontaine-lès-Dijon, a suburb of Dijon, capital of the eastern département of Côte d'Or in Burgundy. Bernard's father Alexandre Roux was a notary in Dijon, where he lived with his wife Marie-Bernarde, née Mugnier. The Roux family was affluent and also owned a country property 35 km outside Dijon, midway between the communes of Trouhans and Les Maillys. Bernard had at least one older brother, Marcel, who died in 1918. Roux's father, meanwhile, had died in 1913.

At an unknown date Bernard Roux moved to Paris, probably for his education. He remained there for the rest of his life. For some years, Roux listed his father's country residence as a second address in Paris directories; this country property was sold during the 1920s.

No record of military service by Bernard Roux during World War I has come to light. Aged 39 in August 1914, he was perhaps too old to be called up, or else he engaged in industrial production deemed of national importance.

On 30 August 1924, in Paris's 14th arrondissement, where he had lived for several years, Bernard Roux married Marguerite Françoise Andrée Eléonore Baudevin, born in Saint-Etienne in 1897 and also resident in Paris. The couple lived together in Paris for the rest of their lives, and had three children: Marie (born 1927?), Marcel (born 17 February 1929) and Charles-Bernard (born 2 December 1930).

Roux's wife Marguerite predeceased him, on 21 October 1955, at the Salpêtrière hospital in Paris.

Bernard Roux died on 7 May 1958 at the family's Paris home.

Education

Bernard Roux received his primary education in Dijon, where he attended the Catholic École Saint-François de Sales from 1887 to 1892.

In October 1892 Roux was enrolled into Paris's prestigious Collège Stanislas, a private Catholic secondary school.

Roux commenced his higher education by enrolling in 1897 as a physical engineer at the École Municipale de Physique et Chimie Industrielles, or EMPCI, in Paris (now the École Supérieure de Physique et de Chimie Industrielles de la Ville de Paris, or ESPCI). Details of his specialisation(s), graduation and qualification(s) are yet to be obtained.

It is not clear whether Roux later obtained a further higher qualification. From 1919 onwards, Paris trade directories list him as 'ingénieur E.P.C.' This usually denoted a graduate of the state engineering academy, the École nationale des ponts et chaussées (now the École des ponts ParisTech); but the École was closed for the duration and in any case Roux was of the wrong age to enrol. The acronym presumably referred to his training at the EMPCI.

Early career

Patents

Soon or perhaps immediately after graduating from the EMPCI, Roux entered the industry in which he would remain active for the rest of his working life: the manufacture of rubber derivatives and their application in a range of products. This career was built on the exploitation of more than forty patents, granted over as many years. Approximately two thirds were granted in France, the remainder elsewhere. The first few patents relate to insulating plastics derived from rubber and techniques for producing them; these were soon followed by others relating to gramophone records and electronics. Roux's own patents, and others naming him (as applicant or assignee), can be found and accessed in the online databases of France's National Institute of Industrial Property, the European Patent Office and the World Intellectual Property Organization. Where relevant, patents are cited below, but they are not described or examined in detail on this page.

Ebonite and Ebonitine

In 1902 Roux was granted his first known patent, in Denmark, for a 'Procedure for the production of an insulating material'; he perhaps devised this method while still at the EMPCI. Its Danish title is an almost exact equivalent of that of a French patent which he would be granted five years later.

1902 was also the year in which Victor Karavodine, a Russian émigré resident in Paris, was granted a patent in France for a method of manufacture of a new material, derived from vulcanized rubber; its name, Ebonitine, clearly echoed that of the older, well-established hard rubber, Ebonite. (The name Ebonitine was not itself new: a British company had marketed a very similarly-named substance in the 1860s, and in the mid-1890s another acted as the British agent for France's Société Industrielle des Téléphones, among whose products were both Ebonite and Ebonitine.)

Karavodine also had an interest in Ebonite, obtaining a patent for its 'regeneration' in 1904, which he assigned in 1905 to Bernard Roux; Roux later obtained a patent of addition to it. Also in 1904, Roux applied on Karavodine's behalf for a patent in the USA on the 'Treatment and Utilization of Waste Vulcanized Rubber and Ebonite', granted in 1906 and likewise assigned to Roux. Another patent transfer by Karavodine to Roux had been made in Italy in 1902. The two men clearly had some kind of partnership, although there is no clear trace of it outside these patents. And although Roux would build his business on making and selling Ebonitine, and always advertised it as produced by the 'Roux process' or 'processes' (procédé/s Roux), no report has yet been found of a transfer by Karavodine to Roux of the patent on Ebonitine.

Ebonite was already used in the manufacture of many products – among them, audio cylinders, discs and player parts; Ebonitine, meanwhile, was patented specifically as an insulator, but soon also attracted interest from the gramophone trade. In 1908 the marque 'Ultima' was registered in Paris for discs made of 'ebonitine or any other material'. The same year, Louis Boduin of Amiens registered Ebonitine as trademark of a product 'intended to lubricate talking machine discs', and by 1910 Boduin was advertising Ebonitine as a preservative which, he claimed, protected discs against wear, giving them an 'unlimited lifespan' and 'reviving worn records'. In 1909 another entrepreneur, based in Auxerre in Burgundy, had also advertised Ebonitine for the same purpose. Across the Atlantic, a US textbook described Ebonitine as 'A hard rubber substitute used for phonograph records.' In 1911, 'Ultima' was again registered in Paris as a marque of ebonitine discs.

1907

By 1907 Roux was involved in a business apparently named Ateliers « Ebonitine », which manufactured and traded in this and other 'moulded insulating components' (pièces isolantes moulées). The constitution of this business and its ownership are still unclear: no notices or reports of its formation have yet been located, despite the fact that making such details public was a legal requirement.

The first trade directories to list Ateliers « Ebonitine » associate it unambiguously with Roux, who clearly had a leading interest in it, and was presumably exploiting his and Karavodine's patents, assigned to him. The business's main address was 144, rue du Chemin-Vert, in Paris's 11th arrondissement, not far from the western corner of the Père Lachaise cemetery. In addition, it maintained workshops (ateliers) and an 'industrial testing laboratory' (laboratoire d’essais industriels) approximately one hour's walk away, at numbers 2 and 4, rue Esquirol, in the 13th arrondissement, near the Salpêtrière hospital. As far as can be ascertained from modern photographs published online, the premises in the rue du Chemin-Vert were located in an ordinary Parisian residential apartment block, possibly in a courtyard. The rue Esquirol was more obviously light-industrial: the area contained many workshops, while a notorious nearby haunt of rag-pickers had only recently been demolished. As Roux's businesses evolved over the following thirty years, they remained at these locations (with variations) for his entire career.

Trade directories afford detailed snapshots of Roux's activities during these three decades. In the earliest relevant directory, for 1907, 'Roux process' Ebonitine appears under the following headings:

Bois durci (a mid-19th century plastic, for which Ebonitine presumably advertised as a substitute)
Ebonite (Ebonitine advertised as a substitute)
Electricité (Ebonitine advertised for 'moulded insulating components')
Isolants et isolateurs (Ebonitine advertised for insulators)
Ivorine (another mid-19th century plastic, for which, again, Ebonitine presumably advertised as a substitute, in 'moulded components for electrical and photographic equipment, knife-handles, ink-wells etc., black and in colours; gramophone discs a speciality')
Matières plastiques ('moulded insulating components')
Phonographes (two entries for 'Ebonitine diaphragms', plus another stating, 'phonograph and gramophone pressing a speciality. Metal matrices, discs, diaphragms.')
Sonneries et signaux électriques, télégraphiques, à air, etc. ('moulded insulating components' for bells, electric signalling systems etc.)
Télégraphie ('moulded insulating components')
Téléphones ('moulded insulating components')
Traction électrique ('moulded insulating components')

Of course, the above listings are not proof that Ebonitine was used for any of the advertised purposes. Nor does the directory reveal anything about the size and health of Roux's business in 1907. Still, the Ebonitine works and its successors would continue to appear in all directories consulted for this page until 1938, suggesting a demand had been identified and was being supplied by Roux.

1908-1913

The 1908 edition of the same Paris directory again listed the Ebonitine business at the same address, but its range of products was now expanded to include Ebonite, Amiantine (an asbestos product) and manufactured rubber.

No significant changes are observed in the editions of 1909 to 1911, except that in the latter the main site had moved across the street from 144 to 147 bis, rue du Chemin-Vert, and Ebonitine was no longer advertised for use in telegraphy or telephones.

In a 1911 directory of the cinema and photography industries, Bernard Roux was listed under headings for batteries and asbestos, offering moulded insulating components; but it seems that this attempt to break into the sector did not bear fruit.

The general Paris directory for 1913 carried new entries from Roux under seven additional headings, and revised entries under existing ones:

Amiante (revised: 'Ebonitine, Amiantine, Ebonite and rubber[.] All asbestos items stocked')
Appareillage électrique (new: 'moulded insulators for [electrical] equipment. Melt-proof components […]')
Calorifuges (new: 'Heat- and fire-resistant Amiantine conduits and coatings. Sold in sheets or mouldings of all shapes')
Chauffage (appareils de) (new: 'insulating, fire-proof control wheels and levers for all heating appliances, radiators, etc., a speciality')
Chauffage à vapeur et à eau chaude (new: as Chauffage)
Cylindres pour phonographes (new: 'Moulds and wax blanks')
Disques pour phonographes (new: 'Specialized workshops for custom pressing of recordings. Supply and installation of all accessories')
Ebonite (revised: 'Range of Ebonite in sheets, bars and tubes in stock. Moulded and machined components. Ebonitine, Amiantine and rubber')
Isolants et isolateurs (revised: 'Specialist manufacture of insulating sheets and bars for all applications. Moulded components for transmission and equipment in Ebonitine, Amiantine and rubber')
Ivorine (revised: 'Manufacture of all Ivorine components using B. Roux process. Moulded components for electricity, telephony, photography and all applications. Gramophone records a speciality')
Radiateurs (new: 'insulating, fire-proof radiator caps a speciality. Supplier of major brands, French and foreign')
Traction électrique (revised: 'moulded plastic or fire-proof components for overhead or underground lines, equipment, power plants, etc.')

Société franco-anglaise de caoutchouc manufacturé

In December 1913, a new rubber manufacturing company was formed in Paris, the Société franco-anglaise de caoutchouc manufacturé. The founder was an engineer turned entrepreneur, Jean Nouzaret (1866-1929). Roux was not named as an officer of the company but clearly had an interest in it. Among the company's stated assets were Karavodine's patent, assigned to Roux, for the regeneration of vulcanized rubber and Ebonite, with its patent of addition by Roux, and Roux's own patent for the manufacture of rubber tubes and rolls or bars (boudins), also with patent of addition. Another asset, reflected in the company's name, was the benefit of agreements with a British company, the Millwall Rubber Co., Ltd., of Harpenden, which had been exploiting the Roux rubber patent (also granted in Britain) for some time, although the recent early death of its owner had cast doubt over the Millwall Rubber Co.'s future. As manufacturing premises, the new Société franco-anglaise was to take over a rubber works in Levallois-Perret, a north-eastern suburb of Paris. (This factory had recently burned down.) Presumably, Bernard Roux's interest in the new firm was mainly financial; his name continued to appear in trade directories only alongside the Ebonitine business.

1914-1919 and after

Roux's entries in the 1914 Paris directory showed no changes from those of 1913.

No information regarding Roux's activities during World War I has been uncovered in sources accessible online. Whether the Ebonitine workshops and Levallois factory were commandeered or re-purposed to serve the war effort is not known.

In August 1918, some months before the end of hostilities, Roux bought additional manufacturing premises in Paris, on the Avenue du Maine. The new site was presumably handy for Roux, being located only five minutes' walk from his flat on the Boulevard du Montparnasse, and it remained in use by his business for the next two decades. It was the workshop of a well-established mechanic, Gaston Roquejoffre, who was kept on by Roux, and appears to have handled the precision engineering side of the business.

Meanwhile, Paris directories for 1918 and 1919 showed the main Ebonitine business in the rue du Chemin-Vert expanding into yet more new lines:

Boutons (fabricants et march[ands]. de) ('Roux process-moulded buttons – imitation horn and black corozo. Fancy items for tailors and seamstresses')
Eclairage électrique des voitures ('Moulded insulators for electrical equipment. Melt-proof components and sheets')
Jetons pour casinos, cafés, usines, jeux […] ('Moulded tokens in all colours')
Jumelles (fabr[icants]. de) ('Eyepieces and components in Ebonitine and Ebonite')
Magnétos ('Insulating components for magnetos, Ebonite distributors')
Radiotélégraphie ('Moulded insulators for electrical equipment. Melt-proof components and sheets')

Thereafter, Bernard Roux's range of advertised uses for Ebonitine, Ebonite and its other products remained more or less unchanged, until a significant development in the mid-1930s took a branch of his company in a new direction (see below).

The Société franco-anglaise de caoutchouc manufacturé, too, continued to be listed in the main Paris directory and other trade publications. Perhaps a sign of post-War social change, its specialities were now rubber-soled shoes for tennis and the beach, as well as a fashionable footwear fad, the talon tournant (a circular heel plate, deeply cut to leave an impression – an image or logo – in soft surfaces such as sand or soil).

Etablissements Bernard Roux

In October 1923, the Société agreed to sell its factory in Levallois to another company. After divesting itself of its main asset, the Société Franco-Anglaise then changed its name to Etablissements Bernard Roux, confirming Roux's continuing interest in the company, which was now restructured. Joining Roux on its board were Pierre Ibled, described as an industrialist, and Gaston Widmer, an engineer. The first of these names is significant: originating in the Pas-de-Calais, one branch of the Ibled family had become prominent in Dijon and was closely intertwined by friendship and marriage with that of Bernard Roux. Etablissements Bernard Roux remained in business until about 1952, and then, with a slight change of name, until 1961.

For most of the remainder of this page, the company's core business of plastics and insulators will be relegated to the background, so as to concentrate on its activities in audio recording, disc manufacture and electronics. As far as is known, the plastics and insulators side continued to flourish. No evidence has been found to indicate which was the more productive or profitable, although Roux ultimately exited from disc manufacturing first (see below).

Records and recordings

The original prompt for this page was a known if overlooked private recording by Bernard Roux of modern chamber music. Its considerable historical importance set me wondering who Roux was, why his firm made it, what else he did and why he is not better known today. It did not take long to find other Roux discs in public collections; several of these are also of historical and musical interest, yet no one, apparently, has written about them. In fact, the only people outside France who know of Roux today seem to be aficionados of early European jazz recordings, a few of which are preserved on Bernard Roux test pressings. The pressings themselves are rarely much help: those I was able to inspect in person or online usually carry little or no information on their labels about their recorded contents or intended use. But the main reason for Bernard Roux's current obscurity, I have concluded, is that his firm issued very few discs under his own name, and none for the retail trade. Private commissions, such as the recording which first piqued my interest in Roux, were probably a minor sideline of his audio business: the bulk of this production consisted of contract recordings and pressings for other firms, hardly reported in the press and largely invisible. Roux's clients were mostly small, local labels, often short-lived and usually leaving little documentary trace behind them. As a result, Bernard Roux has been overlooked by most historians – despite a strong tradition of discographical study of the French industry of the period. Yet Roux was involved in records and recording almost from the start of his career. Why he went in for this line of business is not hard to imagine. Rubber and allied materials had been used to make gramophone records since before 1900; to a young entrepreneur with expertise in plastics, this novel, growing market must have been very attractive.

1906-1920s

One of Roux's earliest patents, obtained in 1906, was for a new process of disc record manufacture; in 1907, he was granted another, for a disc-pressing mould. In that year's Didot-Bottin directory, 'Phonographes' was the only category with three entries from Roux, advertising Roux-process Ebonitine as a suitable material for sound-box diaphragms, and offering to press discs and supply metal stampers.

In a 1910 trade directory aimed at the theatrical and musical professions, a boxed display advertisement, placed between a piano dealer's and a zither-maker's, spelled out Roux's full range of gramophone products and services:

Ebonitine (Procédés B. Roux)
Bureaux: 147bis, Rue du Chemin-Vert :: Usines: 144, Rue du Chemin-Vert
Enregistrements à façon de Disques à saphir et à aiguilles en toutes dimensions
Seul presseur à façon de disques phonographiques et gramophoniques
Cires à enregistrer — Diaphragmes — Matières à disques — Pièces moulées pour applications électriques en tous Genres et toutes Couleurs — Pièces infusibles.

This is the first advertisement or listing located to date in which Bernard Roux offered not only pressing but also recording services, as well as disc-cutting diaphragms and wax blanks for those who wished to make recordings in their own studios. (Incidentally, the advertisement reveals that the Ebonitine business occupied both the original site at 144 rue du Chemin-Vert and the new premises at 147 bis.) These products and services suggest a fully-fledged record-producing operation, although much about it remains unknown. How large was it? What proportion of Roux's overall business did it account for? How many clients did it have? Where were the recordings made and by whom (almost certainly not by Roux in person)? And was Roux's really the sole custom recording and pressing facility in Paris? The latter claim seems justified, going by the 1909 to 1911 Didot-Bottin directories, which list no obvious competitors: one company offered 'blank discs for recording', presumably waxes, but no recording facility; another advertised the 'Only device permitting the recording of blank discs on any make of disc machine, with instant replay', alongside the 'planing' (rabotage) of discs, presumably to remove existing grooves. None of the entries for the larger, well-known companies – Columbia, Compagnie du Gramophone, Pathé – mention custom recording or pressing services. (Pathé did manufacture discs for labels such as Aspir, Dutreih, Idéal and Phono, but did not advertise this service in directories.)

The reason such services were needed is that small record labels did not own matrix processing or disc pressing plants, and so were obliged to contract out these industrial stages of record manufacture. How many French firms possessed such facilities has not been established – far fewer, certainly, than the plethora of labels marketed during this period. If the 1910 directory quoted above suggests Roux was already active in the contract processing and pressing business, or aiming to enter it, his firm's entry under 'Disques pour phonographes' in the 1922 Didot-Bottin directory shows this activity expanding even further:

Ebonitine
procédés B. Roux
Pressage de disques.
Enregistrements.
Création de répertoires.
Machines à enregistrer.
Cires, galvanos, diaphragmes.

Up until 1910, Roux had advertised disc recording and pressing services, materials and equipment; now, he was offering 'catalogue-building' (création de répertoires) – Artists & Repertoire ('A&R'), in modern parlance. (Note also the advertised 'recording machines'; these were probably destined for professional studios, not for home or personal use – but see below.) Presumably, this catalogue-building service was aimed at firms and retailers wishing to launch their own labels but lacking both the necessary editorial and commercial expertise, recording and/or pressing facilities.

Which labels, if any, took up Roux's offer? Unfortunately, Roux did not name clients in trade directory entries. It is sometimes possible to identify contract recordings and pressings from their physical features or characteristic markings, but most Roux pressings currently documented are not distinctive enough for that. It seems unlikely that documentary evidence for collaborations between Roux and long-defunct, small commercial labels will emerge, although not impossible (such evidence does survive from the electrical period and will be discussed below). For a fuller picture of Roux's contract production, test pressings have to be matched up with issued commercial discs, and extrapolations made from observed patterns.

Cylinders

In 1913 and 1914, entries for Ebonitine appeared in the Paris directories under Cylindres pour phonographes et graphophones – somewhat surprisingly, in hindsight, as by this time the format was all but obsolete. In 1913, two other businesses had entries in this category; in 1914, just one, while a few more companies appeared under Graphophones, the category for cylinder players. No evidence has yet been found that Ebonitine was actually used in the production of cylinder records. Perhaps the very fact that most leading producers had dropped the format prompted Roux to see if anything could be made from its last gasp. At least one other business apparently saw the same opportunity: in the 1918 Paris directory, cylinders had disappeared but players were listed for the last time, attracting a single, new entry from the Palais de la Nouveauté, a budget department-store chain. For lack of both evidence and, regrettably, expertise, this topic is not examined further here.

Electrical recording

After more than a decade in audio disc production and manufacturing, Roux would have been well aware of new trends and openings in the industry. In the early 1920s, developments in telephony, microphones and electronics (notably, valves) combined with the rise of broadcast radio and cinema to trigger an international race to replace acoustical recording with electrical techniques. The best known and best documented of these is the Westrex system, developed in the United States by Bell Telephone Laboratories and Western Electric, who licensed it to leading national and multi-national companies. Other systems were developed and exploited with varying degrees of success and recognition – one, possibly the least well documented of all, by none other than Bernard Roux. A 1936 newspaper article described him (with questionable accuracy) as

the first who thought of using triode valve amplification for [sound] recording

Questions of priority aside, in March 1923 Roux had applied in Paris for a patent in a new recording process 'employing the electrical techniques used in telephony'; it also proposed using multiple microphones and a mixer, as well as a headphone feed permitting real-time monitoring. The patent was granted a little over a year later, by which time Roux had applied for Swiss and British patents in the same technology; both granted. Another patent soon followed, for an electrical method of duplicating sound recordings and transferring them between different media (see next section). These patents await investigation by historians of technology; in particular, the construction and properties of the transducer(s) used seem not to be fully described in these patents.

By early 1925, Bernard Roux's engineers had made their first successful electrical experiments. An academic article by a member of the team, published in early 1932, claimed these as France's first commercial electrical recordings. A newspaper article printed a few weeks later related that a number of electrical disc recordings, notably of the piano, had been made by the engineers Georges Laudet and Jean Ibled in Roux's laboratories in 1924-25, and later deposited with Hubert Pernot (1870-1946), head of the Musée de la parole. Laudet had been a sound engineer since the turn of the century, and is best remembered now for his work on film-sound synchronization for Léon Gaumont; Ibled (1890-1965) was a decorated War veteran and related to Roux by marriage. A test recording, clearly deriving from these experiments, is discussed below.

For whatever reason, whereas Karavodine's patented Ebonitine production process (ceded to Bernard Roux) was known as the 'Procédé(s) Roux', the Roux electrical recording technique is not known to have been given a name.

Duplication and transfer

In November 1924, Bernard Roux applied to patent a 'Technique for electrical duplication of phonographic and photographic sound recordings', granted a year later. A subsequent application in Switzerland was also granted; a third, in Great Britain, became void. As the patent descriptions state, Roux's proposed electrical method for duplicating or transferring recordings aimed to supersede the mechanical ones employed until then (notably, by Pathé, which continued to make master recordings on cylinders, even after the format had become obsolete, and to transfer them mechanically to disc masters). Roux claimed his system was capable of transferring any sound recording from one medium or format – cylinder, disc, even optical film – to another, without loss of quality; on the contrary, defects in the original could be corrected by electrical filtering and amplification. In the competitive market Roux was targeting, the facility to change size and cut might be a considerable advantage (as, too, would the ability to transfer recordings to and from film). No press reports mentioning this technology have been encountered to date. If Roux exploited it commercially, considerable detective work may be required to identify its products.

