1 Context and motivation
The Alliance of Ten Salt-making Provinces of Japan (1809–1888) data set was produced in the first phase of the European Research Council (ERC) funded research project entitled Beyond Eureka: The Foundations of Japan’s Industrialization, 1800–1885 (J-InnovaTech). Developed under the Horizon 2020 framework from 2020 to 2027, the goal of the project was to stimulate a shift in the approach to the study of Japan’s early industry and industrialization process by combining history-of-technology approaches with selectively deployed quantitative historical methods. Drawing on the historical material from 19th century Japan, J-InnovaTech’s principal intervention is in providing a more nuanced understanding of diversity and fine mechanics of innovation processes in the human past. While the classic imaginary of innovation is that of a breakthrough resulting from the invention of a solitary genius — the Eureka model — this project to colloquial: brings forth historical examples to complement a growing body of literature in social sciences indicating the importance and impact of distributed innovation patterns and open innovation regimes. Distributed innovation refers to a phenomenon in which actors draw on resources (technical and human) from multiple socio-political spheres or production sectors to achieve a shared goal. Open innovation refers to the phenomenon in which new production or design techniques circulate in the absence of patent systems or equivalent knowledge ownership regimes. Both concepts are empirically rooted in communities of production and challenge the notion of competition as the dominant structuring force of industry. A community of production refers to a group of actors working in collaboration towards an agreed goal within a specific production context while also competing for shared resources and/or markets. Distinct but connected, these two kinds of innovation in “production thinking” were key to Japan’s industrial capacity-building prior to the opening of its ports and its markets in the 1860s. Yet, they remain empirically poorly understood, in part because they do not fit with the prevailing understandings of innovation.
The present data set pertains to one such example of innovation in the salt industry, a strategic production sector in early modern Japan with a vital role in the economic development of the country. The innovation consisted of the change in the industry’s understanding of itself. This happened in a process of rethinking its functioning in society and working together to deliberately slow down the production. This was done by the producers coordinating the shift through yearly consultations. The aim of the alliance was to foster stability in an industrial sector that had gone through a dramatic period of boom-and-bust in the second half of the 18th century. Seeking to avoid reentry into a boom-and-bust cycle and cultivate a degree of stability in production, producers started getting together around the idea of reducing overall industrial output as a way of avoiding a repeat. This meant coordinating a moratorium on salt production for several months each year across as many salt-producing zones as were willing to take part in the alliance.
The idea of a production slow-down, a concept contemporary actors referred to as the “resting fields principle,” first emerged in the 1770s. It was formalized and institutionalized around the turn of the 19th century in the format of the Resting Fields Alliance. The building of the alliance, which involved the coming together of all salt producers active in Japan’s Seto Inland Sea, amounts to social innovation with significant industrial impact. The alliance grew in size and therefore in efficacy and came to encompass all ten salt-making provinces of the Seto Inland Sea basin by the 1840s. This Resting Fields Alliance can be said to have lasted in its original form until 1888, the year in which the Meiji government took a more hands-on role in coordinating the industry while keeping in place part of the social and organizational infrastructure along with most of the operating assumptions.
The emergence and expansion of the Alliance left a documentary trace. Much information about the early efforts to get the alliance going and grow it comes from an early 19th-century technical treatise on salt-making entitled The Secrets of Making Salt, written by Miura Genzō (三浦源蔵 (?–1835), a little-known salt entrepreneur from the Mitajiri region, one of Japan’s major salt-producing industrial zones well into the mid-20th century. The main goal of the treaties was to argue for the introduction of an innovative salt-producing technique: the resting field principle. In the process, the treatise gives an account of the efforts of producers to form an alliance among themselves dating back to the last quarter of the 18th century. Historical sources about the alliance survived in the archives of the Nozaki family, a family of salt entrepreneurs who were latecomers to the alliance when they joined in 1842 but became one of most diligent attendees of annual meetings from the time they joined until 1876. In addition, Nozaki family documents included important data points on the functioning of the alliance through decades; these documents were transcribed and included in the Compendium of Historical Sources on the Salt Industry of Japan.
