by Patrizia Rebulla, Valeria Luti and George Ionita
1. Presentation of the Archivio Storico Ricordi
Compared with the rest of the music field, the music publishing industry is relatively little studied, in part because of the scarcity of primary sources available which shed light on its history and decision-making processes. The Archivio Storico Ricordi, however, provides a remarkably rich and complete insight into two hundred years of music publishing.
In the nineteenth century, Ricordi was one of the largest music publishing houses in the world, and its network extended across the continents. Ricordi was particularly famous for Italian opera, and Rossini, Bellini, Donizetti, Verdi and Puccini, to name a few, were among its composers. The company was therefore particularly involved in the theatre industry. The theatre business is a multi-layered enterprise which actively involves not only company managers and their skills but also composers and librettists, stage designers, illustrators, conductors, singers, artists, managers and theatre impresarios, not forgetting the private customers who purchase scores or librettos, and journalists who regularly update the audience with reviews and information on new releases. This is why the Archive is so rich and varied.
Currently, the Archive holds some 8,000 scores (see Figures 1 and 2), and more than 13,000 iconographic documents, such as for stage design, costumes and any kind of stage fittings to set an opera (Figures 3 and 4):
Figure 1: Giacomo Puccini, La Bohème autograph score, 1896
Figure 2: Giuseppe Verdi, Otello autograph score, 1887
Figure 3: La Bohème by Giacomo Puccini, Turin, Teatro Regio, 1896 Act 2, set design by Adolf Hohenstein
Figure 4: Costume and prop designs
Of course, operas also mean libretti (see Figures 5 and 6), including both print and manuscript copies, some of which have been carefully annotated by company managers, who always attended débuts in theatre:
Figure 5: La Bohème by Giacomo Puccini: handwritten libretto with autograph annotations by Luigi Illica and Giacomo Puccini
Figure 6: La Sonnambula by Vincenzo Bellini: libretto with autograph annotations by Giovanni Ricordi, Milan, Teatro Carcano, 1834
Beyond the libretti, the Archive also keeps some rare ‘staging notes’, again annotated by managers:
Figure 7: Aida by Giuseppe Verdi, Cairo, Khedivial Opera House, 1871, staging notes by Giulio Ricordi
And of course, even before Instagram, the life of the company was documented through photographs of the company’s shops, composers, managers, performers, and theatre performances:
Figure 8: Ricordi store in London, early 20th century
Figure 9: Composers, managers, performers…
Figure 10: Gianni Schicchi by Giacomo Puccini, first performance, 1918
Nonetheless, one should not forget that Ricordi was a commercial company. Therefore, of course, accounts and administrative documents of all kinds are an invaluable part of the Archive: ledgers, inventories, catalogues, contracts, and the very rich amalgamation of forty years of proceedings of the administrative board and of the board of oversight (see Figures 11, 12, 13 and 14 below).
Figure 11: Page of Ricordi publishing catalogue, 1820
Figure 12: Summary of contractual conditions for purchased works
Figure 13: Administrative Board Proceedings, 1889-1919
Figure 14: Board of Oversight Proceedings 1900-1919
2. The Letters of the Casa Ricordi Project
Most significantly of all, the Archive holds an impressive quantity of letters. Before the invention of the telephone, things were done by writing them on paper. This is why (paradoxically) the business of the nineteenth century is, in a way, better documented than in the following years. We can still read, for example, the hundreds of letters that managers exchanged with Verdi; we don’t know details of what was agreed with other artists over the phone during the twentieth century. The Archive holds more than 30,000 received letters (see, for example, Figures 15 and 16).