Studios

Two venues used by Bernard Roux to make sound recordings are currently documented. Almost certainly, there were others, but to date no evidence for them has been uncovered. In particular, no mention has yet been found of venues used by Roux before 1922, during the entire first decade and a half or so of his firm's possible recording activity.

The 1932 account of Roux's first electrical recordings, cited above, stated that they were made in the 'studios Bernard Roux' in the rue Rochechouart in Paris. The plural 'studios' suggests that more than one space was in use there, but in the earliest known published reference to the facility, a mid-1924 Paris newspaper advertisement, just one 'recording hall' was mentioned, and its exact address given:

On dem. jeune homme sérieux, 17 à 19 ans pr trav. manuel facile. Se présenter Salle d’Enregistrement, Bernard Roux, 24, rue Rochechouart, après-midi

(A very similar advertisement, published fifteen months later, gave the same address but did not mention the studio.) This address housed part of the head-quarters of the well-known instrument-maker Pleyel et Compagnie, which at various times occupied numbers 20, 22, 24 and 24 bis, rue Rochechouart, in Paris's 9th arrondissement. (Pleyel's main factory was in the northern suburb of Saint-Denis.) The best known of these addresses is no.22, that of the 'old' Salle Pleyel, the concert hall inaugurated in December 1839 and immortalized by the celebrated pianists who performed there, notably Chopin. To either side of this hall were workshops, offices, showrooms and salons, as well as a second, smaller hall. Situated at no.24, this became known as the 'Salle des quatuors', since it was used mainly though not exclusively for chamber concerts. Its other documented use was pedagogical; as a child, the future conductor and composer Désiré-Émile Inghelbrecht (1880-1965) was taken there to perform by his mother, a piano-teacher. Not long after, the charity L’Oeuvre de Mimi Pinson, founded in 1900 by the composer Gustave Charpentier, put on free classes in music and dance for unmarried working women in the hall at no.24; a magazine article by Charpentier about his venture was illustrated with some of the only photographs found to date of the smaller 'salle Pleyel'. After World War I, unlike the larger main hall, the smaller hall seems to have fallen out of use for commercial concerts. Instead, it was used for student concerts and private lessons by singers and other teachers, notably the pianist-composer Lucien Wurmser (1877-1967): initially based elsewhere, from 1916 until 1927 Wurmser's annual courses were held in a 'salle Pleyel', probably the smaller.

Based on the above evidence, my current hypothesis is that this hall, Pleyel's former 'Salle des quatuors', was used as a recording studio by Bernard Roux from at least 1922 until 1927. Recording equipment was perhaps installed and left in place in an adjoining room; this would surely not have been possible with the main hall, which was in regular public use and opened onto other public spaces, notably showrooms. The smaller hall would also have been easier to drape or otherwise treat for the purposes of recording. Finally, Roux's recording set-up might have been available for hire by teachers who shared the venue: in November 1928, a classified advertisement in a Paris newspaper offered for sale a

Machine à enregistrer électrique, permettant d’enregistrer soi-même, conviendrait pour professeur, s’adresser p[ou]r essai. Blanc, 24, rue Rochechouart.

The advertisement did not name Bernard Roux as the seller but the address given was that of Pleyel's smaller hall, so it seems reasonable to conjecture that the recording machine in question had been part of the Roux equipment there, while the mention of a professeur (i.e. teacher) suggests that its pedagogical use was a familiar notion.

Besides the early electrical experiments conducted by Bernard Roux recordists in the rue Rochechouart (mentioned above and described below), at least one other surviving recording is known to have been made there, acoustically, a little over two years earlier (see below). The presence of this still relatively new musical technology within the precincts of Pleyel was not alien to the firm's director Gustave Lyon (1857-1936), described admiringly by Inghelbrecht as an inveterate inventor who 'unflaggingly demanded of science that it serve art'. As early as 1889, an Edison phonograph was installed there, in its own, dedicated room, and demonstration recordings were made on it; a cylinder previously recorded in New York was reportedly also replayed on it. In late 1906, Lyon had been a founding board-member of the Société Générale d’Impressions Phonographiques, a short-lived record company dissolved not long after. In 1908 and 1909, its showroom was in the Maison Pleyel, at no.22, but it is not known to have made recordings there. Twenty years later, Lyon developed his own electrical record-player and planned to make his own recordings; Lyon himself demonstrated the machine in public in late 1929, but whether any 'Pleyel' or 'Lyon' discs were made is, again, unknown. Further unknowns: when Bernard Roux started using Pleyel's premises; the terms of their relationship (did Roux pay Pleyel rent or session fees?); whether the Maison Pleyel was the only location used by Roux during that period; and whether other firms also recorded there.

In 1927, Pleyel et Cie moved its head-quarters from the rue Rochechouart to a new, purpose-built complex on the rue du Faubourg Saint-Honoré in the 8th arrondissement. At its heart was a large concert hall, the Salle Pleyel, still in use today. Roux's association with Pleyel is not known to have been maintained after the move; recordings were made in the 'new' Salle Pleyel for one of Roux's clients (see below), but not by Roux, who was contracted only to press these discs, not record them. The equipment in the rue Rochechouart would not have been needed in the new venue, which probably explains the advertisement quoted above. The old Salles Pleyel seem to have closed definitively in 1928; after that year, they disappear from Paris directories, although the buildings around them remained in use by various tenants.

Still, Roux's association with Pleyel may have borne other fruit, in the form of a new direction of research and development for his own company. Within the old Maison Pleyel was an office specialising in architectural acoustics, led by Gustave Lyon himself. No evidence of personal exchanges between Lyon and Roux has been uncovered, but it is striking that Roux's firm later developed technology for shaping studio acoustics electronically (discussed below), in association with the composer and organist Eric Sarnette. This was demonstrated publicly in early May 1936, in a 'studio-laboratoire Bernard Roux' which is currently known only from two press reports of that demonstration. It was located in the rue Jenner, in Paris's 13th arrondissement. The exact address is not known: it was certainly not numbers 6 to 8, later renowned as the birthplace of many famous recordings; this already belonged to the Société phonographique française Polydor (and, later, to Philips). It is more likely to have been number 25 bis, a disused factory later used as a film studio by the director Jean-Pierre Melville and almost completely destroyed by fire in 1967. Two trade directories of the mid- and late 1930s list a 'recording studio' (studio d'enregistrement) among facilities offered by Etablissements Bernard Roux, without stating its location; Roux's engineers surely did not use the space in the rue Jenner solely for research and development, so this may have been the facility referred to in the directories.

It remains to be determined whether Roux also used other spaces for recording sessions.

Pressing plant(s)

Bernard Roux had an interest in disc-pressing even before the Ebonitine works came into being, and maintained it for much of his working life: one of his earliest patents, applied for in late 1905 and granted in 1907, was for a disc-pressing mould, while one of the last applied for by his firm, in May 1945, and granted in 1946, was for a method of pressing claimed to offer great savings of energy and metal, then scarce.

From the start of Roux's manufacturing career, Paris directories gave full addresses for his business, but they rarely stated explicitly what was being done at each location. As noted above, the 1907 Didot-Bottin, the first with detailed entries for the Ebonitine works, gave the main address and site of its workshops (plural) as 144 rue du Chemin-Vert, with a workshop (singular) and 'industrial testing laboratory' at 2 & 4 rue Esquirol, while all entries under individual product categories gave only the main address.

This remained true until 1910, when another directory clarified, 'Usines: 144, Rue du Chemin-Vert'. Taken at face value, this means that within the ordinary-looking Parisian apartment building at no.144, alongside other manufacturing plant, was a disc-pressing facility – of what capacity, and how productive, is not known. This directory also stated, 'Bureaux: 147bis, Rue du Chemin-Vert', a new location for the Ebonitine business. In the following year's Didot-Bottin it was given as the main address, with no mention of no.144, while a workshop and lab remained at 2 & 4 rue Esquirol; correspondingly, product entries now listed only 147 bis, rue du Chemin-Vert. From now on, offices and disc presses were apparently in one place again, where they remained until 1919 (the 1920 directory is not accessible).

From 1921, directories listed yet another new main address for Ebonitine, 142 rue du Chemin-Vert, with a workshop and lab still at 2 & 4 rue Esquirol. If all the Ebonitine plant and presses really had been moved back across the street from 147 bis, at least it was not far. But this may have been a simplification: in the 1926 Didot-Bottin, the main entry for Ebonitine again listed it first at no.142 and secondly at rue Esquirol, but now the entry listing all businesses at 142, rue du Chemin-Vert, stated:

« Ebonitine » (Bernard Roux, ingénieur E.P.C.), pièces isolantes moulées; ateliers au 144, comptabilité au 147 bis.

If the main office was at no.142, the workshops at no.144 and accounts at no.147 bis, perhaps the plant and presses had never moved from no.144, and it had not been felt necessary to mention this in listings aimed at potential clients. (Nor had directories since 1924 reflected the fact that the business was officially named Etablissements Bernard Roux.) Whatever the case, this set-up remained in place until at least 1929, possibly until later, although subsequent directories are not as detailed or consistent. In May 1933, a tragic accident befell two children who had climbed onto the roof of the works at 144, rue du Chemin-Vert and fell to their deaths; newspaper reports described it as a 'gramophone record factory' and noted that 'the Roux firm has long been manufacturing raw material for gramophone records'. These sad reports confirm that no.144 had been and was still an industrial production site, possibly of both 'biscuit' and pressed discs.

Around this time, though, either the Bernard Roux presses were on the move, or the firm was adding a second pressing facility. In April, September and October 1932, and July 1933, it applied for permits to carry out building work at 12, place Pinel, a small, oval circus at the southern end of rue Esquirol. Nothing seems to remain of the buildings then standing, but no.12 appears to have abutted the Bernard Roux workshop and laboratory at nos.2 and 4, rue Esquirol. The new constructions on place Pinel were more workshops (ateliers), a small electrical sub-station (cabine électrique) and large shed (hangar). By June 1932, Roux was inviting record-press operators to apply to no.4, rue Esquirol, who would have been needed for the new facility at place Pinel, presumably not yet ready to open its gates. That pressing did take place there is confirmed by notices of an auction at 12, place Pinel seven years later, when Bernard Roux sold off large quantities of heavy industrial machinery, including at least fifteen record presses. Otherwise, this location was almost never mentioned in business reports and directories, surely because it was treated as part of the complex sited at 4, rue Esquirol, listed in two trade directories of the mid- and late 1930s as the address of one of Bernard Roux's three factories (the others were at 144, rue du Chemin-Vert, still, and 43, avenue du Maine, formerly Gaston Roquejoffre's wokshop). Another advertisement soliciting applications from record-press operators (specifying that they should be female and French nationals), in December 1934, again gave the address as 4, rue Esquirol.

(Another activity carried out by Roux at the firm's rue Esquirol complex was the production of moulded Bakelite components; as far as can be ascertained from classified advertisements recruiting experienced mould-workers, this went on roughly from mid-1929 to 1934 or later.)

Disc-manufacturing was heavy, dirty industry; nor was it the only potentially harmful activity at these sites. As early as July 1933, a local councillor passed on to Paris's Prefect of Police complaints he had received about emissions from the factories in place Pinel and rue Esquirol, 'highly noxious' to local residents and a nearby maternity ward, and about the lack of remedial action by the cuplrits. He was assured that up-to-date boilers and smoke abators were being installed, existing chimneys raised and a new one built. No firms were named but the addresses given leave no doubt: as well as those of Bernard Roux's works, Polydor's address on the rue Jenner was mentioned, and a fizzy drink factory on a nearby boulevard.

Test pressings

Below is a summary list of all Bernard Roux test pressings located for this page – a little over fifty sides, to date. The recordings themselves will be discussed in later sections and, where possible, documented more fully.

The disc which prompted this page is one of six Roux tests held by the Bibliothèque nationale de France in Paris (for full catalogue entries, follow links below):

• 'Bernard Roux – Octuor Finale', no matrices given, 30 cm, BnF shelf mark SD 78 30-12984
• 'Guitare hawaïenne' / 'Chœur v.f.[?] av. Grand Orgue', E 584 / 3586, 27 cm, BnF shelf mark AP-2175
• 'Ecoles internationales Phonogramme de français N°2' / 'Ecoles internationales Phonogramme de français N°3', 27 cm, 3670 / 3671, BnF shelf mark AP-2174
• 'Premier disque à aiguille enregistré électriquement le 28 janvier 1925' / 'Orchestre enregistré électriquement le 28 juillet 1926', Essai spécial no.57 / 3698, 27 cm, BnF shelf mark AP-2173
• 'Chant Hébreu' / 'Chant Hébreu', 4120 / 4121, 25 cm, BnF shelf mark AP-2172
• 'Etablissements Bernard Roux 8510', 8510 / [?], 25 cm, BnF shelf mark SD 78 25-19949

Only two Bernard Roux test pressings have been found in national and university libraries elsewhere (catalogue entries via links):

• [no titles], 1106 / PK 200, 25 cm, British Library, London, shelf mark 9TS0002803
• 'Don Goyo / Bernstein. El matrero / Soler.', 4318 / 4319, 25 cm, University of California, Santa Barbara Library, shelf mark Bernard Roux 4318 / 4319

(All remaining tests listed below are 25 cm in diameter, except where stated.) In April 2016, I bought a Bernard Roux test pressing, and in April 2023 two more, from Phonopassion, of Horben, Germany, who described them as follows (labels from the first and third are reproduced below):

anonymous (cello solo ) Prelude en Ut mineur pt.1/2 (Bach) / published sides ? ? ak[ustisch] ? Bernard Roux test

anonymous (sop.) +piano: Se tu m'ami (Pergolesi) mx.4811 AB/ Lasciate me morire (Monteverdi) mx.4810 AB Paris ca.March 1929 released sides ? Azurephone is a very rare label. Usually less than 100 copies had been pressed once a record was released. Handwritten white label test Bernard Roux (Azurephone) test

(sop.)/(contr.) el. Bernard Roux (Azurephone) test unidentified (mx.9576/ 9577) Paris ca.1930

Revisiting past Phonopassion sales yielded two more Bernard Roux test pressings:

anonymous 8 2/3” [22 cm] el. Bernard Roux test (unidentified pianist) Finale: Presto agitato “Mondscheinsonate” (Beethoven) mx.6256 MB/ Fantaisie impromptu (Chopin) mx.6257 MB Paris ca.1930 Released recordings?

anonymous 8 2/3” [22 cm] el. Bernard Roux test (unidentified pianist) Moment musical (Schubert) mx.6343 MB/ Perpetuum mobile “Sonata in C” (Weber) mx.6342 MB Paris ca.1930 Released recordings?

Phonopassion's proprietor, Andreas Schmauder, kindly sent details of five other Roux tests known to him:

anonym (Orchester), Danse des souliers, mtx 11677 / ? mtx 11678, Bernard Roux test

Berson, Roger (European Ramblers) + Jean Farnèse (Gesang), Un deux... toute la compagnie, mtx M 142 A (Foxtrot with vocal) / Nana, mtx M 139 A (Rumba with vocal), Paris 1932, Bernard Roux test

Hamilton, Jack & his Entertainers, That’s my weakness now (Foxtrot with vocal), mtx 4748 AB / Virginia (Foxtrot with vocal), mtx 4749 AB (released on Azurephone in a different coupling), Paris 1929, Bernard Roux test

Briggs, Arthur & his Boys + vocal duet, Glad rag doll (Foxtrot with vocal), mtx 4806 AB / Rudy Bayfield Evans (Refraingesang), Only for you, mtx 4807 AB (Waltz with vocal), Paris 1929 (released on Azurephone), Bernard Roux test

Sterman, Paul (Orchester), Mai-Thou (Foxtrot), mtx AB 10966 / Débinons-nous (One Step), mtx AB 10967, Bernard Roux test

The last item above is also known in a finished pressing which, unusually, bears the name of Bernard Roux and belongs to a special category of issues discussed below; a copy was sold online in mid-2021:

Jazz Paul Stermann Maï-Thou 78RPM Bernard Roux 10966/67

This and other closed auction listings are usefully archived by websites designed as price guides for record collectors and sellers, mainly of vinyl discs but also of other media; these listings document a few more Bernard Roux tests:

Mario Podesta Tenor Wagner Lohengrin 78 RPM Bernard Roux 7131/32

Etablissements Bernard Roux 7330/7331 (10”) Hot Melodic Band (Julien Porret) Great Fun Test Pressing 78RPM Bernard Roux

Etablissements Bernard Roux 7334/7335 (10”) Hot Melodic Band (Julien Porret) Pupazzi Test Pressing 78RPM Bernard Roux

Wurmser Trio Rimsky-Korsakoff Sadko 78RPM Bernard Roux S 25040/42

(The Podesta test above is coupled with an unidentified song, and the Wurmser Trio one with a song by Reynaldo Hahn, both sung by unnamed sopranos.)

In late March 2024, Jerome Moncada helpfully published images of other Bernard Roux tests, supplied by fellow-collectors or retrieved from the internet:

• Bernard Roux 4431; size, cut, system and coupling (if any) unknown; label annotations (in later hand): ‘Fox Trot. Orch. Musette’
• Bernard Roux 18325; size, cut, system and coupling (if any) unknown; label annotations (possibly contemporary?): ‘Marche funèbre de Chopin Jazz’
• Bernard Roux 21506; size, cut, system and coupling (if any) unknown; label annotations (in later hand?): ‘Moi je m’en Balance Chanson comique’
• Établissements Bernard Roux 1748; size and coupling (if any) unknown (probably lateral-cut electrical); label annotations (contemporary or later hand?): ‘Le chaland qui passe’
• Etablissements Bernard Roux AB 11778; size and coupling (if any) unknown (probably lateral-cut electrical); label annotations (in later hand): ‘Bourrée Valse pastourelle’

The last of these images is especially welcome, as it depicts a previously unseen test label design confirming the sequence of Roux pressing plants set out in the previous section.

Further sides are documented only in discographies:

The Red Beans
Robert de Kers tr. Francis “Sus” van Camp tbn. David Bee al-bs. René van Dijck pf. Charlie “Chas” Dolne ch-vl. Leopold Serluppens bt. Rudy Bayfield Evans v.

Parigi, ca. agosto 1928

35 AB Wonder Why (You Made Me Cry) – RBE v Bernard Roux test […]
36 AB Do Let Me Try [ditto] Bernard Roux test

4849 JPRM Sunny Boy (Sonny Boy) – RBE v Bernard Roux test

Marcel Dumont, de l’Empire, accomp. d’orchestre
janvier 1931?
6120 AB Publicité pour le [vin] blanc des Magasins du Louvre
6121 AB [ditto]
Disque publié par Les Editions [sic] Bernard Roux, 142 rue du Chemin-Vert, Paris. Echantillon invendable.

Clearly produced for promotional purposes (see below), this disc was not identified as a test pressing in the source cited; but the legend Echantillon invendable ('Sample not for sale') strongly suggests it carried test labels of the second design described in the following section.

Test labels

Three paper labels for Bernard Roux test pressings currently are known. The first and oldest type is illustrated below, along with the segment of the runout or 'dead wax' in which the matrix number is stamped:

Bernard Roux matrix 4502 label, with runout or 'dead wax' detail (Collection: the author)

The printed text reveals nothing about Bernard Roux, his firm or the purpose of the pressing. A single field is defined, '№'; as seems to have been standard practice, the matrix number stamped in the runout was duplicated here in pencil. (Most Roux tests have a matrix number in this area but not all.) Typically, this number is a sequence of three or more digits, sometimes with a one- or two-letter prefix or suffix.

The florid name and monogram are reminiscent of similar marks used before World War I on 'Beaux Arts'-style record labels and players, and in bulletins, catalogues and related merchandise. Neither mark has been seen in Bernard Roux advertisements in the French press and it is not known if they were used on the firm's other, non-gramophone products (none of which have been seen either).

The second type of Roux test label is more modern-looking and slightly more informative:

Établissements Bernard Roux matrix 9576 label, with 'dead wax' detail (Collection: the author)

The upper half carries the name Établissements Bernard Roux and address 142, rue du Chemin Vert, Paris, suggesting the design was introduced after the Établissements was formed in late 1923 – but with some delay, perhaps, as test pressings are known which are dated later yet carry the earlier label. The legend 'Échantillon invendable' is an exact equivalent of the '[Factory] Sample Not for Sale' often seen on British and American test pressings; this did not necessarily mean that Roux tests with this label were as destined for commercial issue, as this very recording was almost certainly a private, non-commercial commission (see below). No fields are defined, not even for the matrix number, which has been duplicated, again in pencil, in the label's blank portion.

The third and latest type of label, not white but buff in colour, carries the last address of Établissements Bernard Roux and location of its last pressing plant, 2, rue Esquirol, Paris, XIII^e, and revises the 'Sample Not for Sale' legend to 'Épreuve non vendable'. As before, a little over half the label is left clear for hand-written annotations.

Bernard Roux issues

To date, only one disc issued under a Bernard Roux imprint has been documented. Known from an online sales listing, this unusual disc was almost certainly not aimed at the general public but at a little-known business-to-business trade, akin to what is known today as 'production music'; it is discussed in the relevant section below.

No published discs bearing the name Bernard Roux and destined for commercial retail sale are known.

Bernard Roux publications

No printed matter, other than disc labels, published by Bernard Roux or emanating from his businesses has been located.

Identification and interpretation

When this page began to be drafted, only a dozen or so Bernard Roux test pressings had been documented, and even fewer had been auditioned – just the five digitised by the BnF. The assumption that these recordings were non-commercial was reinforced by the BnF's catalogue entries, which describe its Roux tests as made for archival purposes ('Circuit de distribution: archives') on behalf of the University of Paris ('Marque: Université de Paris'). As noted above, the labels of these and the few other tests then known convey little to no information about the discs' contents or intended uses and users. To begin with, there was nothing for it but to listen to the tests and scour the internet for vaguely related recordings and similar-looking matrix numbers.

Fortunately, a long-standing expert came to the rescue: Andreas Schmauder, of Phonopassion, shared details not only of other Roux tests known to him but also of commercial discs issued from the same matrices. The latter were a discographical Rosetta stone, unlocking the code of Roux matrices and opening a window onto the little-known trade which Roux served for almost three decades.

The following sections set out the current, occasionally rudimentary state of knowledge about Bernard Roux tests documented to date, grouped into categories in no particular order. Inevitably, some identifications and interpretations involve speculation, and some guesses made in earlier drafts of this page were wrong.

An important distinction to bear in mind is between recordings made by Roux's firm and those which it pressed but did not itself record; among the tests examined below, several can now be assigned to the second category.

Private recordings

As stated at the outset, this page was prompted by a private recording known to have been made by Bernard Roux. It happens to be the earliest securely dated example of his firm's work, and is of considerable historical and musical interest, so it seems appropriate to begin exploring Roux's recordings with it.