The records relative to the Resting Fields Alliance of Ten Salt-making Provinces of Japan were transcribed and included in the Great Encyclopedia on Japanese Salt-making (Engyō Taikei 塩業大系). This marks the first wave of historiographical interest in the alliances. However, this compilation was not used because the currents of history in Japan went in a different direction when, starting in the 1980s and maturing in the 1990s, a methodological shift in the historical discipline such as it is practiced in Japan got greatly invested in research on the margins of the social status in Japan during its period of being the society of orders. Salt-making being inconsequential to such debates, it did not receive much research attention. It was only in the early 2010s that these documents started attracting research attention of Japanese scholars again.
With the exception of Watanabe Norifumi and the social historiography of Japanese salt published in the Social History of Technology (Nihon Gijutsu no Shakai Shi 日本技術の社会史) of 1985 (Nagahara & Yamaguchi, 1985), no major work had been done on salt since. Case studies were produced by Ochiai Kō pertaining to the salt fields of the city of Edo (present-day Tokyo) in 1999 (Ochiai, 1999) and later on the salt fields of Takehara in Hiroshima Domain in 2010 (Ochiai, 2010), in which alliance-related documents were not used. Most recently, Itō Akihiro published one article on the salt fields of Suō (Itō, 1999). In 2006, Yamashita (2006) proposed a comprehensive case study of the trading of salt and soy sauce in the domain of Akō which also makes no use of alliance-related sources.
The historiography above had made no use of documents concerning The Resting Fields Alliance of Ten Salt-making Provinces of Japan. Transcribed and published in the 1970s, it was practically set aside until the 2010s. In 2013, Nishihata Toshiaki published Early Modern Hydraulic Salt-making, a monograph about structure of salt production in the Bingo province which includes a chapter that draws on the use of alliance-related sources (Nishihata, 2013). Ochiai Kō published using those documents but with a goal of analyzing the network of the salt producers of Seto established via the alliance. (Ochiai, 2020). However, even Ochiai’s use of this source pool remained limited to a few documents and centered around the Hiroshima domain. In 2022 he published a paper on the Alliance during the early Meiji period (Ochiai, 2022b) before going back to case studies on salt fields in Harima and more recently in Kojima, province of Bizen (Ochiai, 2022a).
The data we propose in this paper are the results of the first quantitative approach to the Alliance of the ten salt-making provinces of Japan. To date, the Japanese researchers, such as Ochiai Kō, remain focused on qualitative historical approaches. We argue that a quantitative historical approach to the Alliance can complement traditional historical approaches and thus providing further nuance to of the Alliance.
The data set under discussion here uses the minutes of the annual meetings of the alliance from 1809 to 1888. These documents provide a wealth of information on the political context in which the Alliance emerged and how it grew over time. In this discussion paper Kinsei (近世) in Japanese refers to the Edo period (1600–1868) while the Kindai (近代) refers, in this context, to the Meiji period (1868–1912).
The innovation process that led to the alliance becoming one which encompassed all ten salt-producing provinces was a process that took decades and deserves careful historical attention. Consequently, the data set described in this paper was designed to enable and facilitate the use of new quantitative historical tools and methods to complement classical historical methods in advancing historical understanding of the phenomenon of the production slow-down by industry-wide consensus.
2 Dataset description
Repository location
Repository name
Nakala
Object name
actors.json, meetings.json, salt_sites.json, actors.csv, meetings.csv, salt_sites.csv
Format names and versions
JSON, CSV.
Creation dates
From 2021-03-16 to 2025-11-15
Dataset creators
Mathieu Fauré – Transcription, Data curation, Hiroki Yamashita – Data curation
Language
English and Japanese
License
Data set: Open Data Commons Open Database License v1.0
Publication date
2025-11-17
3 Method
Figure 1 shows the overall data production procedure. We followed a two-step process to create the dataset. First, data harvested from archival sources were entered into Heurist, a relational database platform in a collaborative environment. Second, the data were cleaned and consolidated for publication.
Figure 1
An overview of the data production procedure. Data harvested from archival sources were entered into Heurist, a relational database platform in a collaborative environment. Then, the data were cleaned and consolidated for publication. Red arrows indicate steps carried out by the authors.