Figure 15: Letter from Arrigo Boito to Giulio Ricordi, February 1881
Figure 16: Letter from Giacomo Puccini to Tito II Ricordi, 19th January 1907
The archival process for received letters is easy, since it is sufficient to keep them in a safe letter-box and catalogue basic metadata such as sender and date. But what about copies of sent letters, before the invention of carbon paper or the typewriter? Before being dispatched, letters were simply copied by way of a press into ‘copybooks,’ and each letter’s addressee was indexed in a specific index volume:
Figure 17: Press for copybooks, Museo Nazionale della Scienza e della Tecnologia Leonardo da Vinci, Milano
Figure 18: Copybook index page
These very rich indexes are extremely accurate and reliable ‘filters’, that allowed the employees of the house to retrieve a whole thread of letters exchanged between a given correspondent and the company. The Archive keeps copies of some 600,000 letters such as the one in Figure 19 below, stored in more than 1,200 surviving volumes and indexed in 73 index volumes.
Figure 19: Letter from Giulio Ricordi to Giuseppe Verdi, 11th February 1888
However, from this large volume of letters, the ‘pearls’ of the Archive are three private copybooks of letters sent by Giulio Ricordi and his son Tito II for some thirty years, between 1888 and 1918 (Figures 20 and 21). They tell the story of the company better than any historian, and are available online. These will undoubtedly be an invaluable tool of research for scholars and readers for years to come.
Figure 20: Copybooks of Giulio Ricordi, 1888-1909
Figure 21: Copybooks of Tito II Ricordi, 1912-1918
The letters document personal relations with artists (see Figure 22), as well as the quality of these relations, sometimes very familiar and informal (Figure 23), to the extent of becoming outspoken (Figure 24):
Figure 22: Example of a letter
Figure 23: Example of a letter
Figure 24: Example of a letter
They also document both the competence and self-confidence of the managers, who did not restrain from artistic interventions:
Figure 25: Example of a letter
Furthermore, they document the public image of the company in its relations with institutions (Figure 26) and political figures (Figure 27), taking political positions and fighting against competitors (Figure 28).
Figure 26: Example of a letter
Figure 27: Example of a letter
Figure 28: Example of a letter
They do of course discuss business, patents and money (Figure 29), and give strong and clear managerial direction (Figure 30), also addressing letters direct to the personnel (Figure 31).
Figure 29: Example of a letter
Figure 30: Example of a letter
Figure 31: Example of a letter
The company made use of a secret code (unfortunately lost and still not cracked), often used to keep details sent by telegrams confidential:
Figure 32: Example of a letter
Moreover, as the company was family-run, the letters also document family relations, sometimes difficult, until they almost broke down between father and son:
Figure 33: Example of a letter
Figure 34: Example of a letter
3. Towards the Website
Figure 35: Homepage of the website ‘The Letters of Casa Ricordi’
Received letters were digitised and catalogued with basic metadata (see Figure 36), as currently not all have been read. The Archive intends to add a transcription tool to allow for transcription of these letters; this is our project for 2019.
Figure 36: Letter page of The Letters of Casa Ricordi website
The three private copybooks of Giulio and Tito Ricordi were fully transcribed (Figure 37) and translated into English (Figure 38), and each letter was annotated, identifying (where reasonably possible) persons, places and musical works, and connecting these entities to the Archive’s database (Figure 39).
Figure 37: Letter page of The Letters of Casa Ricordi website
Figure 38: Letter page of The Letters of Casa Ricordi website
Figure 39: Letter page of The Letters of Casa Ricordi website
If a particular letter mentions another letter also held by the Archive, this is linked, in order to rebuild the conversation:
Figure 40: Letter page of The Letters of Casa Ricordi website
Persons and works are connected to VIAF and ISNI, and – when the entry exists – to the Italian national Encyclopedia Treccani, and to Wikipedia, due to its linguistic richness.
Figure 41: Letter page of The Letters of Casa Ricordi website
Obscure or unusual passages are explained (Figure 42), and places are georeferenced and, when possible, illustrated by pictures and plotted on a map (Figures 43 and 44).