In 1923, the émigré Russian composer Igor Stravinsky (1882-1971), then living in Paris, made his concert début as a conductor. On 18 October, at one of the Grands Concerts Symphoniques at the Paris Opéra conducted by his compatriot Serge Koussevitzky (1874-1951), a single item on the programme was entrusted to Stravinsky: his Octet for Wind Instruments, receiving its world premiere. Beset by money worries, the composer was turning towards performing himself, as a way of supporting his large family and dependents (and fairly lavish lifestyle), and he had been persuaded to bring forward the premiere, originally promised to another promoter, in return for a fee he could not refuse.

Some days or weeks later, Stravinsky made his first disc recording, again conducting the Octet. Two published accounts of the recording are based on primary sources. The first is very brief, a footnote in the second of three volumes of Stravinsky's correspondence, selected and edited by the writer, musicologist and conductor Robert Craft (1923-2015):

In November 1923 Stravinsky recorded the Octet in the “salle d’enregistrement” at Pleyel, 24, rue Rochechouart, and a number of copies were distributed. The sides are numbered 796-797 and 798.

Born two days after the Octet's premiere, Craft was for decades Stravinsky's amanuensis, companion, adviser and chronicler; he must have obtained this information from the composer's personal archives. The address of the recording room, not widely circulated at the time or later, is given exactly as in the 1924 newspaper advertisement cited above. And Craft's enumeration of the side numbers, in the present tense, suggests he was transcribing them from the discs themselves or from a document such as an invoice. (The numbering presumably meant that the first two sides were coupled together on one disc, with the third on another, single-sided). It may seem odd that Craft did not note that three sides are too few to accommodate the entire Octet: but if he did have test pressings in front of him, there were probably no inscriptions on the labels. The recording company, by then long defunct and probably not familiar to Craft, did not get a mention.

The second account was published in 1991 by the leading discographer Philip Stuart, in his reference work on Stravinsky's recordings of his own music. A portion of the opening chapter, chronicling Stravinsky's first encounters with the gramophone and with reproducing-piano rolls, is devoted to the Octet recording. It enlarges considerably on Craft's footnote, with important details drawn from correspondence between the composer and the recording company, now named – seemingly, the first mention in print of Roux alongside Stravinsky:

This was a private recording made by Bernard Roux, an engineering firm who had a recording studio in the Pleyel building in Paris. Stravinsky acknowledged the receipt of test pressings in a letter dated 3 December 1923. The Finale of the Octet was incomplete as the cutting engineer ran out of wax before the music ended, but the composer requested delivery of this side despite the defect.

(This may explain why Craft listed only three sides in his footnote; these did not include the side with the incomplete Finale.) A little further on, emphasizing that 'what follows is conjecture', Stuart surmised that

A reference in the correspondence to an intention to record another work, as fill-up for an odd side, was perhaps a ploy to establish that Roux owed him a side in lieu of the one they had failed to capture in full.

These appear to be the only published references to correspondence between Stravinsky and Bernard Roux, which is otherwise undocumented.

Stuart cited one other letter, from the British aristocrat and composer Gerald Tyrwhitt (Lord Berners), who was anxious to obtain the Octet records and asked Stravinsky 'where I must write to order them?' (This was the letter to which Craft appended his footnote about the Roux recording; Stravinsky's answer, if any, was not relayed.) Although 'a number of copies were distributed' – possibly prompting Tyrwhitt's misapprehension that the set was for sale – Stuart argued, convincingly, that it was 'never intended for publication' by Stravinsky but as 'a private recording, for his own exclusive use', to 'give him the chance to study and learn from his mistakes.' More than a decade later, Stravinsky's first memoir related how, at the Octet's premiere, his lack of experience as a conductor had hampered his efforts to realise its unusual sonorities. Implying that no further recording took place, Stuart observed,

The incomplete Finale would not have mattered enough to justify the cost of a further session to put it right.

Stuart's discography entry for the Octet recording reprised the date, location and three side numbers from Craft's footnote, added a fourth side (number unknown) for the incomplete Finale and ventured that the discs were 30 cm / 12 in. in diameter. The latter suppositions are supported by the two subsequent commercial recordings of the Octet, under Stravinsky (1932) and Leonard Bernstein (1947), which both took up four 30 cm / 12 in. shellac sides.

One detail of the Finale débâcle will puzzle anyone familiar with the Octet and with the playing times of 78 rpm discs. The movement usually comes in well short of a 30 cm side's maximum duration of over four minutes: in Stravinsky's 1932 commercial recording, it took some 3 min. 23 sec, and in his first for LP, in 1954, 3 min. 43 sec. The Roux recordist simply could not have run out of wax – unless the side contained more than just the Finale. In his discography, Stuart made no guess as to how the Octet's first two movements were laid out; the simplest scheme, anticipating the lay-out of Stravinsky's 1932 recording, would be:

Side no.	Selection	Matrix no.	Diameter
I.	(i) Sinfonia	796	30 cm
II.	(ii) Tema con variazioni (start)	797	30 cm
III.	(ii) Tema con variazioni (end)	798	30 cm
IV.	(iii) Finale (incomplete)	unknown [799?]	30 cm

But perhaps the 1923 Roux recording was not laid out as above, but as in Bernstein's 1947 recording. In that scheme, the Finale follows the end of the Tema con variazioni on the same side, preserving the attacca transition between them, a moment which Stravinsky might well have wanted to hear himself negotiate:

Side no.	Selection	Matrix no.	Diameter
I.	(i) Sinfonia	796	30 cm
II.	(ii) Tema con variazioni (start)	797	30 cm
III.	(ii) Tema con variazioni (continuation)	798	30 cm
IV.	(ii) Tema con variazioni (end) (iii) Finale (incomplete)	unknown [799?]	30 cm

If enough of the Tema con variazioni was recorded along with the Finale, the last wax could certainly have run out. Such an elementary mistake seems hard to credit, but before the Octet session, Bernard Roux is not known to have recorded any item spanning more than one side, so his recording manager would not have been in the habit of planning side-breaks, while Stravinsky, in later years no doubt closely involved in this tricky aspect of recording short-play discs, had no experience of it either.

But another puzzling detail casts doubt on both schemes: the plan, mooted in the correspondence, 'to record another work, as fill-up for an odd side'. An even number of sides would have created no room or need for a fill-up, so was the lay-out different? Were the discs not 30 cm in diameter but smaller? Several Bernard Roux tests in the list above are 27 cm discs, most are 25 cm and a few are 22 cm. The last can be discounted for present purposes, and 25 cm sides are too short for the Octet, but it would probably fit on five 27 cm sides, leaving a sixth free for the fill-up. Unfortunately, this hypothesis rides roughshod over Craft's footnote, which lists only three sides, and it remains pure speculation, unless new information about the 1923 Roux recording comes to light.

And so it has. Some years ago, after consulting Stuart's discography, I searched the online catalogue of the Bibliothèque nationale de France and found the following entry:

Titre(s) : Bernard Roux – Octuor Finale [Enregistrement sonore]
Publication : [S.l.] : [s.n.], [s.d.]
Description matérielle : 1 disque : 78 t ; 30 cm
Référence(s) commerciale(s) : [sans marque] [000]

I suspected this was the ill-fated fragment of the Octet's Finale but I was not in a position to act on my hunch. In late 2022, finding the entry still unchanged, I requested a digital transfer of the disc and an image of its label from the BnF's Image and Digital Services Department. The first surprise came when I was informed that the disc is double-sided; the second, when I received the audio and image files in December 2022 and heard the Finale, complete and accident-free.

Musically, despite occasional, very slight ensemble problems, the recorded performance is confident and accomplished, and sounds as if the work was fresh in the artists' hands, lips and minds. Technically, too, the recording can be counted a success: the sound, despite the limitations of the acoustical system, is lifelike and clear, with a natural – and naturally uneven – balance between woodwind instruments and brass, just what one might expect from a competent and experienced recordist faced with difficult material. In principle, the BnF should at some point publish the transfers and images on Gallica, although as of April 2024 it has not done so.

In the meantime, some additional details about the recording. The sides are 1 min. 26 sec. and 2 min. 17 sec in duration. The matrix numbers are 801 and 802, close enough to those known from Craft to indicate that these sides were recorded very soon after the reportedly unsuccessful first attempt, if not immediately after. The numbers, stamped in a concentric row in the runout area, are laid out unconventionally: the digits are rotated with respect to each other and read not from left to right but from top to bottom (apologies for the clumsy presentation below – permission has not been sought to reproduce the BnF's photographs here):

8
0
1

This style is also seen on other Roux tests, including one held by the BnF (see below).

The test's labels yield other important details. The inscriptions are in garish crimson ink, hand-written well after the pressing was made, over earlier (original?) inscriptions in pencil, now mostly erased:

VI Octuor – Stravinsky Finale Ire Partie

[V]II Octuor – Stravinsky Finale II partie

On both labels, the composer's name is somewhat illegible, perhaps explaining why the cataloguer missed it. On the second label, the Roman numeral is partly hidden by one of the provenance stickers affixed to each side, which name the disc's previous owner:

Musée de la Parole Legs Ferrant

Collection Guy Ferrant

Guy Ferrant (1898-1954) was an actor and tenor, and the partner of the composer and performer Reynaldo Hahn (1874-1947). Ferrant amassed his collection of some 3,000 records in the 1930s; devoted mainly to vocal music, it was greatly admired by Hahn, who praised it in an impassioned defence of private collections as valuable recorded archives of historic singing, then neglected by music teachers and the record industry itself. How Ferrant acquired the Bernard Roux record of the Octet's Finale is not known, but Hahn and he moved in exalted musical circles, and would have been in a position to acquire rarities. In 1938 Ferrant donated the bulk of his collection to France's early sound archive, the Musée de la Parole et du Geste; from there, 1863 shellac discs and 65 cylinders made their way, via a series of institutions and departments, to the BnF. Reportedly, some of Ferrant's discs were disposed of separately, but it does not seem probable that he ever owned the rest of the Octet and discarded it, keeping only the re-recorded Finale.

The provenance of the BnF's copy of Bernard Roux 801 / 802 is the least of the uncertainties surrounding it. A few of the others are: when was the recording made? Who performed the Finale and who conducted it? Why was it split across two sides? What do the Roman numerals on the test pressing labels mean? Why did neither Craft nor Stravinsky's correspondence with Bernard Roux mention it? And how did Roux come to record it?

After many months of pondering the identities of the conductor and players, even agonizing over them, I believe the players were more or less those billed (by surname only) on the poster advertising the Grand Concert Symphonique Serge Koussevitzky of 18 October 1923 as giving the world premiere: Marcel Moyse (flute), Gaston Hamelin (clarinet), Fernand Oubradous and Alexandre Rible (bassoons), Pierre Vignal and Lucien Body (trumpets) and Victor-Jean-Baptiste Hudier and Emile Lauga (trombones). If they did record it for Bernard Roux, this would be only the second recording made by Marcel Moyse, although the leading flute discographer Susan Nelson notes, ‘Moyse’s participation in this early recording has not been documented.'

I also believe that Stravinsky conducted the remake of the Finale. The side-break (in the fifth bar after fig.62 of the first, 1924 edition of the score) does not of itself indicate Stravinsky's presence, but its precise execution does: with modern audio-editing software, the sides join up cleanly and all but imperceptibly, revealing a firm and well internalized grasp of the tempo and careful avoidance of slowing down. The lack of any mention of the re-recording in the known sources has led me to wonder if another conductor stood in, or even whether the players coped without one. Both ideas are far-fetched, for various reasons - not least, as Philip Stuart convincingly argued, that the point of the exercise was not to make a perfect recording but to give Stravinsky objective feedback on his conducting, and also that the cost-conscious composer felt Bernard Roux owed him another chance at the Finale. Speaking of cost, the decision to re-record it on two very under-filled 30 cm sides seems perverse. But it was a way of minimizing the amount of music to be recorded without a mistake or a break-down, and thus ensuring the remake took as little studio time as possible. As a response to the original débâcle, it is perhaps of a piece with Stravinsky's well-known superstitiousness.

The fact that Craft did not list these numbers is, in truth, not a discographical deal-breaker. It does not mean that Stravinsky did not at some point own a copy of 801 / 802, and later discarded or lost it (there is now no sign that he retained the first three sides either). And if paperwork was all that survived into the late twentieth century, it might well have been silent on the subject: an invoice would have listed only the sides Stravinsky was required to pay for. Stravinsky also requested a pressing of the botched Finale, yet its probable matrix number (799?) was not listed by Craft.

Clearly, nothing can be proven. And if one is willing to speculate, other scenarios can be entertained. For instance, the Roman numerals VI and VII on the labels of the BnF's test pressing suggest the disc formed part of a sequence of at least seven sides. This tallies with the span of seven Roux matrices used for the Octet project, from 796 to 802 (including 799 and 800, the two not accounted for). This might make sense, if original lay-out was indeed on five sides. In such a scheme, the first movement is split over two sides, while the Finale again follows the end of the Tema con variazioni on one side. The layout also leaves a sixth side free, perhaps for Stravinsky 'to record another work, as fill-up', as Philip Stuart reported.

Side no.	Selection	Matrix no.	Diameter
I.	(i) Sinfonia (start)	796	unknown (27 cm?)
II.	(i) Sinfonia (end)	797	unknown (27 cm?)
III.	(ii) Tema con variazioni (start)	798	unknown (27 cm?)
IV.	(ii) Tema con variazioni (continuation)	unknown [799?]	unknown (27 cm?)
V.	(ii) Tema con variazioni (end) (iii) Finale (incomplete)	unknown [800?]	unknown (27 cm?)
–	Title unknown ['fill-up'; not recorded]	unknown [801?]	–

In this admittedly fanciful scenario, the session ended prematurely, with the débâcle of the Finale; the planned 'fill-up' would have been abandoned, freeing its allotted matrix number and informal side number to be reused for the re-recorded Finale, continuing the above sequence:

Side no.	Selection	Matrix no.	Diameter
VI.	(iii) Finale [remake] (start)	801	30 cm / 30 cm
VII.	(iii) Finale [remake] (end)	802	30 cm / 30 cm

Certainly, some explanation is needed for the small gap between highest side (matrix) number listed by Craft, 798, and those on the re-recorded Finale test. If matrices 799 and 800 were not used for Stravinsky's Octet, 801 and 802 must have been recorded very soon after the débâcle, effectively during the next session. Bernard Roux was a busy firm; had days or weeks elapsed, considerably higher matrix numbers would have been allotted to the re-recorded Finale.

Whatever happened exactly and when, Stravinsky would not have had to go far out of his way to record for Bernard Roux: the firm was not alone in having the use of a room in the headquarters of the piano- and harp-manufacturer Pleyel et Cie on the rue Rochechouart. Stravinsky himself had been installed there for some time: in 1921, he had contracted to produce rolls of several of his works for the Pleyela, the firm's player piano. To help him realise this project, Pleyel had given him a studio on its premises, which doubled as a sort of bachelor flat. (Like so much about the Pleyel buildings on the rue Rochechouart, the exact location of Stravinsky's studio has not been determined.) It seems unlikely that Stravinsky spent time scouring Paris trade directories to find a firm competent and available to capture him conducting the Octet; in all likelihood, Bernard Roux was already using the hall at Pleyel to make recordings, and so was just along the corridor, so to speak. In fact the recordists' presence might even have piqued Stravinsky's interest and planted the idea. One can only hope that, now that the Octet's re-recorded Finale has been traced, previously unknown or inaccessible documents will be made available and even - who knows? - sides from the earlier session will be rediscovered.

Patchy and perplexing as the sources for the Octet recording are, no documentation at all survives for a Bernard Roux test pressing one of whose labels is shown above (second image). Both labels carry no information beyond matrix numbers and voice types, suggesting the content is vocal music:

Label inscription	Diameter	System; cut	Matrix no. (in runout)
Soprano 9576	25 cm / 10 in.	Electrical; lateral	9576
Alto 9577	25 cm / 10 in.	Electrical; lateral	9577

The contents have recently been identified. Matrix 9576 contains what on first hearing appears to be a French mélodie from the first third or so of the twentieth century, sung by a soprano with piano. No match for the text can be found online. Matrix 9577 contains the same music, sung by an alto at a lower pitch, and setting a German text. This is a poem by Hugo von Hofmannsthal (1874-1929), of which the text on the other side turns out to be a French translation. The original is 'Die Stunden! wo wir auf das helle Blauen / des Meeres', the second of Hofmannsthal's four Terzinen ('Tercets') of 1895. Of two settings of these poems by composers living in France, one seemed the likelier source of the recordings on the Bernard Roux test pressing. Published in 1934, this was by the French aristocrat, author, publisher, composer and esotericist, Count René Philipon (1869-1936). A copy is held by the Bibliothèque nationale de France, from which a digitisation for private perusal was requested.

Examination of the score confirmed that Bernard Roux matrix 9577 is a recording of the second of Philipon's settings for voice and piano of three of Hofmannsthal's Terzinen. The BnF's print presents all three with German text underlay only. Its front cover carries a printed dedication, 'À Madame Hélène Suter-Moser'. The Swiss-born alto and contralto Hélène Suter-Moser (1893-1965) lived in Paris in the 1920s and '30s with her husband, a Swiss sculptor, and made one commercial recording there; she may be the singer recorded on matrix 9577. The inside back cover of the score states, 'Vertcœur 1934'; the Château de Vertcœur was the country seat Philipon had had built in 1902-04 in the Chevreuse Valley, south west of Paris. Philipon presided over a literary and musical salon both there and in the capital; a regular performer at these was the soprano Hélène Baudry, who also broadcast his mélodies from Paris radio stations. Baudry too made commercial recordings; she might be the singer of the French version of 'Die Stunden' on this disc. Its text may well have been the work of Philipon himself; his joint translation of Hofmannsthal's play Jedermann had been published in 1932 and was performed in 1934 by a student literary society in the French-speaking Swiss city of Neuchâtel. Harvard University Library holds a separate print, with bilingual German and French text underlay, of 'Die Stunden' / 'Les heures' only, which also has the colophon 'Vertcœur 1934': clearly, the French translation was an integral part of the work, although the print does not credit Philipon as translator.

Unlike the BnF's score of the complete set, the print of 'Die Stunden' / 'Les heures' states that it was not for sale (Hors commerce). This reinforces the suspicion that Philipon commissioned Bernard Roux to record his German and French settings of Hofmannsthal for his own private use, probably in 1934, the year he had his Terzinen printed. While the singers could perhaps be identified by ear, the pianist may never be, unless documentation of the project comes to light. If the Count's own playing was good enough, aristocratic decorum may nevertheless have precluded his participation in the recording, which could have been undertaken by any of the professional pianists who performed at his salons. Other pressings may survive, with labels carrying more detailed information and, possibly, a personalized design, rather than Bernard Roux's workaday test ones. Better still, Philipon may have had his other two Terzinen recorded. Even so, on its own, quite apart from its considerable intrinsic interest, this disc suggests that, more than a decade after the Stravinsky Octet set, in fact well into the last phase of its involvement in the gramophone business, Etablissements Roux was still producing private recordings, alongside its much more extensive output for commercial record labels.

Educational recordings

As stated above, the Bibliothèque nationale de France in Paris holds six Bernard Roux test discs, of which five have been digitized and four (to date) made available online. Among the latter is an educational record – an aid to learning French:

Label inscription(s)	Diameter	System; cut	Matrix nos. (in runout)
3670 3671	27 cm	Acoustical; lateral	3670 3671

As often with Roux tests, the labels disclose nothing about the content; at some later date, the hand-written classification Diction ('spoken word') was added to both. Fortunately, each side is clearly announced at the start by an adult male French-speaker:

Ecoles internationales Phonogramme de français numéro deux

Ecoles internationales Phonogramme de français numéro trois

The announcements identify Roux's client as the French partner of the International Correspondence Schools (ICS) of Scranton, Pennsylvania. The ICS was a leading provider of distance-learning materials and an early adopter of audio recording (initially on cylinders). In 1912, it launched an ambitious marketing campaign in Europe; the following year, in Paris, the Ecoles internationales began its own press campaign for 'Méthode I.C.S.' language-learning courses, ramping it up – with unfortunate timing – in 1914, when gramophone-based courses in English, German and Spanish were also advertised. Coincidentally or not, the founder of ICS was a mining engineer, as was an enthusiastic French champion of the method, who described it at length in professional journals and compared it with its main rival, Pathé's ambitious audiovisual system the Pathégraphe. The ICS had already produced French-language cylinders and an associated book in the USA, but the Ecoles Internationales commissioned new discs as well as new printed materials. The Bibliothèque nationale de France holds a large quantity of the latter but none, it seems, of the associated records.

A number of Ecoles Internationales discs for learners of Spanish have survived and are documented online. Their size is given as 25 cm, smaller than the 27 cm Roux test, although the sketchy nature of these entries means such details shoud be accepted with caution. The label images, welcome as they are, are tightly cropped, so that it is not possible to see the runout area or any matrix information, which is not repeated on the labels themselves. Those on what was presumably the first disc, with Fonograma de español N^o 1 on one side and Fonograma de español N^o 2 on the other, branded Méthode I.C.S [sic], appear to be of pre-War design. Four other discs form part of a later series with two fonogramas per side, running from Nos.15-16 to Nos.49-50; their plainer labels cannot be much later in date, as they give the pre-War Paris address of the Ecoles Internationales. For the moment, it remains unclear who recorded or pressed these examples.

The BnF's transfer of its French-language Bernard Roux test is accompanied by widely-framed label images, clearly showing the runout area and matrix numbers, made up of four stamped digits in a row. While it is not possible to state for certain that the recordings were issued by the Ecoles internationales, it seems highly likely that they were. The contents correspond closely to a contemporary description of 'Méthode I.C.S.' lessons: after the opening announcement, each test side starts with a selection of basic vocabulary, read by the same male native French speaker; continues with a simple conversation, the reader impersonating both speakers (including a woman); and ends with a reading of a short series of numbers. Of particular technological interest is the fact that the ICS method required students to record their own voices on disc and post the resulting recordings to the Ecoles Internationales, where native speakers would assess them and post back written corrections. If any words were judged to have been poorly pronounced, students were advised to re-record them after the following lesson. A dual-mode (playback and record) machine for this purpose was marketed by the Ecoles, although exactly how it captured these home recordings was not explained.

After the War, the 'Méthode I.C.S.' programme was relaunched; this time, it was also advertised to English-speakers living in France. The dual-mode gramophone was also relaunched; a surviving example held by the BnF is dated to circa 1921. This interesting venture deserves further research, not least to establish Bernard Roux's role in it, if any.