The first step in data production was tagging the names of participants for each identified meeting of the Alliance from the meeting minutes found in the historical sources listed below. Typically, meeting minutes provide the names and places of representation for each individual participant, meaning the salt field, village, or province represented by the actor (not their place of family origin). The second step in data production was to produce as complete data as possible on the social profile of the participants. We cross-referenced the names of actors with Kaei Yo-nen Torikawase Shōmon no Koto to obtain additional information on their status specifically their social responsibilities (state offices, village offices, etc.). This work enabled us to fill gaps and compile more complete social profiles of alliance participants. We also categorized the types of salt-production sites (salt production sites or zones) and identified their geographical coordinates using online mapping tools for sites whose names are provided in the sources and could be identified in other materials. The data set was produced by harvesting data from the following sources:
Miura Genzō 三浦源蔵 Ensei Hiroku (Secrets of Salt Production 塩成秘録), in Nihon Shomin Seikatsu Shiryō Shūsei 日本庶民生活史料集成 (Historical Materials on the Life of the Commoners in Japan), edited by Watanabe Norifumi 渡辺則文, vol. 10, San.ichi shobo, Tokyo, 1994 [1970], p. 476–481.
Kaei Yo-nen Torikawase Shōmon no Koto 嘉永四年「為取替証文之事」(Compendium of exchanged documents), document cited in Ochiai Kō 落合功, Kinsei Seto-uchi Engyō-shi no Kenkyū 近世瀬戸内塩業史の研究 (History of Salt Production in the Seto Inland Sea in the Early Modern Period), Tokyo, Azekura Shobō, 2010, p. 239–242.
Meeting minutes of the Alliance of Ten Provinces, Nozaki Family Archives, transcribed and published in Nihon Engyō Taikei Henshū I.inkai 日本塩業大系編集委員会 (Editorial Board of the Compendium of Historical Sources on the Salt Industry of Japan) (eds.), Nihon Engyō Taikei Shiryō-hen Kinsei 日本塩業大系史料編近世 (Great Encyclopedia of Japanese Salt-making, Early Modern Period), vol. 2, Tokyo, Nihonsenbai Kōsha, 1976, p. 997–1086.
Meeting minutes of the Alliance of Ten Provinces, Nozaki Family Archives, transcribed and published in Nihon Engyō Taikei Henshū I.inkai 日本塩業大系編集委員会 (Editorial Board of the Compendium of Historical Sources on the Salt Industry of Japan) (eds.), Nihon Engyō Taikei Shiryō-hen Kingendai 日本塩業大系史料編近現代 (Great Encyclopedia of Japanese Salt-making, Early Modern, Modern Period), vol. 1, Tokyo, Nihonsenbai Kōsha, 1976, p. 1–614.
Nihon Chimei Taikei 日本地名大系 (Dictionary of toponyms of Japan) Heibonsha, (http://www.japanknowledge.com).
Salt sites mentioned in meeting minutes were cross-referenced with Nihon Chimei Taikei to identify their present-day location. Using Google Maps, we identified their coordinates. As for the meeting dates, they were recorded using the Japanese era-based calendar system. Japanese dates were converted to Gregorian dates using HuTime’s API. Details of the conversions and variables relating to the dates are explained in Section 4.1.1.
The harvested data from the archives were entered in Heurist, an open-source relational database hosted by CNRS’s Huma-Num. This platform was developed by Ian Johnson, designed to address research challenges in the social sciences and humanities. There are two notable benefits of the platform for researchers in the humanities. One is that it requires little SQL knowledge to manage and edit the dataset on the platform. The other is that it supports simultaneous editing of the data while maintaining the coherence of the data inputs. We created JSON files from Heurist and cleaned them. Heurist allows researchers to download JSON files of tables in a relational database. Cleaning improves readability, as original JSON files generated from Heurist include many keys that originate from the platform-specific structure of Heurist. Additionally, data entry was initially done in French and Japanese. French text was translated into English at this stage. We prepared datasets on meetings, actors, and salt sites in JSON and CSV formats. The JSON format has an advantage over CSV in terms of machine readability, while the latter is more readable for humans. Data on meeting dates and locations, actor names and offices, and salt site names and coordinates are nested in JSON and flattened in CSV. In JSON, multiple elements are stored in lists, whereas in CSV they are separated by semicolons. For details of the variables, see the codebook in the repository.
4 Data details and repairs
4.1 Meetings
The dataset contains 60 meeting records from 1809 to 1888, 34 of which were from Nippon Engyō Taikei Kin.gendai Shiryō, 23 taken from Nippon Engyō Taikei Kinsei Shiryō and three from Ensei Hiroku. 50 meetings were held across seven different domains at 12 locations and 10 meetings were held at unidentified locations. The top panel of Figure 4 shows the change over time in the number of participants for each meeting.