Figure 42: Letter page of The Letters of Casa Ricordi website
Figure 43: Letter page of The Letters of Casa Ricordi website
Figure 44: Map of places of The Letters of Casa Ricordi website
Finally, copybook indexes are accessible and searchable on a specific page:
Figure 45: Copybook indexes page of The Letters of Casa Ricordi website
Last but not least, the whole project aims to be open-ended and welcoming to contributions from other archives and scholars around the world. The first, and excellent, example of cooperation began with the correspondence between composer Vincenzo Bellini and the founder of the company, Giovanni Ricordi, which is fully integrated into the website and connected to the Archivio Ricordi database. We hope that this sets a seminal and fruitful example for future collaborations.
Figure 46: Letter page of The Letters of Casa Ricordi website
4. How to Migrate a Standalone Database into a Live Website
Archivio Storico Ricordi was created in 1994, and from its very earliest days the necessity of using digital techniques to establish electronic catalogues and metadata was recognised as crucial, both to provide information to scholars and to track down the appropriate documents.
Figure 47: Paper card index
The Ricordi Archive preserves a huge amount of heterogeneous materials. Its information was transmitted from each archivist to the next without tools smart enough to substitute, even partially, individual knowledge. The digitization of Archivio Ricordi started about ten years ago, with the purpose of updating historical heritage preservation. Before this, the only way to “browse” the archive documents was through interaction with the archivists themselves.
Figure 48: From paper to Excel files
The first archivists only produced paper lists. These were later replaced by Microsoft Excel and Access files, which stored basic information about the materials, such as what they were and how to find them. It was still necessary for the researcher to know the peculiarities of a certain document to be sure it was the one they were looking for. However, it is crucial to allow access to the archive which does not depend on the archivist, not only for the sake of internal management but also for the conservation and sharing of historical heritage, which are the core valued of the Archive.
Figure 49: Archivio Storico Ricordi FileMaker Database homepage
With the improvement of the process, and the need to include more and deeper metadata about the documents preserved, the Archivio Ricordi created a management tool to make the search simpler and faster, and to allow for a more accurate cataloguing of the collections. In our case, that tool is FileMaker. FileMaker Pro is a cross-platform relational database application. It integrates a database engine with a graphical user interface and security features, allowing users to modify the database by dragging new elements into layouts, screens, or forms.
Our FileMaker database was initially created for the internal workflow, and its design was based on both scientific and management needs.
Figure 50: From Excel to FileMaker database
At the very core of this database were the above-mentioned Excel files, which were no longer able to provide the sustainability or functionality required, not only by the Archivio’s diverse collections of written documents, but also of photographs, scores and iconographic records.
Figure 51: Value lists to uniform data
Therefore, the focus was to restructure and make uniform the metadata in these datasets, making their structure more robust and consistent. It is widely established that the quality of metadata is crucial for obtaining exhaustive answers to a search. Thus, metadata was defined according to national and international cataloguing standards, in order to ensure an accurate description of each document.
Over time, we remodelled and adapted the data structure and metadata fields according to the needs of each type of collection we were cataloguing.
Figure 52: Score data entry interface
We identified the appropriate dataset for scores, letters, iconographies, etc., and shaped the database tables and interfaces accordingly. The result is highly accurate description forms; however, the limitation of this is that they can only be consulted and compiled inside the Filemaker application.
Figure 53: Archivio Storico Ricordi website homepage
Recently, the Company’s policy towards the online accessibility of the archive has evolved, making Archivio Ricordi’s collections available on the internet whilst still adhering to standards that make those collections interoperable and reusable. For this, the existing stand-alone database and the proprietary data formats, created for internal management purposes, are not appropriate, as it would be difficult to make its contents available online. The only sustainable solution is to export data to a new database which generates the website. The problem is working out how to ensure continuity without stopping the running machine and synchronising data across different technical solutions. One of the main issues is the duplication of data or the overwriting of new and updated information.
Therefore, the new goal is to convert the back-end FM tool into a web-based tool, which offers the biggest advantage of being accessible from anywhere, both in terms of consulting and data entry. We are currently in the process of migrating data to the new tool, and trying to integrate and incorporate previous incarnations of that dataset.