It may be that Roux also made recordings for another gramophone-based language course, devised by and marketed under the name of composer, conductor, pianist, organist, teacher, author and entrepreneur Louis-Julien Rousseau (1873-1950). The Bibliothèque nationale de France holds thirteen of these 30 cm / 12 in. vertical-cut discs, whose labels state:

Méthode dialoguée de L. Julien Rousseau […] Français allemand Franzosich [sic] deutsch […] Système unique à répétition. Copyright by L. Julien Rousseau 1922

(Linguistically, the material is somewhat amateurish, not least because the male speaker, possibly Rousseau himself, is fluent in French but not in German, while his 'système unique' seems to consist merely of repetition.) As before, the BnF's label images make it possible to see the runout area and matrix number of each side; here, the matrix numbers consist of three stamped digits in a row but rotated, to be read not from left to right but top to bottom, a detail seen on another Roux test held by the BnF (see below). Each runout area also contains an alphanumeric sequence, in outsized characters, of the form LJR 2, LJR 3 etc. – effectively face numbers; neither these nor the matrix numbers are repeated in the labels.

In addition to these records, the BnF holds a related printed item, either a marketing pamphlet or a course book. Rousseau also published a French-English course, documented in printed form only; whether it too was accompanied by records has not been established. No press advertisements or notices have been located which might throw further light on Rousseau's venture. In 1918, he had been engaged by the Compagnie française du Gramophone as an orchestrator, and before that he had been musical director for Odeon, but it seems unlikely either of these companies produced these records, as their technical quality is below the standard one would expect from these firms. This leaves open the strong possibility that the series was produced by Roux's firm, which made musical recordings with Rousseau (see below).

Experimental recordings

Also among the BnF's holdings is an early experimental electrical recording by Bernard Roux:

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
Premier disque à aiguille enregistré électriquement Le 28 Janvier 1925	27 cm	Electrical; lateral	Essai spécial N°57 BR 28/1/1925
Orchestre enregistré électriquement Le 28 Juillet 1926	27 cm	Electrical; lateral	3689

On this disc, both labels show neat, hand-written inscriptions in ink, which, unusually, appear roughly contemporary with the recorded contents. This suggests that, unlike Bernard Roux tests whose labels lack all but the barest annotations, this pressing was not a 'disque de travail' – an intermediate stage in the production of a published disc, to be discarded later – but was intended to be preserved in that form, to document the results of research and development already deemed historic.

The label of 'Special Test No.57' identifies it as the 'First electrically-recorded needle-cut disc 28 January 1925'. The content is unexpectedly light-hearted: an unnamed male speaker describes the recording session itself in jocular doggerel, egging on a pianist who punctuates the patter with short solos. The latter, addressed as 'Rousseau', was almost certainly Louis-Julien Rousseau (see above), who was born and died in the north-western Paris suburb of Taverny. To judge from the compère's teasing, Rousseau was rather attached to his birthplace:

Pour rendre le phono pratique
faut l'enregistrement électrique;
et c'est pour celà qu'entre nous
on le travaille au studios Roux.
Faut du boulot, de la constance,
car quand c'est fini on r'commence.
Et notre maestro est verni
quand i' r'gagne à l'heure Taverny!
M'sieur Rousseau! Faites-nous de la musique!
[piano solo]
Lààà... Que'qu'chose de bien et d'harmonique!
[piano solo]
Ça va... Et pour essayer ce micro,
encore un p'tit air de piano:
[piano solo]
Lààà... Maintenant, ça va bien, je l' proclame:
M'sieur Rousseau! Encore une gamme!
[piano scales]
Celui-ci va pas mal, ma foi;
jouez-nous donc l'air de l'autre fois.
[piano solo]
Mais toute peine a sa récompense:
M'sieur Rousseau, jouez un air de danse!
[piano solo]
Lààà... Enfin, voilà que'qu'chose qui plaît.
Félicitons Monsieur Ibled!
Notre patron en sera fort aise;
M'sieur Rousseau, jouez la Marseillaise!
[piano solo]

This side clearly derives from Bernard Roux's experimental electrical recording sessions (see above), known to have been held at 24, rue Rochechouart. As can be heard, the compère mentions Jean Ibled, one of the two engineers responsible, by his surname, and also alludes to 'Notre patron' ('Our boss') – surely Bernard Roux, whose initials 'BR' appear alongside the date in the runout or 'dead wax'.

The other side of this test pressing contains an upbeat march with occasional syncopation, not titled or credited on the disc label, which states, 'Orchestra recorded electrically on 28 July 1926'. Presumably this too was an experiment, in recording a larger ensemble. Conceivably, it was conducted by Rousseau, who made recordings for commercial labels with which Roux seems to have collaborated. The matrix number, 3689, is only a little higher than those on the Ecoles internationales tests discussed above (it is also stamped in the runout in same, conventional way), which presumably means that the Phonogrammes français were recorded not much earlier. Their constricted sound suggests acoustical recording, which remained in use until suprisingly late in places.

Unidentified recordings

The great majority of Bernard Roux test pressings documented for this page contain music, and it seems likely that music accounted for most of the firm's production of recordings. Such is the seeming plethora of information about recorded music available on the internet today that it may come as a surprise to learn that the output of most labels thought to have been Roux clients has barely been documented, if at all. (A particular difficulty is that very many shellac discs in the huge collection held by the Bibliothèque nationale de France are catalogued without matrix numbers.) At the same time, much of the repertoire and many of the artists on these discs were in their day little known outside France, and are now familiar only to collectors, again mainly French. Without the relevant expertise or access to physical collections, it can be difficult to identify the contents of Roux tests or their intended clients, and to determine whether they were experimental recordings, business-to-business samples, private commissions or for commercial issue. This section presents a few such cases; perhaps you can help solve them?

One Roux test held by the BnF perplexingly couples sides with very different matrix numbers, one containing a Hawaiian steel guitar solo (with backing banjo or ukulele) and the other a short sacred choral work:

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
E 584 Guitare Hawaienne [added later?] B	27 cm	Undetermined; lateral	E 584
3586 AB Chœur [illegible] av. Grand Orgue [added later?]	27 cm	Electrical; lateral	3586

As indicated (clumsily, with apologies) in the table, the labels appear to have been annotated at different times. As often, matrix numbers are in grey pencil and were probably entered first, whereas titles are in blue pencil and were perhaps added later (they look very like those on the language-learning test pressing discussed above). Individually, each of these sides might have been issued commercially by, say, Inovat, or other, similar labels with very varied outputs. But would they have made for a viable commercial product as coupled on this test, with very different music presumably aimed at different audiences?

The instrumental side has not been identified. The Hawaiian guitar was popular in France in the 1920s and '30s: labels large and small, including Inovat, issued recordings by leading players such as the Italian-born Gino Bordin (1899-1977) and Edouard Jacovacci (1875-1939). Purely as a recording, this has an odd quality: the BnF's catalogue does not specify the system, which sounds restricted enough to be acoustical – a subjective judgement, admittedly.

The choral work is identified in the BnF catalogue as a sacred canticle by the French composer and musicologist Charles Bordes, setting a prayer to the Virgin Mary by the Benedictine Dom Jean Parisot (1861-1923). The (librarian's?) label annotation, describing the accompanying instrument as a 'grand organ', is a perhaps forgiveable exaggeration of what appears to be a harmonium. This side does sound electrical and might represent another experimental ensemble recording; its matrix number is relatively close to that of the mid-1926 orchestral side discussed above, except that this number appears to be suffixed (on the label only) AB. That is more commonly seen as a monogram (in various configurations) in the 'dead wax' or runout of many commercial records produced by Roux for a wide variety of labels. Conspicuously absent from the latter, though, are the French labels best known for issuing mainly or only sacred music, namely Lumen (c. 1933-1970), La Musique au Vatican (commonly known as SEMS) (c.1938-1950s), and Studio SM (1946-2007), while Voix chrétiennes (c. 1936-1939) can also be eliminated as an outlet for this side on the grounds of probable date. Perhaps this test was a business-to-business sample, a demonstration of what Roux's engineers could do with such different material (if so, it must be said, it is none too convincing).

Another Bernard Roux test disc held by the BnF contains recordings identified on both labels simply as 'Jewish song':

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
4120 Chant Hébreu	25 cm	Undetermined; lateral	4120
4121 Chant Hébreu	25 cm	Undetermined; lateral	4121

A male voice announces each side in Turkish and plays a brief prelude on the oud or ‘ūd (the instrument is specified by the BnF but not the language), and then sings unaccompanied for the rest of the side. Despite considerable efforts, it has not yet been possible to identify definitively the performer (whose speech was recorded somewhat indistinctly) or the exact selections on this disc, but the latter are certainly part of the repertoires of Sephardic Jews in Ottoman and post-Ottoman Turkey. Rivka Havassy, a scholar of Sephardic music, has tentatively identified the song on matrix 4120 as a seliḥa, BeZokhri ‘Al Mishkavi (בְּזָכְרִי עַל מִשְׁכָּבִי), and the singer as the renowned Hayim Efendi or Haim Effendi (1853-1938), who had previously recorded it in 1913 for the Turkish label Orfeon. His birth name was Haim Behar Menahem, and it is interesting to note that a M. and Mme. Haïm Behar travelled on the SS Lamartine from Istanbul to Marseilles in February 1926. That is within the range of possible recording dates for this disc, although before jumping to conclusions one must establish whether the singer Haim Effendi was one of the passengers named, and whether he or the couple travelled on to Paris.

The reason for the second proviso is that no evidence has yet been found to indicate that Bernard Roux's firm ever made recordings outside Paris, let alone abroad – which, in turn, begs the questions of this disc's intended purpose and possible client. Both multinational concerns and local French labels issued Jewish music from the Near East and North Africa commercially; while companies such as Columbia could record in Istanbul and elsewhere, smaller ones such as Perfectaphone recorded eastern musicians in Paris. In 1927, during a visit to the city, celebrated Syrian violinist Sāmī al-Shawwā (1899-1965) set down several sides for the company, some of which appear in an undated Perfectaphone catalogue listing records of Greek and Turkish music. (Note that there is no evidence that Bernard Roux made these or any other recordings for Perfectaphone.) These repertoires were also the focus of academic recordists; but in France that activity was dominated, if not monopolised, by Pathé, and again there is no evidence that Roux's firm was involved in it. On balance, it seems more likely this was a commercial production – perhaps for an as yet undocumented label, perhaps unissued – or a private commission, which seems less probable, when Haim Effendi made so many commercial recordings.

Of the two Roux tests whose labels are illustrated above, one is especially tantalising:

Label inscription(s)	Diameter	System; cut	Matrix nos. (in runout)
4502 Prelude en Ut mineur (Bach) 1e Pie	25 cm	Electrical; lateral	4502
4503 2e Partie (Fugue)	25 cm	Electrical; lateral	4503

An unusual feature of this disc is that annotations on both labels, in pencil and in the same handwriting as the matrix numbers (suggesting they are contemporary?), clearly identify the music as the opening movement of J.S. Bach's Suite for solo cello in c minor (BWV 1011). It is performed complete and interpreted confidently, thoughtfully and distinctively, surely by a professional cellist – who is not named on the labels, frustratingly. The recording itself is somewhat out of the ordinary. At this date, presumed to be the late 1920s or early 1930s, this was uncommon repertoire for small companies in France or indeed anywhere; on disc, cellists were usually limited to salon fare such as Handel's so-called 'Largo', recorded by Jean Vaugeois for Perfectaphone, or Gounod's 'Ave Maria' (then commonly credited to Bach), recorded by, among others, Victor Pascal for Gramophone. Meanwhile, not a single movement of Bach's solo suites was recorded in France before April 1930, not even by the leading multinationals. Movements were recorded in Britain and the United States by Pablo Casals (Columbia), Beatrice Harrison and Guilhermina Suggia (Gramophone Co.), while Casals' famous complete cycle (HMV) lay some years in the future (Casals also recorded arrangements for cello of other items by Bach). Might this be the earliest French recording of this music?

The only commercial label currently known to have been a client of Roux, and which might have issued such recherché repertoire, was Optima, registered in January 1930 by a Paris wholesaler of musical instruments (see Other labels, below); but this test's matrix numbers are much lower than those on the admittedly few documented Optima issues. Another candidate might be Electrovox, for which Bernard Roux is suspected though not confirmed to have produced recordings (see also below); Electrovox issued one rather unusual disc of Baroque keyboard music, whose matrix numbers are close to those on this test, and whose physical make-up is similar. Conceivably, as with the disc of Turkish-Jewish music just discussed, the client was an undocumented label which awaits discovery by a specialist such as Jerome Moncada, or one which is known but poorly documented. Finally, another possibility, discussed below, is that this recording was not for retail sale but produced for one or more radio stations.

It is easier to identify the intended commercial client of a test pressing when its matrix number(s) are not only close to a range known from published discs, or even within it, but also have in common with the latter a distinctive configuration or mark. One such detail is the AB monogram, mentioned above as occurring on many published commercial issues, as well as several Bernard Roux test discs. One example carries repertoire even more specialised than Bach, recorded by an unidentified soprano with piano:

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
4810 Lasciatemi morire Monteverde	25 cm	Electrical; lateral	4810 AB
4811 Se tu m'ami Pergolesi	25 cm	Electrical; lateral	4811 AB

Complete with AB monogram, these matrix numbers are one pair higher than the range known from issues on the Azurephone label, all of which also show the monogram and were probably produced by Roux; Roux test pressings of two Azurephone discs survive (see above). Accordingly, this coupling has been tentatively attributed to the label. The marque was registered in April 1929, as it happens by an entrepreneur who, like Roux, had started off in rubber and allied products and by 1928 had formed a retail business selling musical instruments, radios and even vacuum cleaners. Azurephone discs may have been for sale only direct from the business's one retail address, and the catalogue may not have been very extensive. What little of it is known consists solely of jazz and dance music, including tangos, making Monteverdi and (as then attributed) Pergolesi seem unlikely fare for this publisher. Again, the recordings might have been made for another, unknown label or for Optima (again, though, the matrix numbers are rather distant); or this might have been a private commission. Both selections are sung in versions with piano by Alessandro Parisotti (1853-1913), who most probably himself composed the 'Arietta' long attributed to Pergolesi (and arranged by Stravinsky in his ballet Pulcinella); both were popular in France, sung by artists well-known and not so well known. If the singer on this test made any other records, it may be possible to recognize her voice from the distinctive Italian erre moscia or erre francese, although this is not unvarying – she could trill her /r/ – while other features of her pronunciation also leave some doubt as to her native language.

Another monogram associated with matrix numbers of some Roux productions is MB (combined into a single digraph), seen on issues by Angelus, Champion and other labels. It may have been present on a sketchily documented Roux test of classical vocal music sold in 2021:

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
7131	25 cm	Electrical; lateral	7131 MB(?)
7132(?)	25 cm	Electrical; lateral	7132(?)

The sole image archived from this auction consists of the later type of Bernard Roux test label, with the first matrix number written in pencil; a suffix MB appears to be faintly visible in the runout. Perhaps the original listing contained more information; otherwise, it is not clear how the seller identified the artist on the first side as the French tenor Mario Podesta (born Henri Lemoine, 1892-1977) with, presumably, an unidentified soprano. The music is the so-called 'bridal scene duet' ('Das süße Lied verhallt') in Act III, scene ii of Wagner's Lohengrin, given in French as 'Déjà se perdent leurs voix'. The coupling, described as 'Unknown Song by Uncredited Soprano with Piano', was possibly recorded by the same soprano as the duet. Podesta made only a small number of commercial records (and none of Wagner), of which this is not one, and he is remembered now mainly as the teacher of the soprano Mado Robin (1918-1960). Robin began vocal studies in her teens and sang in public as early as 1934, but Wagner was never in her mainly Franco-Italian repertoire, and in any case this test was probably recorded earlier. Recordings by operatic tenors and sopranos were issued by labels such as Champion; this disc could have been made for the commercial market.

For once, another Bernard Roux test of classical music, also sold online in 2021, is associated with a considerable amount of information:

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
Wurmser [other illegible names?] Chanson Indoue Rimsky-Korsakoff (trio) 25042 Juin 1931	25 cm	Electrical; lateral	S 25042
L'heure exquise [? (no label image available)]	25 cm	Electrical; lateral	S 25040

Not all of this information is currently retrievable. The one (later-type) Roux label imaged in the archived listing is too reduced to decipher the two names below 'Wurmser', but the date, and much else, is clearly visible; as often, these annotations may have been added after the pressing left the Roux factory with just the customary pencilled matrix numbers. The original listing attributed the first side to a Wurmser Trio: certainly, the composer, pianist, conductor and teacher Lucien Wurmser did lead trios named after him, in 1901, 1907, 1910, 1934, and from 1939 to 1944, the last of which made two commercial recordings in 1942. But these formations played mainly original classical chamber works, rather than salon arrangements of items such as the 'Song of the Indian Guest' from Rimsky Korsakov's opera Sadko. It is tempting to read the second name on the label as Vaugeois, the cellist just mentioned, if optimistic: Vaugeois and Wurmser did perform at one of a series of recitals organized by a singer, Yvonne Besneux-Gautheron, but not alongside each other. The seller did not identify the artists on the coupled side, a trio of soprano, violin and piano, presumably because their names were not inscribed on the other label, but did supply a vital detail, the matrix prefix S (not shown on the label). This has allowed the probable publisher of these recordings to be pinpointed with some confidence (see below).

A final example is included in this section – despite the fact that its contents have also been identified – for two reasons: the identification took several months, and not all the problems posed have been solved. The disc in question is the only Bernard Roux test held by the British Library Sound Archive in London:

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
[under paper label] 1106 – La Favorite (Fantaisie) – Ultima –	25 cm	Acoustical; vertical	1106
Werker 17/2/30 Paris. PK 200	25 cm	Electrical; lateral	–

Both sides carry an early Bernard Roux label. The label for matrix 1106 has no inscription in French, only a pencilled note in English, 'Phono Cut', drawing attention to the vertical-cut recording. Originally, no inscription was needed, as the paper label only partly covers an engraved (or 'etched') label, visible underneath and clearly legible, a type used by several French companies before World War I. The music is a fantasy – one of many – on themes from Donizetti's opera La Favorite, played by a military band. No arranger or performers are credited. Ultima, as noted above, was a marque first registered in late 1908; within weeks, new releases were briefly advertised in the French press, after which Ultima disappeared from sources currently accessible online. In late 1910, the company's assets were put up for sale, and in November 1911 the marque was sold and re-registered. The same label design was registered on both occasions; it corresponds exactly to the markings on the British Library's pressing, identifying it as an Ultima master which was very probably issued (no published copy has yet been found). The recording date is therefore no later than 1911; whether Roux was pressing discs for Ultima at that date is not known.

The other label on this test, matrix PK 200, does carry an inscription, in ink: unusually, it includes a date and general location, as well as an enigmatic word, read here as 'Werker'. This has no known connection with the music, which is another orchestral selection from an opera, the Overture of Si j’étais roi by Adolphe Adam. 'Werker' might therefore refer to the performers – perhaps one of several Belgian socialist bands called 'De Werker'? – except that no recordings by them have been traced. Matrices with suffix PK, of similar vintage and later, are known from discs issued by Paris-based Perfectaphone. The BnF holds 14 such discs. A recent discography of Perfectaphone, admittedly a work in progress (entries for unseen discs lack matrix numbers), lists many more PK matrices but none match this test pressing. As it happens, this line of enquiry has proved to be a red herring, and the client for this test has now been identified as a completely different company (see below), although the annotation 'Werker' remains unexplained.

Once again, it should not be assumed that this test was intended for issue as coupled, any more than the BnF's experimental electrical Roux test (see above). Rather, it may have been a sample of what Roux could produce for a prospective client. Even so, why couple a relatively antique, vertical-cut acoustical side with one recorded electrically nearly two decades later? The answer may lie in the recycling of Ultima masters, on a variety of labels, after World War I and well into the 1920s. Nearly sixty years ago a discography of vertical-cut recordings noted reissues of Ultima masters on Arya, Corona, Opéra (possibly) and Perfectaphone; to those labels can be added Apollon, Lutétia and, probably, others. On many discs, original Ultima engraved labels are visible under later paper labels, as on the British Library's Roux test. Ultima matrix 1106 may have been recycled on Perfectaphone 515, a vertical-cut disc held by the Bibliothèque nationale, whose catalogue shows the identical matrix and title. Whereas the engraved Ultima label credits no performers, the Perfectaphone issue names the Garde Républicaine, which also contributes the coupling, another operatic fantasy, from Bizet's Carmen (an original Ultima issue of that side, matrix 1236, is also known). Perfectaphone issued other recordings by the Garde Républicaine band, on e.g. 505, 530, 536 and 559: of these, 530 matches a known Ultima original in size and content; the matrix numbers also match but on the Perfectaphone pressing are prefixed 22-, presumably denoting the issued disc size. This prefix is seen on several Perfectaphone discs, as are other prefixes seemingly denoting disc sizes, notably 29-. In all likelihood, many similar matches remain to be documented.

One perplexing detail is that the BL's pressing is 25 cm in diameter, whereas Perfectaphone 515 is 22 cm, as were most Ultima discs, though not all. If it turns out that Perfectaphone 515 is the same recording as Ultima 1106, was the latter decreased in size by Roux's electrical duplicating technique, for reissue by Perfectaphone? (Roux's duplicator cannot have transferred markings such as the Ultima engraved label from the source to the target medium; if these are markings are present on Perfectaphone 515, the duplicator cannot have been used.) No less intriguing is the question of how Roux obtained Ultima masters. A clue emerges from a report of the 1911 sale: the seller (not Ultima's original owner) was one François Biat, whose business was located at 146, rue du Chemin Vert, next to the Ebonitine offices. So perhaps Roux had indeed pressed Ultima discs for Biat; their relationship, if any, certainly merits investigation. Meanwhile, a probable provenance for the BL's test pressing has now been established, which in turn raises further questions (see below).

Jazz and dance music

With European jazz and dance music we reach surer ground. This field has been well documented by discographers and collectors, so that commercially published discs can often be reliably matched up with Bernard Roux test pressings. This, rather than detailed discographical presentation, is the purpose of this section, starting with one of two Roux tests auctioned online in early 2023:

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
7330	25 cm	Electrical; lateral	7330 MB
7331	25 cm	Electrical; lateral	7331 MB

No titles or artists were inscribed on the (later-type) Bernard Roux labels, only matrix numbers, but the contents were identified by the seller, allowing this test to be matched with the following 25 cm disc, reportedly recorded in about April 1933:

Selection	Artist(s)	Matrix no.	Label cat. no.
Great Fun (Fox-trot) Julien Porret	Orchestre-Jazz Hot Melodic Band	J.P. 161 7330	Champion 1403
My Love (Fox-trot) Géo Linat	Orchestre-Jazz Hot Melodic Band	G.L. 31 7331	Champion 1403

Géo Linat was one of several pseudonyms of Julien Porret. The alphanumeric codes preceding the matrix numbers correspond to printed sheet music.