4.1.1 POSIX — standard handling of Japanese dates
As for meeting dates, the archival sources record them in the Japanese calendar system. To facilitate use by a wider audience, we have used HuTime1 to convert the source dates to the Gregorian dates. Some meeting records provide only the year and month of the meetings. In those cases, we recorded a potential range of dates for the meetings in two variables, “date_range_start” and “date_range_end” in the ISO date format, so that end users of the data can choose how to represent the date, for instance, by using the midpoint of the range. When an exact date is available, we recorded it the “exact_date” variable using the ISO date format.
4.2 Actors
Actors.json and actors.csv contain a total of 866 actors mentioned in the historical sources. Actors are generally recorded as individuals (by personal name), not a firm or family, except for certain non-human actors, as explained bellow. In addition to the individual participants in the meetings, those who are mentioned as being represented by another person are also included.
4.2.1 Unnamed actors
We have kept all the records of attendees as close as possible to the original archival sources. This means that the data set of actors also includes actors referred to as hikyaku, a Japanese courier or representative of the region whose name is not provided. Facilitate analysis focusing only on human actors, we added the “is_name” variable to actors.json and actors.csv. The datasets include 832 human actors and 34 non-human actors.
4.2.2 Inconsistent spelling of actor names
Even for human actors, from one set of meeting minutes to another, we identified cases where entries appear to refer to the same individual: identical associated sites, names differing by only one character, and apparently identical status. The repetition of these names within a series of successive reports (a sequence of reports close in time) associated with the documents highlighted by Ochiai (2010) allowed us to identified duplicate entries as referring to the same person. In this case, we merged the entries into a single actor record under a single name, using the name recorded in the most recent source. In cases where several duplicates could not be identified as the same person, due to disparities in the reports and/or too large a time gap between appearances, we considerded them as separate actor records. Additionally, we identified that some Nozaki family members changed their names in the course of their lives.2 Nozaki Bunzaemon (野崎文左衛門) changed his name to Buzaemon (武左衛門) in 1841, and after his death in 1864, his grandson Nozaki Bukichiro (野崎武吉郎) started using his grandfather’s name, becoming himself Buzaemon (武左衛門) until 1869. Thus, all three names appear in the archival sources. We have standardized the name of the Nozaki person who attended meetings from 1841 until 1860, to Nozaki Buzaemon (野崎武左衛門), and his grandson, who attended the meetings from 1864 until 1876, to Nozaki Bukichiro (野崎武吉郎). Nozaki Zaijirō (野崎材次郎) also changed his name to Shinshichirō (晋七郎) and integrated himself into the Ogino family. We have recorded him as Ogino Shinshichirō (荻野晋七郎).
4.3 Salt sites
The dataset contain 97 salt sites. They may not cover all the salt sites that existed during the period. However, we assume that we have captured the most important salt-production sites, as their representatives took part in the series of meetings with historical importance that were held throughout a great political regime change, the Meiji restoration. Figure 2 shows a map of the identified salt sites, as well as the meeting places.
Figure 2
Map of the Seto Inland Sea region in Japan. The highlighted area shows the ten domains in the Salt Alliance. Meeting locations (except Anyoji, where we could not identify the present-day location) are shown as red dots. Orange dots and grey areas indicate salt sites and salt production zones, respectively.
4.3.1 Site type
As with the personal-name category, we kept the records as close as possible to the original historical source. Of the 97 identified sites, 46 entries (e.g., “Unspecified Awa salt field”) could not be geolocated. We geolocated 42 salt sites and nine production zones. Salt sites refer to specific salt-production fields, whereas zones designate larger areas that typically includes multiple salt-production sites.
5 Results and discussion
5.1 Tempō famine (1833–1839) as a major moment of rupture and reconfiguration
Traditional historical analysis reveals that over seven decades, the alliance grew in size and came to encompass all ten salt-making provinces of the Seto Inland Sea basin by the 1840s. However, quantitative historical analysis enables a more nuanced understanding of the pivotal moments within this period by providing insight into changes over time in the structure and composition of the meeting participation and venues.