5. Tooling Up for Scholarly Editing
As the developer, my task was to provide a way to transform the archive into an open dataset that could be explored by a variety of users, including students, researchers, and music history enthusiasts. To do so, it was necessary to transform the archive into an interlinked system, with the collection of letters as the central piece, and a large number of related entities that appear in the letters, such as people, places, theatres, performances, manuscripts and stage descriptions. This task posed a number of challenges.
One of the challenges was understanding an already existing complex ontology – in Italian. Figure 54 is a picture of the database schema; as can be seen, this is a fairly complex ontology, and it was therefore important to understand how the collection was archived in the first place.
Figure 54: A complex ontology
Visiting the physical archive in Milano provided me with the necessary understanding of Casa Ricordi’s working processes. This was also an opportunity to clarify some of the database issues and unknowns with my colleague Valeria Luti, and to understand the decisions made when designing the database, which resulted in those issues occurring.
Another significant challenge was designing the system architecture with the final interface in mind. It wasn’t enough just to design a system that could link text and entities together; we had to anticipate what kind of explorations a user would expect from the dataset once all the data was linked up, for example, which entities would be accessible from which pages, which entities will be filtered by other entities, and what would those filters be, etc.
Working with an outdated database system was the biggest technical challenge. Anyone who has used FileMaker Pro before will know how difficult it is to work with. It only exports CSV files, and does not impose the same restrictions that an RDBMS would. This meant that a lot of work had to be carried out before we could use the database in a web environment. For example, we had to find workarounds for having multiple foreign keys in one field, stored as comma separated values. In a Relational Database, such a thing is not possible; instead, you would use a linker table to break the Many-To-Many relationship into two One-To-Many relationships.
There was also the fact that there were no existing tools for the editor to carry out the work with. The team’s vision was for a user to be able to read through a letter and simply click on a word within the letter (for example, a person) to display information about that entity. However, this link between the two items needs to be created somehow. How?
5.3 The Solution
Our solution was to create an Editing Workbench tool using CKEditor’s base code. CKEditor is a WYSWYG open-source editing tool, and is very customizable.
Figure 55: An instance of CKEditor
We were able to create an extension to the CKEditor functionality that would allow the editor to make a text selection and communicate with the database to create direct links in the form of an in-text annotation.It works as follows:
- The user makes a selection with the mouse, then presses the annotation button on the toolbar.
Figure 56: Annotating text in the editing workbench
- A panel is displayed where, the user can search for any entity in the database. As the user types each letter, an AJAX request to the server is fired and returns the list of results.
Figure 57: Searching for an entity to be annotated
- Once the entity is found and the user presses “OK”, a link is created between the document and the entity. An AJAX request is sent to the server, with the IDs of the document, entity, and an auto-generated link ID, so that each fragment of text is identifiable by an ID. This is useful when displaying the letter text to the user, allowing them to click on a link, and for that link to match a record in the database.
5.4 The Result
The result is a system that allows the exploration of all linked entities, beginning from any part of the collection, and which is able to filter by any of its related entities. Here are a few examples of explorations:
- For example, by accessing a letter you can see all its related entities on the right-hand side, as well as where they occur in the text.
Figure 58: The annotated entity on the live site
- Clicking on an underlined word in the text gives you the information for that item.
Figure 59: Viewing an annotation
- Clicking on ‘Otello’ then shows all information for that work, as well as its related letters where it was annotated.
Figure 60: Viewing an entity
- Because everything is interlinked, filtering by all related entities to this work is also possible. We used ElasticSearch for aggregating the results to create facets.
Figure 61: Example of facets
- Letters can also be retrieved by browsing a map of all theatres relating to them.
Figure 62: Retrieving a letter using a map
5.5 Going Forward
Using all the linked information, we are planning to implement visualisations, such as word-clouds, to further aid the exploration of the dataset.
Crowd sourced translations are also something we are considering implementing, as only a fraction of the letters have been translated. This is a task that poses its own set of challenges, but we hope to give users the opportunity to not only interact with, but play an active role in digitising, this fascinating piece of history.