The other Roux test auctioned in 2023 coupled sides which must have been recorded at the same session or soon afterwards:

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
7334	25 cm	Electrical; lateral	7334 MB
7335	25 cm	Electrical; lateral	7335 MB

The titles were identified by the seller as fox trots by Julien Porret, both performed, as on the other Roux test listed above, by the Hot Melodic Band. These sides match the following 25 cm disc closely but not exactly:

Selection	Artist(s)	Matrix no.	Label cat. no.
Pupazzi	Hot Melodic Band	N.S. 134	Julien Porret 1340
A Brivélé	Hot Melodic Band	N.S. 141	Julien Porret 1340

N.S. stands for Nat Sing, another of Julien Porret's pseudonyms, and the alpha-numeric sequences transcribed as matrix numbers are again codes corresponding to printed sheet music; it seems the original matrix numbers were not shown on the published disc. Another discrepancy is that the session(s) for the latter have been dated to 1936, some three years before the Roux tests; yet a coupling of Pupazzi and A Brivele was available for broadcasting in April 1935. That may be explained as an instance of a little-studied practice, namely the distribution of commercial recordings on non-commercial or pre-release pressings for use in broadcasts, theatres, cinemas and similar outlets. Jerome Moncada's Label Gallery Française covers a host of such 'production' labels, several with the word 'Production' in their names (e.g. Production Maurice Brun, Production Louis Closset, Production Yatove etc.). Julien Porret's own label clearly operated on this basis, distributing discs in sleeves with detailed instructions for logging plays and reporting them to the rights agency SACEM; the sleeves also promised more free discs in return for compliance as well as replacements for worn discs, and asked users to help publicize Porret's productions, listed on the other side (including both the above discs).

Later, seemingly after World War II, Porret recycled some of his recordings on a new label, Constellation, coupling each number with a newly-recorded instructional 'choreographic divertissement' based on it: Pupazzi reappeared on Constellation 16 (these discs were pressed by Pathé, not by Roux, and catalogue entries do not always show full matrix information, hindering precise identification). Based on the above, a number of Julien Porret recordings, issued on his own label, Champion and others, can be considered as possible Bernard Roux productions.

Other documented Bernard Roux tests of this repertoire include a coupling of sides recorded by Berson's European Ramblers, with vocals by Jean Farnèse:

[Selection]	Diameter	System; cut	Matrix no. (in runout)
[Nana (Rumba with vocal)]	25 cm	Electrical; lateral	M 139 A
[Un deux… toute la compagnie (Foxtrot with vocal)]	25 cm	Electrical; lateral	M 142 A

These match the following sides:

Selection	Artist(s)	Matrix no.	Label cat. no.
Nana (La Marchande d'Ananas)	Le Jazz Berson's European Ramblers, Jean Farnèse (vocals)	M 139 A	Mag-Nis 527

Selection	Artist(s)	Matrix no.	Label cat. no.
Un, deux… Toute la Compagnie	Le Jazz Berson's European Ramblers, Jean Farnèse (vocals)	M 142 A	Mag-Nis 528

Mag-nis was the record label of the department store chain Les magasins réunis, which issued a considerable number of such recordings. These were low-priced discs, made of cheap materials; one commentator has deplored their poor quality, although it must be remembered they were not intended to be heirlooms capable of withstanding a century's wear and tear. The recordings stand up well, as listeners today can judge for themselves: transfers of both sides of Mag-Nis 528 have been posted online, alongside images showing both labels pasted over with half-labels of another budget marque, Franceco. Roux certainly pressed discs for other grands magasins (see below) and probably also produced recordings for these house labels.

Another Bernard Roux test disc proves to have had more than one known commercial issue:

[Selection]	Diameter	System; cut	Matrix no. (in runout)
[That's my weakness now (Foxtrot with vocal)]	25 cm	Electrical; lateral	4748 AB
[Virginia (Foxtrot with vocal)]	25 cm	Electrical; lateral	4749 AB

These were released by Azurephone on neighbouring issues:

Selection	Artist(s)	Matrix no.	Label cat. no.
That's my weakness now	Jack Hamilton and His Entertainers, Maceo Jefferson, Rudy Bayfield Evans (vocals)	4748 AB	Azurephone 1008

Selection	Artist(s)	Matrix no.	Label cat. no.
Virginia	Jack Hamilton and His Entertainers, Maceo Jefferson, Rudy Bayfield Evans (vocals)	4749 AB	Azurephone 1009

(Virginia is a song by George Gershwin from the 1924 musical Sweet Little Devil, book by B.G. DeSylva.)

One of these sides was reportedly also issued on a novel, flexible plastic record, coupled with a side from another Azurephone disc:

Selection	Artist(s)	Matrix no.	Label cat. no.
That's my weakness now	[as above]	unknown	Discolor 12
A room with a view	[as above]	unknown	Discolor 12

The Discolor marque was registered in March 1930; not stated in the registration is the fact that the label was launched by the film company Gaumont Franco-Film Aubert S.A., revealed in contemporary newspaper reports. One writer noted that Discolor's lightweight, unbreakable discs were particularly suitable for sending by post to France's colonies.

Another artist whose recordings were issued on Discolor, among other labels, was the composer, band-leader and bandoneon-player Auguste-Jean Pesenti (1902-1952). Did his band – or that of his brother René Pesenti (1898-1976) – record two tangos coupled on a Bernard Roux test held by the University of California, Santa Barbara?

Label inscription(s)	Diameter	System; cut	Matrix no. (in runout)
Don Goyo (Bernstein)	25 cm	Electrical(?); lateral(?)	4318
El matrero (Soler)	25 cm	Electrical(?); lateral(?)	4319

Or was the band Jose Soler's? The lower matrix number above is only three away from that on another French disc of a tango by Soler, issued by Electrovox:

Selection	Artist(s)	Matrix no.	Label cat. no.
Decadencia (Tango chanté) Jose Soler	M. Amato	4315(?)	Electrovox 4315(?) / (?)

(The disc has no catalogue number; the matrix number is a tentative reading, yet to be confirmed, from Jerome Moncada's photograph of a very low-contrast label; the other side's contents and matrix number have not been made available.) The singer could be Oscar Amato, who recorded with René Pesenti's and Jose Soler's bands and wrote tango lyrics himself, or, less probably Nicolas Amato; unfortunately, the band is not credited.

Promotional recordings

In the Bilbiothèque nationale's collection, not fully catalogued or digitised, is a Bernard Roux test pressing of an 'advertising record':

Label inscription	Diameter	System; cut	Matrix no. (in runout?)
Échantillon invendable	25 cm	Unknown; unknown	8510

The words 'Échantillon invendable' suggest that this disc carries the later design of Bernard Roux label, incorporating this legend (see above). Until it is digitised, nothing more can be said about the content or client. It may in fact be double-sided; and the matrix number(s) may have a prefix or suffix. The number shown suggests a date in the 1930s.

More details are available for another advertising disc, described by a discographer as 'publié par Les Editions [sic; probably Établissements] Bernard Roux, 142 rue du Chemin-Vert, Paris. Echantillon invendable' – again, copy clearly from the later design of label, suggesting this was a test rather than a finished pressing. It is unclear whether the selections were also transcribed by the writer from the labels, or were simply summaries of what could be heard on both sides, namely 'Advertisement[s] for white [wine] of the Magasins du Louvre', a prominent, long-established Parisian department store. The singer, also on both sides, was Marcel Dumont (whose name will crop up again below), with an unnamed orchestra. The matrix numbers were 6120 AB / 6121 AB (see also below), the date estimated at January 1931; the size, not given, was probably 25 cm. Most interesting is the observation that the advertisements included an announcement, 'Don't leave without seeing our next film', and so were to be played in cinemas – presumably, in a programme of shorts. The date places this disc before Bernard Roux's relationship with another company producing advertising discs, discussed separately below. They were probably jointly responsible for a remarkable disc promoting the Ondes Martenot and produced for the 1937 Paris Expo, again discussed below.

Even earlier, in the 1920s, Bernard Roux may have started producing promotional records for the chocolat manufacturer and retailer Ibled. As mentioned above, Roux himself was related by marriage to the Ibled family, and Jean Ibled was one of the engineers who developed electrical recording for Roux. The family's Pas-de-Calais branch had founded Chocolat Ibled a century earlier; now defunct, the firm is still remembered for its decorative chocolate boxes, posters and other printed items. At an unknown date, Ibled's marketing expanded to encompass discs and even record players. The earliest discs boasted labels with an eye-catching design, echoing the exuberance of Ibled's fancy boxes; they are also somewhat reminiscent of early Perfecta and Perfectaphone labels. The discs' contents were not promotional but consisted of popular vocal and band music, including numbers performed by the Garde républicaine band. Some masters were almost certainly drawn from the catalogues of commercial labels, while others may have been newly recorded by Roux. A later label design was much plainer and more modern-looking, with no visible link to Chocolat Ibled beyond the brand name; this label is very similar to those of other marques which may have also been supplied with recordings by Bernard Roux. Some of these later Ibled records were conducted by Roux's collaborator Louis-Julien Rousseau, which reinforces the impression that Roux was involved in their production. Unfortunately, a shortage of hard discographical data means that, for now, nothing more definite can be about this relationship.

Demonstration recordings

Between 1931 and 1936, during the development of a new electronic sound equalizing system, Bernard Roux engineers made recordings designed to demonstrate the equipment's potential uses. No copies of these demonstration discs are known to have survived; they are mentioned here for completeness' sake. For a tentative list, see below.

Edison Bell

Previous sections have examined Bernard Roux test discs and attempted to identify their contents and the clients and/or purposes they were produced for, usually without the help of documentary sources. This section is devoted to an important partnership between Roux and a high-profile international client, for which there is some documentary evidence, as well as one test pressing.

In September 1928, the business pages of British newspapers published a prospectus for shares in a new company, Edison Bell (International) Limited. An offshoot of Britain's well-established Edison Bell Ltd., it was being formed to produce and market new and existing recordings around the world. Listing operational arrangements already in place, the prospectus announced an agreement with Etablissements Bernard Roux,

'whereby the latter will press a minimum of 500,000 records per annum during a period of 3 years [which] should place [Edison Bell (International)] in an exceptionally favourable position in France and its colonies. […] the negotiations already concluded and now in progress should result in an aggregate sale from the home and foreign factories of 7,000,000 Edison Bell records of “Radio” 8 inch and other brands during the first year of full operation, and they consider that by means of the existing factories at Zagreb and Milan, and the contracts for the production of records by Edison Bell Ltd. and Etablissements Bernard Roux, the Company should at once be able to supply the anticipated demand.'

In addition, Bernard Roux was to be appointed the new company's 'Local Adviser' in France (one of six in its European territories). The French business press added that an Edison Bell recording studio would be set up locally, and that the London share offering was fully subscribed.

Today, the phrase 'the production of records by […] Etablissements Bernard Roux' might be taken to mean that Roux's role would include recording as well as pressing of discs. But, as the prospectus explained, Edison Bell's successful lines, including the Radio series, were already

'produced under the electrical process known as the Voigt process licensed for use by Edison Bell, Ltd., under the terms of Contract No.4 below, and considered by that company equal, if not superior, to any known method of electrical recording.

Those terms, also laid out in the prospectus, stated that the process (developed by the engineer Paul Voigt) was licensed until mid-1942. Clearly, no other recording process, including Bernard Roux's, would be required. Still, in his role as 'Local Adviser', Roux's experience of the French market could surely be useful in other areas, such as A&R (artists and repertoire). As the prospectus specified, Edison Bell International would concentrate on producing low-priced small-diameter discs, mainly the Radio series, whose diminutive labels and smaller-pitch grooves allowed a 20 cm side to accommodate as much programme as a conventional 25 cm one (this was Edison Bell's answer to Vocalion's 20 cm Broadcast series, launched in France at about the same time). Bernard Roux was well versed in the small-diameter disc business, and so was ideally placed not only to press Edison Bell discs but also to help build its French catalogue (création de répertoires).

It is tempting to see Roux's fingerprints on Edison Bell's early 'production of records' in Paris, which seems to have got under way as soon as late 1928. New recordings were already being reviewed in French newspapers in early 1929 and radio and print marketing campaigns were taking shape. Among the earliest reviewers of French Edison Bell releases was a fledgling record magazine, which in February 1929 recommended a clutch of Radio discs of dances and numbers from popular shows such as Virginia and Show Boat. It also disclosed a rare titbit of behind-the-scenes information: Edison Bell's latest orchestral Radio discs (of which one was singled out for mention) had been recorded in the new Salle Pleyel. On 19 July 1928, less than a year after its much-fêted opening, the long-awaited new hall had been destroyed by fire. Remarkably, it was rebuilt within four and a half months and reopened on 30 November 1928, so Edison Bell could have held sessions there in December 1928 or even January 1929. The review also revealed that the performers were '44 musiciens de l’Opéra et de la Garde Républicaine' (credited on the disc's labels as 'Edison Bell Symphony Orchestra, Paris'). Given the venue and conductor, one wonders if this and three companion discs were the brainchildren of Bernard Roux, whose collaboration with Julien Rousseau went back several years:

Selection(s)	Matrix no(s).	Diameter	Catalogue no.
Delibes Coppélia ballet (excerpts unidentified)	88284-? / 88285-?	20 cm / 8 in.	F 87
Suppe Poète et paysan ouverture	88286-2 / 88287-1	20 cm / 8 in.	F 88
Massenet Manon (selections unidentified)	88288-1 / 88289-1	20 cm / 8 in.	F 89
Waldteufel Estudiantina waltz	88291-1	20 cm / 8 in.	F 90
Strauss Le beau Danube bleu waltz	88292-1	20 cm / 8 in.	F 90

(The last disc in this batch was also issued in Britain, on Edison Bell Radio 1367, with the legend 'Recorded in Paris' on the labels.)

Rousseau soon recorded six more discs for Edison Bell, whose matrix numbers suggest slightly later sessions; it is probable, though not certain, that these were also held in the Salle Pleyel:

Selection(s)	Matrix no(s).	Diameter	Catalogue no.
Lecocq La Fille de Madame Angot fantasy	88300-? / 88301-?	20 cm / 8 in.	F 94
Planquette Les Cloches de Corneville fantasy	88302-? / 88303-?	20 cm / 8 in.	F 95
Bizet Carmen (selections unidentified)	88413-1(?) / 88414-1(?)	20 cm / 8 in.	F 126
Puccini La Bohème (selections unidentified)	n/a / n/a	20 cm / 8 in.	F 127
Delibes Sylvia (selections unidentified)	n/a / n/a	20 cm / 8 in.	F 128
Puccini Tosca (selections unidentified)	88411-1 / 88412-2	20 cm / 8 in.	932

(The last disc in this batch appears to have been issued in France with a British-style catalogue number.)

After the sessions partially documented above, it seems Rousseau's services were dispensed with. From September 1929, Edison Bell France was advertising orchestral recordings conducted by Joseph Sieulle (1883-1937); and in the spring of 1931 the conductor Georges Bailly (1891-1940) was added to the label's roster. Tempting as it might be to take this as a sign of the firm's dissatisfaction with Rousseau's recordings, it should be noted that six of his discs were also issued in Britain.

Whether or not Bernard Roux arranged the recordings listed above, he was certainly involved in the new venture. Neither he nor any of his known colleagues were named as officers of the new multinational, yet in 1929 Roux started using the name Edison Bell in classified advertisements offering positions for clerical staff at 142, rue du Chemin Vert. Some months later, a new company, Edison Bell France, was formed in Paris. Apart from the obvious advantages of managing such an ambitious enterprise locally, was the French branch established partly also to exercise tighter control over the brand? Not that there is any sign that Roux's relationship with Edison Bell changed; in fact, a rare item of dated evidence suggests it was still active in 1930. It was proposed above that the Bernard Roux test pressing held by the British Library in London might have been produced for Perfectaphone. But examination of the French Edison Bell Radio catalogue reveals that the Roux test must have been pressed for Edison Bell. The side labelled 'Werker 17/2/30 Paris. PK 200', contains a recording of the overture of Si j’étais roi by Adolphe Adam. In October 1930, Edison Bell France issued a recording of this overture, conducted by Georges Bailly, on Radio F 515; its matrices are PK 200 / PK 201. The name 'Werker' on the label of the test pressing may refer to the performers, as hypothesized above, credited on the issued disc as the 'Orchestre symphonique Edison Bell', following the company's standard practice. (The apparent similarity of these matrix numbers to those seen on Perfectaphone discs is a coincidence and a trap; in Perfectaphone matrices, PK is a suffix, while on Edison Bell discs it is a prefix, one of a series denoting European and Asian cities in which Edison Bell's international branches made recordings.) The BL's test helps to date the change from British-style matrix numbers, seen in the tables above, to the PK- (and PR-) prefixed numbers seen on later Edison Bell France issues, probably to some time in 1929. Presumably, this change went hand in hand with the transfer of production from the home company to Edison Bell France.

What remains unclear is why Roux pressed a 20 cm Edison Bell master on a 25 cm test, not to mention the Ultima master 1106 on the other side. But that side too yields an unexpected link with Edison Bell. As noted above, the Bernard Roux label pasted over the original engraved Ultima label carries a pencilled note in English. The same annotation, 'Phono Cut', in the same handwriting, is seen on an undated, unbranded 6 in. / 15 cm test pressing included in the latest Edison Bell discography. The provenance of this test and its inclusion in the discography are not explained by the authors, while the annotations on both discs could of course have been added later by a collector. Nevertheless, the possibility remains that the British Library's Bernard Roux 10 in. test pressing and the unbranded 6 in. (Marathon?) test were both annotated by the same Edison Bell employee. It is striking, too, that these two test discs both contain early, vertical-cut matrices, seemingly altered in diameter. Was Edison Bell perhaps evaluating Bernard Roux's electrical copying system (see above)?

In general, the story of Edison Bell France largely remains to be written. The company receives a single, brief mention in the extensive account to which the discography is appended. Another lacuna concerns Edison Bell's supposed studio in Paris; it is not known when or indeed if this was set up, as had been reported at the time of the share offer; no further mention of such a studio or studios or other locations has been found. Nor have the French branch's later activity and ultimate fate been chronicled. From early 1932, the home company, heavily indebted, underwent a protracted period of reconstruction, receivership and, ultimately, sale; in May 1932, the Radio label was discontinued in Britain. Edison Bell International was not immediately affected and indeed the French branch obtained exclusive distribution rights for Decca in France in mid-1933, and took over French Crystalate in mid-1934. But in December 1935 it was declared bankrupt and restructured, before its final demise at an unknown date. This story needs to be completed: quite apart from its intrinsic interest, it might illuminate the financial difficulties Bernard Roux experienced in the later 1930s, and possibly the firm's exit from the disc business in favour of electronics.

Le Disque de France

In 1925, a series of advertisements began to appear in French regional newspapers. They invited readers to solve a very simple verbal puzzle and receive a 'splendid gift' (superbe cadeau) by sending their solution in a stamped, self-addressed envelope to an address in Paris's sixth arrondissement. The advertiser, presumably intent on harvesting readers' personal details for marketing purposes, was called Primes Sphinx, and it continued its campaign in the provincial press for month after month. The 'gift' was not specified; it may or may not have been gramophone-related.

In late February 1927, Primes Sphinx moved its address from the sixth arrondissement to a new location in the fifteenth – literally: the street had only recently been created. Opened in 1924, the rue des Quatre-Frères-Peignot was named after a family of prominent Paris printers whose death in the Great War had turned them into national heroes; perhaps it was a coincidence that around this time their successors printed an undated brochure entitled Le Sphinx, promoting modern typefaces designed by Alfred Latour. In May 1927, not long after the change to 22 rue des Quatre-Frères-Peignot, the Primes Sphinx press campaign was interrupted. It resumed very briefly in July and October, with a slight change of name to Le Sphinx, and then stopped (note that this was not the only such campaign being run at that time).

One year passed before the campaign was revived, with the same typography and types of quiz, and from the same address – but by new advertiser, named Arya, offering a new prize:

2.000 phonographes de marque donnés pour rien 

Arya may be familiar: it was one of the labels which reissued pre-War vertical-cut Ultima masters, mentioned above. The marque was not registered, which leaves October 1928 as the first documented date for its existence – surprisingly late, considering how old the Ultima recordings were by then. Arya's puzzle advertisements continued for three years, always stating the address as 22, rue des Quatre-Frères-Peignot, although images of Arya record sleeves show the street number as 22 bis, alongside an unusually spelled surname, Le Carèrès. In early 1929, Arya upped the ante to '5000 record-players for nothing'; eighteen months later, another change of name took place.

From the autumn of 1931, 5000 puzzle-solvers were promised free machines by yet another advertiser, Angelus, still at the same address, whose campaign debuted not in the provincial press but in Paris papers. This too was a record label, and like Arya it was never officially registered. Far fewer discs seem to have been issued on Angelus than on Arya: they appear uncommon today, with only seven documented for this page, of which just two are held by the Bibliothèque nationale de France in Paris, as against thirty-six from Arya. Angelus quizzes appeared in French newspapers for four more years, with decreasing regularity, until November 1935.

Well before the disappearance of Angelus, a new label had taken up residence at 22, rue des Quatre-Frères-Peignot – but, unlike Arya and Angelus, it did not continue the quiz advertisements. The earliest sighting so far of Champion in the press, and of the associated children's label Champion Baby, is an advertisement of May 1933, placed by Etablissements Champion. The name implies a fully constituted company which, the advertisement suggests, had existed for some time, although no announcement of its formation has yet come to light. The same is true of another business, named Disques de Phonographes et Articles de Paris, whose beginnings are undocumented but whose assets were officially sold in June 1933. The buyer was the latest new company to be based at 22, rue des Quatre-Frères-Peignot. Le Disque de France had just been formed, on 23 May 1933, by Ernest-Pierre Le Carrérès, resident at that address. (Only one other person, an accountant, was named in the formation notice.)

The assets sold to Le Disque de France included:

• a shop sign, clientèle and goodwill
• various tradenames and marques including, 'notably' (notamment), Champion, Angelus, E.L.C. and Eclair (NB: Arya not mentioned)
• fittings and stock
• 'a catalogue of disc recording(s)' (Un répertoire d’enregistrement [sic] de disques)
• rights to the lease on the premises

Not included, it should be noted, were any recording equipment, disc matrices or metal stamping parts; this was a retailer (and, possibly, wholesaler), not a recording or manufacturing business. The 'catalogue' (répertoire) was surely intellectual property, for which recordings were made and discs pressed by someone else.