For example, the data reveal that one major turn came in the 1830s. While traditional historical analysis readily reveal a gap in meetings between 1832 and 1841, quantitative data analysis allows us to charactererize the nature of the rupture. More precisely, the ratio of participants per domain provides evidence of the shrinkage in the footprint of the Hiroshima domain, where the Mitajiri salt zone was located, within the Resting Fields Alliance. If we look at the ratio of participants per domain (han), Hiroshima systematically had more men in the room before the Tempō famine. After the famine, Hiroshima-domain representatives comprised the largest delegation for only six years in the four decades of alliance meetings post-Tempō famine.
More evidence of the rupture and reconfiguration can be found in the low continuity between pre-famine and post-famine participants in meetings. We constructed a network of meetings by treating meetings as nodes and the existence of common participants as edges. The weight of an edge is calculated based on the number of participants they have in common. For instance, if five participants attended meeting A and meeting B, then an edge with weight 5 is constructed between the two meetings. The network indicates two lines corresponding to a single actor, a certain Kunio Katsusaburō (都穂克三郎), who attended two meetings before the famine (1818, 1821) and the first post-famine meeting in 1841. We refer to him as a “continuity actor” — someone who was there in the early pivotal years as well as in the first post-famine meeting, which was some twenty years after his last participation in a meeting. Next, we applied the Louvain algorithm (Blondel et al., 2008) to find four clusters with a modularity of 0.394. Figure 3 shows the network of meetings and its clusters. The clusters correspond to a chronological phases in the alliance. The first cluster spans the meetings from 1809 to 1832, the second from 1841 to 1874, the third from 1875 to 1883, and the last cluster comprises meetings from 1884 to 1888. The separation between the first and the second clusters corresponds to the rupture caused by the Tempō famine.
Figure 3
The clusters of meetings based on the Louvain algorithm. Each node represents a meeting, and each edge represents the existence of participants they have in common. The size of the nodes shows the number of participants in the meeting, and the color shows the cluster. The thickness of the edges corresponds to the number of participants two meetings have in common. The Tempō 12 meeting node has a single shared participant with the pre-famine cluster, shown in red.
5.2 The 1875 shift
Figure 4 shows the change in the number of participants in each of the meetings and the change in the average number of salt sites represented by a participant. From the graphical representation, we can observe three changes from 1875.
Figure 4
The top panel shows the change in the number of participants for each meeting. The colors of points correspond to the cluster from the community detection algorithm. On the panel at the bottom, the y-axis shows the average salt sites for each participant, and the solid line shows the trend calculated by LOWESS (locally weighted scatterplot smoothing).
A first change concerns the number of meetings. From 1875 onward, the meetings of the alliance began to be held more than twice a year. Our dataset contains a total of 60 mettings between 1809 et 1888 but almost half of the meetings (27) took place after 1875.
The second change is the number of participants in the meeting. The meeting of the Alliance Meiji 8th (1875) in October had the highest number of attendees with 74 participants in attendance. On average, meetings had 18.48 participants before 1875 and 31.44 participants after 1875.
The last change concerns the representation of salt sites by participants. From 1875 onward, we observe an increase in the average number of salt sites represented by meeting participants. Meeting for the Rules and Regulation of the Salt Producers Association in Meiji 17 (1884) in July is the first to record an average higher than 1, meaning one participant was representing more than one salt site on average.
These three changes show that the salt alliance meeting was more frequently held, more participants were attending them, and a greater variety of salt sites were represented in the meetings. This suggests that they were gaining more importance in coordinating salt production and knowledge sharing.
5.3 Nozaki family: a relatively latecomer turned important figure
Another major advantage of the digitized and standardized dataset compared to classic archival data is that it enables the use of natural language processing techniques and visualization tools. This section demonstrate this by analyzing the participation of members of the Nozaki family in Alliance meetings. The Nozakis were a family of salt entrepreneurs who were relatively latecomers to salt-making and to the alliance. They entered salt making only a few years after the famine but by its end, they had become a major player figure in the industry by virtue of operating over 65 salt fields by 1841, a point when they joined in the alliance as it resumed its meeting post-famine. While being new to the alliance, they became one of the most diligent attendees of alliance meetings from 1841 to 1876.