The owner named in these transactions has been identified as Ernest Pierre Célestin Marie Le Carrérès, who was born in 1898 in Pleuvihan / Pleubian, Côtes d'Armor (Brittany), and died in 1975 in Paris. (The name is Breton, and printers spelled it in various ways, as seen on the Arya record sleeve mentioned above.) Le Carrérès' activities in the capital left traces in the press from at least 1923. In early 1926, he and his wife were named as partners in a firm making imitation pearls, which existed for just over a year. By 1927, Le Carrérès was also trading in 'fancy goods' (Articles de Paris) from the avenue Émile Zola. This is only a short walk from the rue des Quatre-Frères-Peignot, and by May 1927 Le Carrérès had moved his business there and was selling pocket and wrist-watches.

Presumably, Arya was not listed among the assets sold in 1933 because it was already defunct. Whether its demise had coincided with the arrival of Angelus in September 1931, or whether they coexisted for a while, is unclear. Arya did not only reissue Ultima masters, whose original 'etched' labels are visible under its two styles of paper label, Art Nouveau and Art Déco; it also issued new recordings, some made electrically. The latter are presumed to have been produced by Bernard Roux, for reasons which will become clear, although this cannot be confirmed from the scant information about Arya found to date (no discography of Arya exists, while most sales listings and library catalogue entries are too rudimentary to be of help).

ELC appeared as an additional logo or trade mark on the Art Nouveau-style labels of vertical-cut Arya discs: an obvious acronym of 'Ernest Le Carrérès', whose surname was also printed on Arya sleeves, it apparently referred to a retail business, rather than to another, separate disc marque. Like Le Carrérès' other ventures, E.L.C. too advertised using puzzles: an advertisement from February 1926, the earliest for E.L.C. so far located, revived a type of puzzle first used in 1910 and then reused by Primes Sphinx. Also like some of Le Carrérès' other ventures, E.L.C. was based at 40, avenue Émile-Zola, and recycled Primes Sphinx puzzles sporadically over the following years. What exactly it was selling was not stated until November 1930, when a final brief spate of advertisements promised '5000 free record-players'. Later, E.L.C. puzzles were in turn recycled by Angelus.

It is Angelus which provides the earliest tangible link between the business in the rue des Quatre-Frères-Peignot and Bernard Roux. Another record recently added to the online transfer project mentioned above is Angelus 271, a 22 cm disc whose cheaply-printed, unglazed labels are branded 'ANGELUS PARIS XV^e' (the Roman numeral stands for quinzième arrondissement). It couples Schubert's Moment musical in f minor D.780 No.3 with the finale of Beethoven's Sonata in c sharp minor Op.27 No.2, played by the pianist Raymonde Eustache-Lemaire. A Bernard Roux test pressing which included the Schubert side was auctioned some years ago; as often with Roux tests, the artist was not identified. On the test disc's other side was the finale of Weber's Sonata in C major Op.24, issued on Angelus 273. A second Bernard Roux test pressing, auctioned alongside the one just mentioned, contained the same Beethoven movement, but with a lower matrix number than the one issued on Angelus 271 (6256 MB as opposed to 6349 MB), coupled with Chopin's Fantasy-Impromptu in c sharp minor Op.66; no issues of either side have yet been located. All the above records carry the monogram MB after the matrix number.

As an aside, one possible explanation for Mme Eustache-Lemaire's appearance on Angelus is, simply, proximity. From May 1926 until August 1933, she appeared regularly in newspaper listings of concerts broadcast by the Paris station Radio-L.L., launched in March 1926 by the radio manufacturer Établissements Radio Lucien Lévy (Lévy was an important inventor, innovator and entrepreneur). During these broadcasts she performed the few pieces she is known to have recorded for Angelus and many others, mostly but not exclusively of a popular or 'salon' character. Radio-L.L.'s studio and transmitter apparently shared premises with Lévy's manufacturing business, also in the fifteenth arrondissement, on the rue de Javel. This street crosses the southern end of the rue des Quatre-Frères-Peignot; Le Carrérès would have had to walk no more than a few hundred metres to witness one of Mme Eustache-Lemaire's performances in person. (In mid-1935, Lévy sold Radio-L.L. to another entrepreneur, Marcel Bleustein, who relaunched it as Radio-Cité and turned it into a pioneering and popular broadcaster.)

The other record label listed among the assets sold to Le Disque de France in 1933 was Eclair. A disc marque of this name had been registered in 1929, but it is not clear if it was exploited or how Le Carrérès, with no known connection to the registered owner, came to exploit it – as early as the spring of 1930, according to one account. The Eclair logo originally registered was not the same as that used by Le Carrérès on Disques Eclair labels, which show a clear kinship with Arya's needle-cut electrical labels. Unlike Arya and Angelus, Eclair has benefited from a little discographical study: on some discs, matrices with the Bernard Roux monogram AB are coupled with matrices of different provenance, while on others only Roux matrices are found.

The AB monogram also appears on very many Champion issues, as do MB and RB, both likewise associated with Roux; and, as noted above, a Bernard Roux test pressing of a known Champion issue has been documented. Setting aside for the moment the relationship between Roux and Le Disque de France, a question of chronology arises: which came first, Eclair or Champion? The hypothesis advanced here (with some trepidation) is that Eclair, like Angelus, preceded Champion, and may not have continued in production after the 1933 sale. The overlap between the arrival of Champion and the last Angelus newspaper quizzes (cited above) suggests that the older label was being phased out and stocks run down. Not that anything was wasted: Angelus 271 duly turned up unchanged on Champion 3012, and two more of Raymonde Eustache-Lemaire's Angelus sides seem to have been taken over by Champion (pending discographical checks), which also added other sides not known on Angelus:

Selection	Artist	Matrix no.	Diameter	Issue
Moment musical (Schubert)	Raymonde Eustache-Lemaire	6343 MB	22 cm	Angelus 271
Finale de la Sonate "Clair de lune" (Beethoven)	Raymonde Eustache-Lemaire	6349 MB	22 cm	Angelus 271
Capriccio [Rondo capriccioso] (Mendelssohn)	Raymonde Eustache-Lemaire	6344 [MB?]	22 cm	Angelus 272
Nocturne en fa dièse (Chopin)	Raymonde Eustache-Lemaire	6345 [MB?]	22 cm	Angelus 272
Mouvement Perpétuel (Weber)	Raymonde Eustache-Lemaire	6342 MB	22 cm	Angelus 273
Fileuse (Mendelssohn)	Raymonde Eustache-Lemaire	6347 MB	22 cm	Angelus 273
Mouvement Perpétuel (Weber)	Raymonde Eustache-Lemaire	?	22 cm	Champion 3010
Le Bengali au Réveil (P. Gerville)	Raymonde Eustache-Lemaire	?	22 cm	Champion 3010
Fileuse (Mendelssohn)	Raymonde Eustache-Lemaire	?	22 cm	Champion 3011
Menuet (Paderewski)	Raymonde Eustache-Lemaire	?	22 cm	Champion 3011
Moment musical (Schubert)	Raymonde Eustache-Lemaire	6343 MB	22 cm	Champion 3012
Finale de la Sonate "Clair de lune" (Beethoven)	Raymonde Eustache-Lemaire	6349 MB	22 cm	Champion 3012

(To complicate matters still further, at least one of these sides was issued on another label, as will be seen below.)

Arya and, especially, Champion were by far the most productive of Le Carrérès' record labels, but he was also associated with much less well-known productions. In March 1926, a business supposedly named Arts Religieux and based, like Le Carrérès, at 40, avenue Émile-Zola, had invited clerics to write in and receive 'gratis […] a very beautiful art medallion' of the Virgin Mary, Jesus Christ, St. John or Saint Thérèse of Lisieux, also known as sainte Thérèse de l’Enfant Jésus; the following month, now named L’Art religieux, it offered free First Communion medallions with any purchase of a girl's neck-chain or a boy's watch-chain. (Not long after this, as mentioned above, Le Carrérès advertised watches from the rue des Quatre-Frères-Peignot.) Over seven years later, a disc marque named À Sainte Thérèse de l’Enfant Jésus was registered in Paris; a label image states that the disc's issue had been authorised by the Carmelite convent of Lisieux, while the associated record sleeve gives the address of the 'General agent for the entire world' as 22, rue des Quatre-Frères-Peignot – the address of Le Disque de France. (This shows, yet again, how important it is to retain record sleeves and indeed any printed matter emanating from lesser-known companies; Jerome Moncada's Label Gallery Française is a goldmine of sleeve images, many just as informative, not to mention decorative.) It is possible these discs were also recorded and pressed by Bernard Roux, something which could probably be confirmed by physical inspection.

Le Disque de France also apparently acted as a producer and distributor of advertising records. Marcel Dumont, previously encountered above on a c. 1931 Bernard Roux test pressing promoting a department store's white wine, also sang on records advertising the Williot brand of hot chicory beverage. At first sight, nothing identifies the producer of one such disc whose label is emblazoned in gold on black:

Ça vient de la cuisine (Musique de Jack Auldebine)
Chanté par Mr Dumont de l'Empire
Offert par la chicorée Williot

Luckily, a collector who has published a transfer of one side online also photographed the accompanying sleeve, on which is stamped:

Publicité par le disque et la radio "Le Disque de France S.A."

The other side reportedly carried a musette version of the same highly catchy number, without vocals. (Another copy has survived in a sleeve without the above stamp.)

Le Carrérès produced a promotional record for another drink, a wine-based apéritif called Charnay: a copy auctioned online in mid-2023 carried a white label with the same stamp as on the chicorée Williot sleeve, "Le Disque de France S.A." As well as specifying the diameter, 25 cm, the seller noted that only one side was recorded, the advertisement being repeated five times (a partial photograph showed clear scrolls between the bands). Presumably this allowed the disc to be cued up on a broadcast studio turntable, ready to play publicity 'spots' at the press of a button, between other items or programmes, without recueing.

Georgetty, the singer who recorded a film song for Le Disque de France (and for Champion, Edison Bell and no doubt others), also recorded advertising songs, promoting soft drinks made by the firm Secrestat of Bordeaux. Their presentation varied interestingly: one pressing sported yet another gold-on-black label, this one branded 'Secrestat Bordeaux Disque à aiguille Fabrication Champion':

Selection	Artist(s)	Diameter	Matrix no.	Label cat. no.
One step du Toni-Kola (Abel Monestès)	Georgetty, Orchestre musette Champion	25 cm	7700	Secrestat Bordeaux 603
Bitter Secrestat One step (R. Brun, A. Monestès)	Georgetty, Orchestre musette G. Courquin fils	25 cm	10611	Secrestat Bordeaux 603

Champion reportedly issued the first side under its own name, with the same catalogue number but a different coupling, its neighbouring matrix number suggesting it derived from the same session as the second side above:

Selection	Artist(s)	Diameter	Matrix no.	Label cat. no.
One step du Toni-Kola	Georgetty, Orchestre musette Champion	unknown	AB 7700	Champion 603
Orangeade Secrestat	Georgetty, Orchestre musette G. Courquin fils	unknown	AB 10610	Champion 603

The above issue, described summarily and without label images, may in fact have resembled yet another Champion issue of the first side, its standard labels sporting a discreet banner, 'Offert par Secrestat, Bordeaux':

Selection	Artist(s)	Diameter	Matrix no.	Label cat. no.
Toni-Kola One Step (Abel Monestès)	Orchestre Musette Champion, J.B. Ropp (dir.), Arsène (accordion)	unknown	7700	Champion 500
Les Fruits de France Fox-Trot (Abel Monestès)	Orchestre Musette Champion, J.B. Ropp (dir.), Arsène (accordion)	unknown	7701	Champion 500

For some reason, the labels do not credit the vocalist on both sides (presumably, Georgetty) but do credit the band-leader and accordionist, billing the latter as 'Arsène de l’Abbaye de Thélème', a well-known Montmartre cabaret on the place Pigalle.

(As mentioned above, a promotional disc for the electronic Ondes Martenot may be connected with Bernard Roux's diversification into that field in the 1930s, discussed below.)

However successful le Disque de France may have been – no record of sales or profits has been found – its heyday lasted at most two or three years. In late October 1935, an extraordinary general meeting of shareholders was called, to discuss the way forward in view of 'the loss of three quarters of its share capital'. It is not known how this situation arose or who the shareholders were, other than Ernest Le Carrérès, presumably, and possibly his wife. Weeks later, the business went into receivership; intriguingly, so did Etablissements Bernard Roux, at exactly the same time. Both survived: a few months later, details of agreements with their creditors were published, again simultaneously. The board of Le Disque de France had decided to soldier on, despite enormous losses, and in mid-1936 it was apparently still trading from its former address. In September 1936 Le Disque de France was named as majority shareholder in a new company formed jointly with Bernard Roux, which is discussed in more detail below. Le Disque de France remained in existence; the following January, its registered address was transferred to 2, rue Esquirol, also that of Etablissements Bernard Roux, and AGMs were held there in June 1937 and June 1938. In February 1938, Le Disque de France sold its shares in the new joint company to Etablissements Bernard Roux, no doubt to replenish its cash reserves. At its last recorded AGM, on 30 June 1938, the board again decided to continue trading. This is the last sign of life from the company which has been encountered to date; no official report of its demise has been found.

Without wishing to jump to conclusions, one is tempted to wonder about the nature and extent of the relationship between Ernest Le Carrérès' businesses and Bernard Roux's. Le Carrérès had no studio or factory of his own; there seems little doubt that Roux was the main supplier of recordings and pressings to Angelus and Champion, tests for both of which by Bernard Roux survive. It is hard to be sure how far back the relationship went; most likely, as early as 1928 Roux was pressing Arya discs for Le Carrérès from Ultima masters, to which Roux is known to have had access, and perhaps Roux later also made new recordings for the label (a discography is sorely needed), as well as Angelus and Champion. Whose were the brains behind these ventures? What was the purpose of Primes Sphinx? Why did Le Carrérès move from selling jewellery to producing records? Was Le Disque de France partly or wholly financed by Roux, only for the latter's own difficulties to bring about receivership followed by amalgamation? Or was the catastrophic loss in late 1935 caused by the defection of another shareholder? It is interesting to note that this was exactly the time when Lucien Lévy sold his radio station (see above); were the events related? These questions will have to await research in business sources not accessible online.

Note that Le Carrérès' business had no connection with another also named Le Disque de France: this was a Paris record shop seemingly specialising in recordings of conservative, right-wing, monarchist and patriotic material. An advertisement published in the 1935 Almanach de l’Action française, eponymous annual of the right-wing organisation, gives its address as 36, rue Vignon, in the ninth arrondissement, and lists 'royalist records' selected from the catalogue of the Hébertot label, founded by the poet, theatre director and publisher Jacques Hébertot (born André Daviel, 1886-1970). Le Carrérès' products, meanwhile, were occasionally promoted in both right-wing and left-wing newspapers.

Le Carrérès' career and life after Le Disque de France have not been investigated.

Grands magasins

Like their counterparts in other countries, French retailers sold records under their own imprints, produced for them under contract by other firms. The best known of these shops were Paris's grands magasins (department stores), notably the Bazar de l'Hôtel de Ville (BHV), Le Bon Marché, Galeries Lafayette, Le Printemps and La Samaritaine. Some originated in regional capitals, such as Les magasins réunis, originally of Nancy, and Les Nouvelles Galeries, founded in Saint-Étienne. In-house labels were also marketed by the budget chains which sprang up in the wake of the 1929 Crash, perhaps not strictly speaking grands magasins, although some were owned by the upmarket emporia usually so called. This section briefly examines department-store records for possible links with Bernard Roux.

Roux's involvement in this business was hinted at above by a test pressing which stands apart from other known Roux tests because of its unusual matrix numbers, M 139 A / M 142 A; these allow it to be matched with issues on Mag-Nis, house label of Les magasins réunis during the 1930s. As Jerome Moncada has shown, Mag-Nis discs came in three sizes and several series. These must have been produced, possibly at different (overlapping?) times, by different suppliers – including Bernard Roux: as noted by Moncada, the 25 cm 1000-numbered series, with gold-on-blue-green labels, has matrices both with the Roux AB monogram and without. A dozen of these discs have been documented for this page, of which the highest-numbered, Mag-Nis 1084, is of special interest, because one side was also issued on Parnasse and Excelsior, (on both of which, more below). Another 25 cm series, with the same label but numbered 16000, drew on X-prefixed matrices from the French Crystalate label Magra. Roux also recorded some Mag-Nis 15 cm 700-series children's discs, not covered by Moncada; 'Lolotte', an accordion solo on one side of Mag-Nis 473, had previously been issued on Le Rivolia, in-house label of the Bazar de l'Hôtel de Ville (see also below); another disc in the same Mag-Nis series has matrix numbers suggestive of a Roux production. Another series of children's discs, branded Magasins réunis, was possibly marketed before the Mag-Nis marque was launched: the example whose label is illustrated by Moncada recycled a Pathé matrix also issued on Javo, Pygmo and the house label of La Samaritaine (on which, again, more below).

Meanwhile, the M-prefixed matrices on the Bernard Roux test relate to yet another Mag-Nis series: these 25 cm 500-numbered discs with gold-on-red labels were pressed, explains Moncada, not on shellac but on a light, cardboard-based material. Whether Roux's firm also recorded these or merely pressed them, or both, is not clear; the test itself is not apparently made of such material. What is clear is that some Mag-Nis discs were recorded by another producer: a number of labels in the light-green label series with P.S.-prefixed matrix numbers, bill the recordings as made by the Casares process (Procédé Casarès). This is confirmed by a 1932 trade advertisement for the Casares studio, listing Mag-Nis among its clients; although Spanish, the firm was then based in Paris (where its name was always gallicised as Casarès) and made recordings for a range of French record and film companies. Casares was only a studio, though, not a factory; another firm must have pressed these discs for Mag-Nis, as well as its 2000-numbered gold-on-black label series with Salabert matrices (Salabert was also listed by the Casares studio as one of its clients). But here another complication arises: the initials P.S. are those of the Russian-born Paul Sterman (earlier Stermann, 1889-1969), composer, pianist, conductor, band-leader, publisher – and record producer: in May 1934, several newspapers published notices of Sterman's recent bankruptcy, mentioning his recording studio in the Cité Bergère, just off the rue du Faubourg-Montmartre in the ninth arrondissement, and even a record factory on the rue Sainte-Anastase, in the third arrondissement. P.S.-prefixed matrices are seen on a wide range of labels, including Allegro, Excelsior, Floréal, Prix Fixes and Le Rivolia (both BHV labels, see below), Spring and Saturne; of these, only the last was registered by Sterman, in 1930 (and is not to be confused with the better-known post-War producer of highly decorative picture-discs). Of the three main styles of Saturne label, one is closely related to those of Allegro and, in turn, of Mag-Nis. Allegro's issues consist mainly of P.S.-prefixed matrices, with some prefixed S.E., but it was registered after Saturne by a different owner; and, confusingly, labels of some Allegro discs with P.S.-prefixed matrices state that they were recorded by the Casares process. Working out exactly who recorded what for Mag-Nis and these allied labels, and who pressed the resultant matrices, is a task which will require a much larger sample of discs, carefully documented and cross-checked – and, as ever, perhaps even unpublished paperwork.

Such a tangled web of studios and matrices, issues and reissues can seem impenetrable, and it's a relief when serendipity comes to the rescue. Recently, recordings by an unnamed string quartet, issued by an unfamiliar French label, were added to an online transfer project. Restoring and preparing the transfers for upload required repeated listening, with the unexpected result that the two sides began to evoke strong memories of other recent transfers. Comparison with two 1927 Edison Bell discs of the Brosa String Quartet, transferred as part of the same project, confirmed that one side of the French record was identical to one of the Brosa sides, while the other contained a slightly different take from the one issued by Edison Bell:

Selection(s)	Artist(s)	Matrix no(s).	Diameter	Issue
Mignon Gavotte (A. Thomas)	Brosa String Quartette	10942-3	25 cm / 10 in.	Edison Bell Electron 0156
Mignon (Thomas)	Quatuor à cordes	0350	25 cm / 10 in.	Primivox M 0561
Spring Song. Morceau (Mendelssohn)	Brosa String Quartette	10941-2	25 cm / 10 in.	Edison Bell (Electron) 0172
Spring Song (Mendelssohn)	Quatuor à cordes	0349	25 cm / 10 in.	Primivox M 0561

(For fuller details of these respective issues, as well as transfers and label images, please follow the hyperlinks under 'Issue' in the above table.)

Primivox was the house label of Priminime, a French chain of budget department stores established in late 1932 (Priminime is a pun on prix minime, 'tiny price'). Launched shortly afterwards, Primivox seems to have recycled mainly British and French Edison Bell masters. There appear to have been six or seven numbered series and possibly four sizes – the smallest, at 15 cm / 6 in., aimed at children and branded Primivox Bébé. Two styles of Primivox label are known: one resembles Edison Bell Radio labels and usually states, 'Imprimé en Angleterre' (note that 'imprimé' means 'printed', not 'pressed'), while the other is simpler in design, more poorly printed and, on surviving discs, often rather worn; the disc listed above carries this type of label. Most matrices were given new numbers for issue on Primivox, as on M 0561 (see above table), although at least one Primivox pressing with original Edison Bell matrix numbers is documented. Some masters were issued on both Primivox and Eldorado, a label belonging to another chain, much better known and still in business today, the Galeries Lafayette.

Since Etablissements Bernard Roux was contracted to press discs for French Edison Bell, it would have held Edison Bell manufacturing parts, and it seems reasonable to suppose Roux drew on these for Priminime. The Primivox disc listed above derives from British matrices not otherwise known to have been issued in France; the same appears to be true of other Primivox issues, not only confirming the strategy announced by Edison Bell International at its launch in late 1928 (see above), namely to concentrate on the low-priced, 22 cm / 8 inch Radio label, but also another reminder that string quartet music, however tuneful and approachable, was not considered mainstream fare by commercial companies at this period. Although Edison Bell did issue a fair number of 25 cm / 10 in. Electron discs in France, presumably it was happy for the catalogue to be constructively and profitably cannibalized for clients such as Priminime. The paradoxical result was that these popular bonbons, beautifully played by Antonio Brosa's quartet, were only to be enjoyed by bargain-conscious shoppers, unaware that the performers had received rapturous reviews from Paris critics a few years before. It may also transpire that some Primivox discs were original issues, perhaps recorded by Roux for Priminime. As with Mag-Nis, though, other suppliers were involved. One of Priminime's series, launched in 1933, was Primiquatuor, a series of 25 cm / 10 in. discs each containing four numbers, two per side. This series was surely produced by Crystalate, which had been active and highly visible in France for four years. In early 1933, British Crystalate launched the Broadcast Four-Tune series, followed, probably shortly after, by a French counterpart, Cristal 4 Succès. Both these French series were short-lived and are poorly documented.

Priminime was not the only budget department store to which Edison Bell matrices were supplied. A rival chain, Prisunic, also had its house label, Disclair; as Jerome Moncada cautions, the task of identifying source matrices is bedevilled by their renumbering on Disclair issues and by pseudonyms and omissions on labels. For example, a French collector has identified Jane Pyrac as the uncredited singer on one side of a Disclair record which may first have been issued by Edison Bell France; this Disclair label, printed in England, shows unmistakable affinities with Edison Bell lay-out and typography. But another Disclair side, likewise probably an Edison Bell France matrix, has a label stating not 'Imprimé' but 'Fabriqué en Angleterre' ('Manufactured in England'); again, more discographical work is required to elucidate these chains of supply.

This is work a British discographer is poorly placed to do. French department-store records are not at all common across the Channel or, probably, anywhere outside France. It does not help that the Bibliothèque nationale's otherwise useful catalogue routinely omits matrix numbers from entries for shellac discs. When they are included, prefixes and suffixes are sometimes not noted, making it hard to detect characteristic Bernard Roux matrices. Online sales and other listings with label images fill the gap to some extent; but ideally documenting French department-store labels is a job for local discographers with the necessary expertise, contacts and personal collections. As ever, Jerome Moncada is an excellent guide, for instance to Parnasse, in-house label of Uniprix, the budget department-store chain of the Nouvelles Galeries and a well-known competitor of Priminime and Prisunic. Moncada's survey of Parnasse identifies four main suppliers, three styles of label and several series. Of these series, Bernard Roux was partly responsible for two, a 25 cm black-label 6000-numbered series and a 15 cm 600-numbered series for children with dark-green labels. No examples of the latter have been traced for this page, whereas scouring the internet yields some two dozen of the black-label discs with solid to fairly good attributions to Roux. Some were also issued on other labels, such as the disc mentioned above, Parnasse 6099, one side of which surfaced on both Mag-Nis and Excelsior. Another in the series, Parnasse 6021, with AB monograms on both sides, was duplicated on Eclair and Champion, both labels owned by Ernest Le Carrérès (see above); on one side of these discs, the AB monogram preceded the matrix number iself, and on the other followed it, suggesting one should perhaps not read too much into its exact position. A third disc, Parnasse 6101, with two popular numbers from Bizet's incidental music for l'Arlésienne, was recycled from Champion, more evidence of Roux's involvement – and, perhaps, of Le Carrérès': one begins to wonder if the éminence grise in this web of connections was not Le Carrérès himself, clearly a canny marketer and saleman, who could perhaps have acted as a broker between Roux and his clients, not just in the department-store business but across the retail record trade in general.

Other department-store labels for which Bernard Roux is known or presumed to have produced records are, briefly:
– Le Rivolia (BHV); Jerome Moncada names the suppliers of BHV's little documented 'flagship label' as Ultraphone, Aérophone and L’Industrie Phonographique (Lutin matrices, pressed by Pathé), but at least one Le Rivolia 15 cm disc with Bernard Roux matrices is known (see above), and a disc in Le Rivolia's 25 cm 1000-series has matrix numbers typical of Roux productions
– Prix Fixes (BHV), seemingly less developed than Le Rivolia; Moncada mentions only (P.S.-prefixed) matrices from Allegro and Saturne, but examples are known with P.S.-prefixed matrices on one side and a Roux matrix on the other, also a feature of Disques Eclair
– Samaritaine (A la Samaritaine); Moncada names suppliers as Aérophone, French Crystalate (Clarus and possibly Magra matrices), Saturne and Ultraphone; another was Bernard Roux, whose 4-digit matrix appeared in a 2000-numbered series with labels colour-coded according to genre
– Eldorado (Galeries Lafayette), one of the chain's main labels (others, from different suppliers, were Domisol and Galfadisc; NB despite appearances, La Fayette had no connection to the Galeries Lafayette, see Other labels), according to Moncada, supplied entirely by Edison Bell; therefore, presumably, pressed by Bernard Roux?
– Minny (Galeries Lafayette), a children's label (not yet documented by Moncada); the few known discs can all be related to issues on Champion Baby and Disque Eclair
– Cofra (Compagnie française); it is not clear to which chain this budget label, not yet documented by Moncada, belonged (for instance, department stores of that name in Nantes advertised all manner of goods, including records), but some of the few known Cofra discs drew on Bernard Roux matrices also issued on Angelus and Champion, as well as on I.V.E[.], P.A.L. and Stradella, all labels associated with Roux (see Other labels)

Other labels

NB this section still under construction

The early period of Bernard Roux's activity as producer of gramophone records is one of the largest lacunae in this account of his business. As documented above, Roux's Ebonitine works was offering to manufacture discs from 1907, and to record them (in both hill-and-dale or vertical cut and needle- or lateral cut) from at least 1910. Yet the earliest datable Roux recording was made in late 1922 (for a private client, see above), while the earliest surviving electrical recording is preserved on a test pressing dated 28 January 1925 (see above for a transcript of its extensive and unusual content). This leaves a gap of some two decades between the emergence of Roux's gramophone enterprise and the earliest datable commercial issues which can be matched with Roux tests: currently, these are three Azurephone discs issued some time after the marque was registered in April 1929 (see above). Which other commercial labels, if any, did Roux make recordings and/or press records for during this long, undocumented period?

The people best placed to answer this question are French record-collectors and discographers. Discs issued by Bernard Roux's clients are not at all common outside France; yet handling and examining them is the best way to develop an 'eye' for those physical characteristics which allow these companies' products to be identified. Next best is belonging to a community of fellow-enthusiasts willing to share such details from personal collections. For discographers based outside l'Hexagone, looking at photographs of discs and labels (posted online on discography websites, in sales listings, on media-sharing sites etc.) can be a good substitute – though less so when images exclude the 'dead wax' or runout area, as they often (and very carefully) do. Even more frustrating is the habit, widespread in institutional and academic library catalogues, of simplifying or omitting matrix information. Nevertheless, very many discs produced during Bernard Roux's active years in the gramophone business can be examined remotely, and months of immersion in this rich legacy have made it possible to identify, more or less positively, several of the clients Roux recorded and pressed discs for. (As ever, tribute must be paid to Jerome Moncada, whose website Label Galerie Française has been an invaluable guide and touchstone.)

In early 2024, work on this page was intermittently halted, while a database of some 600 discs attributable to Roux's studio and/or factories was compiled from the types of sources mentioned above. What it currently contains is just a fraction of perhaps thousands of records his business must have produced over more than three decades, most of which do not figure in published discographies. Unsurprisingly, it has almost no entries from the first two of these decades, for which Bernard Roux test pressings are almost completely lacking. One exception is the 1923 private recording which prompted this entire project (see above); another apparent exception, the Roux test held by the British Library in London, is somewhat deceptive. Neither side was recorded by Roux: an Edison Bell matrix, dated on the label to 1930, is coupled with an Ultima matrix recorded no later than 1911 but, clearly, pressed in 1930, probably for Edison Bell. Still, the fact that Roux held this matrix opens one avenue of exploration, namely the afterlife of Ultima matrices which, as noted above, were recycled on a variety of labels well into the 1920s. Some of these labels, such as Arion, Corona, Lutètia and Opéra, were registered between 1921 and 1928 by one or both brothers, Etienne and Pierre Chanoit, or associates of theirs; another, Apollon, was registered in 1922 by music and record retailer Adam Morhange. To judge from the striking similarity between the 'Disque à saphir Apollon AM' label and the 'Disque à saphir Arya ELC' label (neither of which was registered in that form), Morhange's recycling of Ultima matrices must have morphed into the operation which ended up in the hands of Ernest Le Carrérès (see above). Indeed, one pair of Ultima matrices was reissued on both labels and on the later Arya Super Disque; the well-documented relationship between Roux and Le Carrérès makes it seem highly probable that Roux was involved in supplying Ultima matrices to him. (The other label prominent in Ultima's afterlife was Perfectaphone, not connected to Roux or the other entrepreneurs just mentioned.) But the lack of discographical data for both original Ultima issues and their recycling on these labels (apart from Perfectaphone, none has been the subject of published discographical work) means that only researchers with access to original copies (or detailed descriptions and images of them) will be able to trace this murky nexus definitively.

What the Roux database can confirm, without the need for a trip to France, is that a substantial number of smaller French labels of the late 1920s and early 1930s drew on matrices originating from the same source as those issued on Angelus, Champion and Eclair – that source being Bernard Roux. As Jerome Moncada explains, many Champion matrices are alphanumeric, consisting of a number accompanied by a distinctive, monogram-like fused doublet of letters AB, MB or RB (in the last, the R is reversed). These, Moncada notes, are also seen in matrix numbers on three slightly earlier labels, PAL, IVE and Inovat. They occur on other small labels such as Azurephone and Stradella, on department-store labels including Cofra, Parnasse and Samaritaine, and production-music labels such as Barthe, Choudens, Eschig, Floréal, Maurice Brun, Yatove and others. This was noticed long ago by French collectors, one of whom wrote a brief guide to identifying French manufacturers from matrix prefixes and suffixes. In a table listing these, the AB suffix is attributed to Champion as manufacturer of discs for the following labels:

Arc en Ciel – Arya – Barthe – BHV – Choudens – Cofra – Delecta – Eclair – Eolian [sic] – Etoile – Excelsior – Faunéas – Floréal – France-Edition – Hebdo Disque – Inovat – I.V.E – Loyraux – Mag Nis – Mateo – Multiphone – Orfe [sic] – Pal – Parnasse – Polydor JAP – Prix Fixes – Bernard Roux – Samaritaine – Vox Populi

(It comes as no surprise to see Arya and Eclair in this list; as noted above, they belonged to Le Carrérès.) The guide's author added a note about the AB doublet:

This mysterious suffix [first] appears in early 1929 on Inovat discs, then on I.V.E. and P.A.L (Etablissements Palma, 99 boulevard Auguste Blanqui). At that point, the letters AB (Auguste Blanqui?) are quite separate. Later, they are joined to each other, fusing vertical strokes of the A and B.

Some additional observations can now be made. Of the three doublets used by Roux, AB appears to have been the most common, outnumbering MB (the second most common) and RB (the rarest) by an order of magnitude or more. On the earliest issues showing the AB doublet, it is neither a prefix nor a suffix but instead stands quite separate in the 'dead wax' or runout area – on several I.V.E. 15 cm discs, for instance, the letters AB are diametrically opposite the matrix digits. Finally, some time after migrating to join the numbers as a suffix, the doublet shifted again to become a prefix. The lowest-numbered such matrix in the rough-and-ready database is AB 8190, although swathes of discs with neighbouring matrix numbers are poorly documented, so that it is not yet possible to determine exactly where in the range of matrix numbers the change from suffix to prefix took place.

The suggestion made en passant above, that AB stood for Auguste Blanqui, is ingenious but probably wide of the mark. As has already been seen, the letters AB are found on Bernard Roux tests which have nothing to do with Etablissements Palma and must predate its first issues. The lowest AB-suffixed matrices documented on that label (5286 [AB] / 5287 AB, on PAL 102, seemingly the second issue) are considerably higher than the lowest AB-suffixed matrix on a commercial issue (4646 AB, on Disques I.V.E. 801) documented in the database. This matrix is also lower than the lowest well-documented AB-suffixed matrices on Inovat (4746 AB / 4747 AB, on Disque Inovat 2012, dated by the BnF to 1931), identified in the quotation above as the earliest label to use AB-suffixed matrices. These are in turn higher than all matrices known to have been issued on Azurephone (the lowest is 4748 AB, on Azurephone 1008). Truth be told, dating these matrices and issues is very difficult; as far as can be ascertained, most companies supplied by Roux, such as Azurephone, I.V.E., PAL and others, did not publish new release supplements or bulletins or submit discs to the press for review. Broad-brush dates for some issues can be obtained from the Répertoire phonographique, a commercial annual published from 1932 onwards, while others have been dated by assiduous sleuthing of jazz sessions, song publication and film release dates.

Jerome Moncada has proposed that the AB monogram was the 'signature' of an unidentified recordist. Presumably, these were not the person's initials: it seems improbable that Bernard Roux employed three recordists whose surnames all began with B-, as implied by the two other monograms known on Roux productions, MB and RB. For the moment, the significance of these letters remains unknown, as are the dates when each was introduced and phased out, and the reasons why. Currently, the lowest known matrix accompanied by one of these doublets is 3586 AB, on a 27 cm Bernard Roux test side containing a canticle by Charles Bordes (see above). This is coupled with a recording of a Hawaiian guitar plus, apparently, banjo, with a different style of matrix number, E 584. As will be seen shortly, couplings of Roux matrices with those of other producers are far from unknown in the output of Bernard Roux's firm, although it is not certain known whether either of these sides was issued commercially.

Rather than indulge in more speculation, the rest of this section will survey, in (very) roughly chronological order, labels for which Roux is thought or known to have produced recordings. Rather than repeat the information provided for each by Jerome Moncada, it will only add any further information or examples which seem interesting or notable.

Disque(s) I.V.E.
On 26 January 2024, Jerome Moncada announced that he had temporarily withdrawn several pages from his website, so as to incorporate information newly available to him about the labels concerned. This was good news, as the five labels in question were all supplied by Bernard Roux, prompting the hope that previously unknown source(s) and/or test pressings have come to light. One of these labels is IVE or I.V.E., about which little is known, since it was not registered; the general opinion in France, reported doubtfully by Moncada, had been that it belonged to Inovat (see below). Moncada reported finding the AB suffix on 15 cm yellow-label Disques I.V.E. issues, in a 4600-numbered matrix series (which in fact went as high as 4701 ABand 4701 AB), and on 18.5 cm powder blue-label Disque I.V.E issues, in a 5000-numbered matrix series. On a smaller number of these issues, the MB suffix is seen in a 6000-numbered matrix series. What is interesting about the latter is that several seem to be replacements of earlier sides with 5000 AB matrices: for instance, on Disque I.V.E 5001, Transibérien, a pas redoublé (matrix 5229 AB), was replaced with Mougeot's Le Tram (Y’a que l’amour) (matrix 6428 AB).

Le Carrérès (or whoever was the brains behind Champion) was certainly skilled at recycling and exploiting recordings to the full. A little-known sideline of Le Disque de France was supplying recordings for a series of discs, at 18 cm smaller still than most Champion issues (except those aimed at children), celebrating films and stage shows. The top half of their labels read:

Le Programme Sonore
Fabriqué par le Disque de France
100% français
Enregistrement électrique

Four Le Programme Sonore discs have been documented to date. Each side usually contained one or two songs from a film or show; these were not soundtrack or original cast recordings, but new recordings by other artists. On two discs, for whatever reason, the coupling was a Strauss waltz. Two have survived with elaborate packaging, a combined sleeve and four-page brochure giving full film credits, lyrics of numbers sung on the disc, and profusely illustrated stills and other images; presumably all discs in the series originally had these 'Talkie Programmes' ('film sonore' was the French term for 'talkie'). Le Programme Sonore 41 promoted the 1934 film L’Auberge du Petit Dragon, coupling its main song, 'Un coup d’jambe en l’air', with 'Aimer, boire, chanter', i.e. Wein, Weib und Gesang Op.333 by Johann Strauss II. Le Programme Sonore 42 promoted another 1934 film, Sidonie Panache; one side has been transferred and made publicly available online, in a video showing the label and sleeve-brochure. Le Programme Sonore 43, promoting the 1935 film Le billet de mille, again coupled its vocal selections with a Strauss waltz, 'Le beau Danube bleu' (An der schönen blauen Donau Op.314), the latter clearly an existing recording (not yet identified) with a much lower matrix number (543, as opposed to 10410). The fourth known Le Programme Sonore disc promoted not a film but Le pays du sourire, a French version of Lehár's popular Land of Smiles, and it was produced not by Le Disque de France but by Cristal. Le Programme Sonore was, clearly, not owned by Le Disque de France, though by whom is not known, as the marque was not registered. Hoped-for hits from new films and shows were recorded and heavily promoted not only by the leading companies but also, opportunistically, by smaller labels. With their small size, bright and busy brochure-sleeves, and Strauss waltz couplings, Le Programme Sonore discs look like unofficial souvenirs, an attempt to stand out from the very stiff competition.

Le Disque de France may not have owned Le Programme Sonore, but it seemingly made sure to exploit to the full its sessions for Le Programme Sonore, recording the songs more than once so they could be issued on its own and other labels. Thus, 'Un coup d’jambe en l’air' was issued on both Le Programme Sonore 41 (diameter 18 cm), and on Champion 1856 (diameter 25 cm), reportedly from the same matrix (AB 10294); more likely, the Champion issue was from a matrix one digit lower, also issued on another 25 cm disc, Excelsior 35, held by the BnF. Both sides of Le Programme Sonore 42 (matrices 10320 / 10322) have parallel issues on Champion from neighbouring matrices (10319 / 10321), while one side of Le Programme Sonore 43 (AB 10410) was again shadowed by an issue from a neighbouring matrix (10405) on Excelsior 81, of which a video transfer has been made freely available. (Lest it be thought that too many assumptions are being made here about missing matrix prefixes, please see the BnF's transfer of Excelsior 74, whose images clearly show matrix prefix AB present in the runout area on one side, and suffix MB on the other, but both absent from the labels.)

NB intervening sections still under construction

Electronics

In mid-1934, when Bernard Roux's second disc-pressing facility was in production, his business began to receive attention in the press, but for a different reason. One of these early reports did not even get Roux's name right – understandably, perhaps, since the new development was very much associated with another figure, then more newsworthy than Roux but now almost as little known.

(André) Eric Sarnette

Joseph Antoine André Sarnette (1898-1993) was born in Tarbes, capital of the French Hautes-Pyrenées. After studies in Marseilles, Nîmes and Paris, in 1917 he entered the Paris Conservatoire's organ and composition classes. He soon showed a special interest in film music and in brass and wind instruments, publishing articles on both subjects in a popular illustrated magazine as early as 1919; Sarnette reportedly also studied orchestration with Adolphe-Edouard Sax (1889-1945), son of the great instrument-maker and inventor. Not yet twenty, Sarnette published a combined manifesto for the wind-and-brass band and blueprint for a band conservatoire. By 1927, when Sarnette was a professor, a public organist (performing on the organ at the Palais du Trocadéro) and in charge of 'mechanical music' (i.e. piano-roll production) at Pleyel, his interests had merged in a militant vision of mechanized music-making, with organ, wind and brass replacing the supposedly superannuated, string-dominated orchestra. He gained considerable publicity, if not notoriety, for statements such as 'I can't stand the violin!' (Moi, j’ai l’horreur du violon), often couched in misogynistic language and provoking controversy at home and abroad.

Five years later, now going by the forenames Eric André, Sarnette was appointed France's first 'professor of the microphone'. The adoption of electrical recording had overtaken Sarnette's objections to traditional string-based ensembles, yet he was still proposing to create new wind-based orchestras suited to the microphone, and to teach a new way of composing for it. (His collaborations with Sax's son and Leblanc company did result in new and modified instruments tailored to the requirements of modern music and media.) Still given to outbursts against 'feminine music', Sarnette was now also working in film production, and it was perhaps his contact with professionals there which suggested a new approach, tackling not sound production but sound capture and reproduction. Of course, Sarnette was far from the only person in France concerned with the state of audio affairs. For years, French radio producers and writers had been wrestling with techniques and aesthetics of studio sound, while in the early 1930s sound cinema exposed ever more clearly the limitations of small, dead studios and unvarying aural perspectives.

In 1931, two young engineers in Paris, Robert Gamzon (1905-1961) and Mario Sollima (1904-1994), applied for a patent in audio playback with 'imaging' (relief) through loudspeakers, granted the following year. It was followed in 1934 by another, 'for improving electrical playback of music', and in the same year they applied for a third, granted in 1936, for 'electrical correction of hall acoustics'. Alongside their names on this patent was another: Bernard Roux. Gamzon had mentioned the firm in a 1932 lecture, possibly speaking from inside knowledge, but only in 1934 were he and Sollima were named as employees of Roux. An article published that year reproduced a high-magnification photograph showing the difference between disc grooves cut by a standard saxhorn and those cut by an instrument built by the Selmer company – in consultation with Sarnette (now plain 'Eric'). Disc and photograph formed part of a technical study of these instruments' properties credited to 'Gamzon and Sollima, engineers from the firm of Bernard Roux', which may have been commissioned by Sarnette for his new book, La Musique et le micro. (No copy of the disc, which must be added to the list of Bernard Roux 'experimental' recordings, is known; there may have been more than one, given Sarnette's eagerness to demonstrate the new 'microgenic' wind and brass instruments.)

At some point Sarnette himself had been hired as 'musical advisor' to Bernard Roux's laboratory: given his long-standing interest in film music on the one hand, and Gamzon's and Sollima's mid-1931 patent on the other – and not forgetting Bernard Roux's possible contact with Gustave Lyon, or indeed Sarnette's time at Pleyel – one is hard pressed to say who initiated this, or exactly how the 'team' (équipe), as Sarnette liked to call it, came together; another possible link is the fact that Gamzon and Sollima had recently helped another innovator, Gabriel Boreau (1880-1957), devise and construct an electroacoustic intrument, the Radiotone. Perhaps unexplored technical archives will reveal what went on out of sight between 1931 and 1934. Perhaps it was this appointment, too, which made Sarnette, previously contemptuous of radio, turn his attention to it. In July 1934 he was a founding member of the Union d'Art Radiophonique, and in the same month, Sarnette organized a small demonstration for music and audio professionals of a new device, designed to improve audio capture (améliorer la prise de sons). He had perfected it (mis au point), said one report, together with Gamzon and Sollima, who in turn had had a 'highly technical discussion' (discussion technique des plus poussées) with one Pierre Bernard – clearly a misnomer for Bernard Roux. A few months later, it was announced that a state committee tasked with evaluating 'M. Sarnette's proposed "reverberation chamber"' had made a favourable report, as a result of which the government's recently acquired station Radio-Paris would start testing it with orchestral and vocal material.

Meanwhile, further demonstrations were held: a second was given by Sarnette and Gamzon to the French Society of Photography and Cinematography in January 1936, followed by a third for the press in May; this was the demonstration held in the Roux studio-laboratory on the rue Jenner (see above). The device was described in articles by the inventors themselves, in the popular science and general press and in a book about film music published in the same year. The latter drew some ridicule from the composer George Antheil in the United States, directed chiefly at Sarnette's 'dazzling theories' about the suitability of brass orchestras for film soundtracks. In France, meanwhile, the device was broadly welcomed – perhaps unsurprisingly, since several press commentators were long-standing partisans of Sarnette's programme. They and others mounted something of a campaign for it to be adopted by the state's stations, in the face of apparent silence on the subject from Radio-Paris. Even Paul Castan (1899-1987), revered pioneer of French radiophonic art, argued that the ideal radio studio must be equipped with variable reverberation. By then, it had already been taken up by the private radio sector (see below).

Gamzon-Sollima device

The Gamzon-Sollima device seems to have been a combined frequency-splitter and reverberation unit, allowing the acoustics of a broadcast or recording to be modified in real time. The audio signal from a studio was fed to a set of three variable band-pass filters which split it into treble, middle and bass signals, a technique named 'split sound capture' (prise de son fractionnée). These were fed through a control panel to three speakers in a small reverberation chamber containing a microphone. An operator at the control panel, following a score or script for the programme being recorded or broadcast, and able to watch proceedings in the studio or sound stage through a window, modulated the split feeds to the reverberation chamber, altering the mix of frequencies output by the three loudspeakers, a technique dubbed 'variable electrical reverberation' (résonance électrique variable). This modified output was picked up by the microphone in the chamber and fed back into the main audio output.

It would be possible today to hear the Gamzon-Sollima device in action, if copies of the demonstration discs recorded by the Bernard Roux engineers could be located. As noted in contemporary accounts, a number of these had been prepared for the presentations, where, presumably, it would have been impossible to set up the device itself. After Robert Gamzon's presentation in January 1936, one of the discs played was described in especially precise detail:

Of the examples presented, the ones we found most interesting were the effects of an actor gradually receding in an interior (by increasing reverberation and decreasing output level) or in the open air (total cancellation of reverberation, attenuation of high and low frequencies by filters in the main signal path, and reduction of output level); we should also mention the very varied reverberation effects achieved in successive recordings of the same orchestral piece or children's chorus.

At the May 1936 talk, the musical recordings made even more of an impact:

Eric Sarnette has just proved the excellence of this technique in public […] not least by playing the same [recorded] piece (orchestral, organ, vocal, spoken word) in turn without reverberation, and then with. It's like night and day…

Another of the discs played at this event presented a poem, Le vent by Emile Verhaeren (1855-1916), again in two versions, both recorded by the actress, dancer and choreographer Marie-Louise van Veen (1900?-1985). On one side, she recited it 'with consummate skill' (avec un art consommé) but without effects; on the other side, varying types of artificial reverberation were applied to her second recitation, making the poem 'strangely captivating' (étrangement saisissant) for one listener. The same account suggests there were two demonstration records involving an actor: in addition to the one offering comparison of simulated receding motion inside and outside, on another

the speaker, plainly standing stock-still a foot from the mic, appears to talk and walk from one room to another, from a staircase to a corridor and from there to a vault

Thus, it seems at least six recordings were made by Bernard Roux to demonstrate the Gamzon-Sollima device:

• Actor moving away from listener, interior / Ditto, exterior
• Actor moving between different interior spaces
• Orchestral music without reverberation / Ditto, with reverberation
• Organ music without reverberation / Ditto, with reverberation
• Song(s) without reverberation / Ditto, with reverberation
• Children's chorus without reverberation / Ditto, with reverberation

(No spoken word example is listed, because it seems Verhaeren's Le vent did duty in this category, even though it was not produced specifically for this purpose, as will be explained below.)

The exact number of resulting discs remains to be determined, and depends on how the contrasting examples (interior / exterior, without / with reverberation) were presented, recorded and pressed. Were complete speeches and pieces of music recorded, or only excerpts? Did they follow each other on one side of a disc or were they pressed on separate sides? All that is certain at present is that Le vent occupied one disc, with one version per side, as dictated by its length.

Société des Procédés Bernard Roux d'Electro-Acoustique

In the meantime, a new limited company was formed, in September 1936, by Le Disque de France, Etablissements Bernard Roux, and Bernard Roux himself. Its aims included 'exploitation of a recording studio, exploitation of a laboratory for study of electro-acoustics problems, as well as manufacture and sale of related equipment'. Its initial capital was a modest 65,000 francs and its head-quarters were at 2, rue Esquirol, the address of Etablissements Bernard Roux. Roux was also managing director but the major shareholder was Le Disque de France.

The following April, the company, also known by the acronym SPEAC, changed its name to Société des Procédés Bernard Roux d’Electro-Acoustique. Finally, in June 1937, fourteen new partner-shareholders were registered, among them the engineers Robert Gamzon and Mario Sollima, Bernard Roux's wife, another Roux family member and, as second largest shareholder, Etablissements Edouard Belin, a company exploiting the inventions of its eponymous founder and also involved in electronics, including early television.

Conspicuously missing from the list of shareholders' names was Eric Sarnette, despite the fact that he had been closely involved in developing the Gamzon-Sollima variable reverberation device, and was reportedly employed by Roux. It is tempting to read too much into this, and into the fact that, over time, his name gradually disappeared from the device's multifarious appellations, until it became known simply as the 'appareil à résonance Bernard-Roux' (and variations thereof). Understandably, once the device began to be marketed, it needed a name, which could only be that of the firm which had developed it and was making and selling it. Sarnette himself became less visible in the press over the same period; the reason for this is not known. It is notable that at the demonstration given on 24 February 1937, just before the device's inaugural broadcast, Sarnette expressly chose to speak about his other project, to create 'radiophonic' orchestras, and not about his work 'in the laboratories' (i.e. with Roux). Sarnette even went so far as to say,

An orchestra is not ‘radiophonic’ only because it includes modified instrments, but because the porportion of these instruments is somewhat different from that of a concert-hall orchestra. One does not obtain a perfect broadcast merely thanks to a studio's reverberation or the quality of the microphones: there is more to it than that

It seems France's radio and film studios did not adopt Sarnette's proposals for recasting their orchestras, to the chagrin of his remaining supporters; likewise, interviews and articles propounding his ideas soon disappeared from sources consulted for this page.

Meanwhile, in February 1938, le Disque de France had sold all its shares in the Société des Procédés Bernard Roux d’Electro-Acoustique to Etablissements Bernard Roux. This is the last mention of the new comnpany found to date in sources accessible online. Soon after, Etablissements Bernard Roux quit the gramophone disc business all together (see #Other labels|above]]) but stayed in business, while the Gamzon-Sollima device also continued to be marketed to new clients (see below), so perhaps the Société des Procédés Bernard Roux d’Electro-Acoustique was dissolved and its business taken over by the older firm. This can presumably be verified from sources not accessible online.

Clients: private

The first known private user of the Gamzon-Sollima device may have been Marie-Louise van Veen, mentioned above. Her two recordings of Verhaeren's poem Le vent, one without effects and one with varied reverberation added by the Gamzon-Sollima device, were played during the May 1936 demonstration at the Bernard Roux studio-laboratory – but they had been made several months before, at the behest of one of France's foremost practitioners of 'radiophonic art', Gabriel Germinet (alias Maurice Vinot), as he himself revealed in the press a year later. In February 1936, van Veen mounted an evening of performances by herself and students from her 'training for cinema and radio' class, at which the Roux disc of Le vent was played. A month later, in a newspaper article about her art, she wrote,

Moving away and closer, swaying, spinning, the play of shadow and light, rapid falling, gliding – all that and more can be rendered visible purely by vocal means. […] In Verhaeren's Le vent, I pushed these experiments to extremes on disc, and a recording like that demanded great technique and hard toil. […] Starting in 1926, I had been studying these problems in relation to the theatre, directing a group of professionals. I am currently revisiting them in a recording laboratory, as well as with a special class at the Russian Conservatoire in Paris

Van Veen played her record at another event in late April 1936, shortly before it was presented as one of several discs demonstrating the Gamzon-Sollima device's capabilities at the Roux laboratory in the rue Jenner. Later that year it was broadcast by Belgian radio, in an evening of programming marking the twentieth anniversary of the Verhaeren's accidental death on 27 November 1916; and in early February 1937, the disc was again heard over the air, during an hour-long broadcast of 'spoken choruses' performed by van Veen and her company, which she had previewed in her own studio for guests including Germinet. Not everyone was convinced: one critic, fed up with mannered radio poetry readings anyway, felt van Veen was over-rated – and however 'intriguing' (curieux) the effects in Le Vent, 'she trots them out at every turn' (elle nous les ressort partout). Which was not going to stop the disc from being aired, until 1939 and possibly later.

Clients: broadcast

No published notice or report has yet been found of the outcome of tests of the Gamzon-Sollima device by the state-funded Radio-Paris, announced in early 1935; one commentator ascribed this lack of interest to the broadcasting minister's obsession with television. Instead, the first broadcaster to adopt the new technology was the private Poste Parisien, a decision possibly taken before the May 1936 demonstration. Actually plumbing the device in and testing it must have taken some time; one account stated that it had been deployed for 'some' of the station's studios, which presumably meant that more than one was purchased, as each studio needed its own control console and, probably, a dedicated chamber (this article includes rare photographs, supplied by Bernard Roux, of the main equipment cabinet and console). Only in February 1937 was the Gamzon-Sollima device ready to be unleashed on the listening public. After another presentation by Gamzon and Sarnette on 24 February, an elaborate inaugural programme was devised for the evening of Friday 26 February, consisting of:

• La Parade sentimentale du Frégoli sonore, a sketch written specially for the programme by Suzanne Malard, with a cast including the well-known radio director André Alléhaut, also producer of the broadcast
• Demonstrations of the Gamzon-Sollima device in use with vocal and organ music
• Demonstration of the Gamzon-Sollima device in use with the Radio-Paris orchestra, conducted by its music director Théodore Mathieu, in Wagner's Tannhäuser Overture
• Cent atmosphères, a playlet by Michel Dulud, evoking a cruise on the passenger liner Normandie
• ‘Alleluia’ by Vincent Youmans ('Hallelujah!', from Hit the Deck)

The broadcast was well received by enthusiasts who had been agitating for the device's adoption, but also by commentators who hadn't. One of these gave an especially complete account, adding details lacking from other reviews: the orchestra, for instance, played instruments modified by Sax to Sarnette's specifiations, and was seated in a novel layout. The list of music, performed first without reverberation and then with, also included unspecified excerpts from Wagner's Parsifal and Bizet's L'Arlésienne, as well as Borodin's On The Steppes of Central Asia – the latter played on the organ, which the writer found somewhat unsuitable. In the shipboard playlet, the new device passed with flying colours:

The actors, placed in the studio at a constant distance from the mic, and speaking without varying their delivery, all but convinced one they were on board the liner, moving from cabin to ballroom or swimming pool, where the watery noises were particularly well imitated and sounded "all together lifelike".

Other listeners remained sceptical: two agreed there was too much talking between the musical examples, while for another, normally an avowed fan of the device, the comparisons had been a little naive. The most questioning critic was a woman, Wanda L. Landowski (1899-1959; not the celebrated harpsichordist but a musicologist), who applauded the Roux team's efforts but felt it had lost its way: in wondering why additional 'echo' was needed for orchestral music, she was perhaps forgetting years of complaints about excessively dry studios, but she was probably justified in calling it 'disastrous' in Fauré mélodies and excerpts from L'Arlésienne, showing in this and other comments a perceptive grasp of the range of acoustics required by different genres and instruments.

But the Gamzon-Sollima device was here to stay, and was soon being used in what was surely its most obvious and suitable application, radio drama and features – so much so that when it was not used, critics complained, while a member of the listening public even wrote to a newspaper castigating the state-funded service for ignoring it. Le Petit Parisien published some judicious propaganda, in the form of a vivid account of a live broadcast from a Poste Parisien studio, during which the duty sound engineer enthused about the 'astonishing' Bernard Roux device, described as resembling an electric furnace. In the end, the ministry gave in and announced that it would equip the capital's station with an artificial reveberation device, although the official parliamentary proposal did not specify the inventors or manufacturer; these were only named some months later, in news items probably prompted by a press release from Bernard Roux. The installation was again described in a detailed popular science article, shortly before it entered service in October 1938. The device seems to have remained in use by Paris radio stations for some years: a 1942 article in a magazine for young people briefly described its potential in 'radiophonic theatre', without naming Roux but proudly noting it was a French invention. Another popular science article alluded to it briefly in 1948, again not by name, but by 1952, it was being written about in the past tense as pre-War technology, for instance in an article which confirmed its main application had indeed been acoustical effects (effets de timbre) rather than routine correction of hall reverberation. It is not known when France's radio studios actually decommissioned their Roux devices.

Nor is it known if the Gamzon-Sollima device was bought by broadcasters outside France. One 1937 article stated that, following the Poste Parisien's adoption of the technology, 'several foreign stations are planning to do likewise', but no further reports on this potential development have come to light. The device was noticed outside France, with brief reports published in Britain, Italy, Spain and, probably, other countries (not all countries' historical press archives are accessible online). Relevant patents were obtained in Britain, Germany, Italy, Spain, the USA and, again, possibly elsewhere.

Clients: 1937 Expo

In June 1937, a Paris industrial newspaper noted that the Gamzon-Sollima device, installed some months before at the Poste Parisien, was proving to be 'indispensable for radiophonic broadcasts', adding, 'New installations are under way, notably as part of the substantial sound system for the Paris International Exhibition'. Officially titled the 'International Exhibition of Arts and Technology in Modern Life', this event ran for six months, from late May until late November that year, and celebrated the power and potential of electricity – memorably depicted by Raoul Dufy in his vast painting La fée électricité, commissioned for the Expo's Palais de l'Électricité. Among the new public spaces and buildings, temporary halls and national pavilions, purely ephemeral displays were also mounted. For an arts magazine published in Brussels, a young French composer, not yet thirty, described one series of these, the 'Festivals of Light' (Fêtes de la lumière):

These festivals – planned by the architects Beaudoin and Lods and brought to life in veritable fairytale extravaganzas of water, light and pyrotechnics – are accompanied by a tightly synchronized commentary in the form of eighteen scores composed specially for the occasion. […] I should add that the scores, recorded onto discs and played back through impressive amplifiers, would run a serious risk of distortion if it weren't for Bernard Roux's split volume-controls, which allow the three audio frequency ranges, low, middle and high, to be individually attenuated or boosted.

The writer, unmistakably describing the Gamzon-Sollima device, was Olivier Messiaen (1908-1992), composer of Fête des belles eaux, now the most famous work written for the 1937 Expo. It is scored for a sextet of Ondes Martenot, the French electronic instrument invented some years earlier but newly improved; several other works commissioned for the Fêtes de la lumière also called for it, in numbers from one to four. (An item played by eight Ondes, from a concert given in one of the Expo halls, has been preserved on sound film). The very varied compositions accompanied displays which took place every day throughout July and August, usually in the evening (although Messiaen's work was first heard during the daytime), along the 1300-metre stretch of the Seine from the Pont de Passy (now Pont de Bir-Hakeim) in the west to the pont de l'Alma in the east. For many reasons, not least the strict synchrony mentioned by Messiaen, the eighteen commissioned compositions had been pre-recorded, and during the displays were played from disc to thousands of spectators thronging the embankments and bridges, through forty-four loudspeakers mounted on eleven pontoons. The complex choreography of music, light, water, smoke, foam and fireworks was a demonstration not only of electricity's power but of electrical control and coordination, as its designers Eugène Beaudoin and Marcel Lods explained in detail for a broadcasting magazine, paying tribute to the 'many technicians of the highest calibre' (nombreux techniciens de tout premier plan) who made it possible.

Among them were the Bernard Roux engineers, who had supplied a Gamzon-Sollima device: its electronics and console were in the control centre, on board a small barge dubbed the 'bateau-studio', while its reverberation chamber was enclosed within the pont d'Iéna. Newly restored and widened, the bridge was at the heart of the Expo and offered a pedestrian walkway between the Palais de Chaillot and the Eiffel Tower; the 'bateau-studio' was moored just upstream of it, in front of the Swiss and Belgian pavilions. Its control centre also contained three turntables, from which the pre-recorded discs were played back, a microphone for announcements and, not least, the nerve-centre of operations, a large electrical control panel. A partial photograph, published in an illustrated magazine, appears to show what may have been the console of the Gamzon-Sollima device to its right, although accounts differ as to where this was located and who operated it. According to the composer and pianist Jean Wiéner, the person in charge of the device was none other than Maurice Martenot, inventor of the Ondes, but it is unclear whether Martenot turned the split volume-controls himself or relayed instructions from the 'duty' composer present to an operator. Another account suggests two architects involved in the light and water shows were at the controls. (As well as these technical and logistical questions, the scrolling graphic 'scores' for each fête, displaying the required water, light, colour and pyrotechnic effects for the control panel operator to follow, also merit closer investigation.) Whoever was in charge of the Gamzon-Sollima device, it cannot have been easy to adjust its output against the din of fountains, water-, foam- and smoke-jets, fireworks and thousands of spectators: one critic called the entire musical venture a failure, while for another the first hearing of Messiaen's score was spoiled by a poor recording – made, it should be noted, by the Compagnie de Gramophone at its Paris studio; a third preferred to suspend judgement on music heard mainly in 'gusts' (bouffées). Clearly, there was not much the Gamzon-Sollima device could have done in these circumstances.

The supplier of the turntables which played back the pre-recorded scores was not named in sources consulted for this page; it may have been Bernard Roux, who had patented a twin-turntable unit that same year. One source also appears to attribute the Expo's loudspeaker system to Roux: in a newspaper interview, Arthur Honegger, another composer commissioned to write a score for a fête, spoke of 'the new Bernard Roux reverberant louspeakers' (les nouveaux diffuseurs réverbérants de Bernard Roux), but this slightly vague statement has not been corroborated and may have arisen from a misunderstanding Honegger's part or the reporter's. Roux was listed as a supplier in the official Expo catalogue but what the firm supplied was not specified.

Barely audible as they sometimes were, the Fêtes scores which used Ondes Martenot probably did less to impress the instrument on visitors' minds than another strand of events at the Expo. From 17 September 1937, the pavilion of the Union Corporative des Artistes Français (UCAF) hosted a daily, somtimes twice daily series of demonstrations of the Ondes and concerts by what was billed as the 'first Ondes orchestra' – actually an ensemble of up to eight instruments, with occasional guests on other instruments, all played by women in flowing white Grecian gowns (seen in the newsreel mentioned above). The concerts were conducted by Ginette Martenot (1902-1996), sister of the inventor, while the demonstrations, billed as 'humorous', were given by another Ondes specialist, Darius Cittanova (1905-1994). The demonstrations went on for not quite a fortnight; no programmes have been found but the contents can probably be guessed from his one commercial record, a 25-cm Pathé disc with 'The Seals' Rumba' on one side and impressions of fairground music and a 'Three Little Piggies' cartoon on the other. The concerts, ranging very widely in repertoire, lasted until almost the end of the Expo – long enough one for commentator to wonder if the Ondes hadn't lost its novelty value. Be that as it may, for the Expo the Martenots produced a publicity record promoting the Ondes, which was given away gratis and entitled the bearer to 'two complimentary entries', presumably to events in the UCAF pavilion. The disc was of an unusally small size, 15 cm, and was attached to a small brochure with a text by Alfred Cortot, no less, and programmes of the Expo's Ondes concerts (no copy of this brochure has been located). As a photograph shows, one label was of a very simple and striking design, while the other listed existing commercial discs of the Ondes (including Cittanova's), though not the contents of the promo disc itself. These were a technical description of the Ondes on one side and a demonstration of its capabilities on the other. Finally, the matrix numbers, 11764 AB / 11765 AB, show that the recording was made by Bernard Roux, while the combination of small-diameter disc and brochure, as well as the black-on-gold labels, are highly reminiscent of the 'Programme Sonore' film-souvenir discs produced by Le Disque de France (see above).

Clients: film

As early as mid-1936, it was reported that a new sound stage at Paris's Billancourt film studios had been equipped with variable reverberation, but the inventor(s) and manufacturer were not named. It would be suprising if no other production facilities or companies considered the Gamzon-Sollima device; but, for the moment, lack of online access to most French cinema periodicals means that nothing is known about its use by the film industry in the first years after its launch. Fortunately, one film scholar has documented its adoption by at least one producer soon after the start of the German Occupation, for the somewhat counter-intuitive purpose of dubbing German-language films into French. The problem was, it seems, that the acoustics of newly dubbed dialogue often failed to match those of the source films' original scenes. The Gamzon-Sollima device was the ideal tool for remedying this – but filmmakers also soon saw that it allowed to them to experiment with sound in novel ways:

The Roux technology allowed filmmakers to increase the ratio of reflected to direct sound radically beyond what had been imaginable, and the simplicity of the control panel made experimentation easy. In short, rather than simply allowing filmmakers to perform old tasks with greater efficiency, the Roux system cleared a space for novel and unforseen artistic practices.

Among the films in which this unsuspected potential was creatively tapped were Le Colonel Chabert (1943, dir. René Le Hénaff) and Le Corbeau (1943, dir. Henri-Georges Clouzot). These films were made by different production companies; it would be interesting to know how widely the device was used, and how long for: another film is also mentioned, made well after the War, Hôtel d’Invalides (1952, dir. Georges Franju), although it is unclear if this also used the Roux system. It is disappointing that, in the quarter-century since this research was published, the periodical sources on which it drew have still not been digitised and made accessible. If they ever are, the reception of the Gamzon-Sollima device by filmmakers and viewers, as well as other instances of its use, may one day be better documented.

Clients: recording

Perhaps even more surprising than the appearance of Bernard Roux's name in the history of film is the discovery that his firm's reverberation device was bought and used by one of the leaders of the French record industry, with whom Roux and his clients had been small-time competitors not many years before. In the spring of 1939, a magazine for young people published an article on the making of gramophone records, which revealed that what by then had become IME- (i.e. EMI-) Pathé Marconi was using the Gamzon-Sollima device at its production centre and factory in Chatou, outside Paris. To illustrate what it could do, the writer chose a rather alarming example:

Suppose we are to record some organ music in a cathedral. No need, these days, to take ourselves off for real to the vaulted spaces of a majestic Gothic monument; an ingenious artificial "reverberation" system, invented by Messrs. Roux and Solima [sic] will allow us to do our work at the factory.

One hopes that Pathé, which made quite a few notable organ recordings, did not in fact pass off any instruments recorded in 'a totally dead studio' (un studio complètement amorti) as cathedral organs; a later passage in the article does suggest the writer was thinking of special effects rather than full-scale musical productions. To date, no other reference to Pathé's adoption and use of the Gamzon-Sollima device has been found. It is not known when it was decommissioned and replaced, and whether any element of it survives. In 2004 the last remaining section of the Chatou factory, closed since 1992, was demolished. Whether any documents in the Pathé archives preserved at the Fondation Jérôme Seydoux-Pathé in Paris relate or refer to Bernard Roux has not been ascertained.

No evidence has been found for the Gamzon-Sollima device's adoption by any other company.

Acknowledgements

Customer services, Department of Image and Digital Services, Bibliothèque nationale de France, Paris
Henri Chamoux, France
Lee Davis, Houghton Library, Harvard University
Dr. Maureen Jackson, USA
David Mason, UK
Jerome Moncada, France
Andreas Schmauder, Phonopassion, Horben, Germany
Philip Stuart, UK
Jonathan Summers, British Library, London