Figure 5 shows the first year of participation in alliance meetings and the duration of participation for all participants. Members of the Nozaki family are shown in red. There are five family members. Nozaki Buzaemon participated in the meeting nine times from 1841 to 1860. Other members did not necessarily frequent the meetings. However, we observe that other family members took over participation in meetings after Nozaki Buzaemon, and the family’s presence was ensured throughout the early 1870s, when we observed a change in the frequency, size, and influence of meetings, as we have seen in the previous section.
Figure 5
Visualization of the year of entry and length of participation in alliance meetings. Each dot represents a participant and year of their first attendance to the Alliance meetings, and lines represent their duration of participation in them. Dots and lines in red represent participants from the Nozaki family.
5.4 Applications
This data set is the first in a series of dataset produced within the framework of a research project entitled Beyond Eureka: The Foundations of Japan’s Industrialization, 1800–1885 (J-InnovaTech). It is geared towards helping develop a better understanding of the innovative ways in which salt entrepreneurs reorganized production in order to avoid boom-and-bust cycles the industry encountered in the second half of the 18th century. The dataset was produced and made available with three principal goals in mind. First, proposing a replicability standard in the discipline of history. It seeks to provide full transparency and replicability of quantitative elements used in the historical analysis that underpins publications coming from this research project. In so doing, the authors seek to be part of a move towards new standards of referencing in the discipline of history. While replicability of the sort described in this paper may be part of standard practice in other social sciences, this is not the case for the general field of history. Second, this data set hopes to provide points for comparative research on salt, producer networks, or on the processes of innovation as part of the project orientation towards future data-driven academic collaboration.
In that context, two kinds of application are of particular pertinence. First, this data set can contribute to research on organizational innovation, especially research on emergence of social institutions aimed at creating new spaces of economic and industrial productivity. The concept of “commonning” has been coined and deployed in describing such organisational innovation and social institution-building in economic history, history of economic thought, and in management studies.3
Second, this data set can also be used as a point of comparison and contrast in studies on the histories of early modern guilds or contemporary cartels. In terms of structure and function, an alliance of the kind described by this data set has certain features in common with both early modern guilds and with modern-day cartels, while being distinct from both. Like guilds, the Resting Fields Alliance provides structures to exchange information about markets, prices, and innovations in production technology. In fact, information exchange is a fixture of the annual meetings that this data set describes. However, unlike guilds, the alliance did not impose limits (numerus clausus) on entry into the structure. On the contrary, the goal of the Alliance is to enroll as many producers as possible into the discussion. Similarly to cartels, Resting Fields Alliance members work jointly to regulate the industry, and by extension exercise some degree of control over supply supply. However, unlike cartels that aim to maintain prices and profit margins, Resting Fields Alliance producers operated within a distribution system in which they had little to no control over supply-side pricing. Setting the wholesale buying price of salt at production point seems to have been de facto a prerogative of merchant houses of Osaka and their network of regional wholesaler collaborators. This explains why the words “price” or “profit” rarely appear in the Resting Fields Alliance meeting minutes in a way it would in the discussion of contemporary cartels.
Notes
[2] A study of Bizen Kojima Nozaki family ―History of establishment of Naikai salt inc. [備前児島野崎家の研究―ナイカイ塩業株式会社成立史] published by Naikai Engyō Kabushiki Gaisha [ナイカイ塩業株式会社] (eds.), p. 36–37.
[3] For work on commoning economies see Commonism: A New Aesthetics of the Real. (Dockx & Gielen, 2018). On the concept of commons as a social system beyond natural resources management see De Angelis (2017). On current examples of theoretical approaches to commons outside economic and environmental history approaches to natural resources see Multitude: War and Democracy in the Age of Empire (Hardt et al., 2004) and Hardt & Negri (2009).
Acknowledgements
The authors would like to thank Catherine Bastien-Ventura, Amélie Marineau-Pelletier, Éric Mermet, Mark Ravina, and Dimitri Tatoyan for their comments on different drafts of this paper. Special gratitude goes to Professor Ochiai Kō of Aoyama Gakuin for his kind guidance throughout the researched leading up to this dicussion paper.
Competing Interests
The authors have no competing interests to declare.
Author Contributions
Aleksandra Kobiljski: Conceptualization, Funding acquisition, Investigation, Project administration, Resources, Supervision, Formal analysis, Validation, Writing — Original draft (in English), Writing — revisions and editing.
Mathieu Fauré: Data curation, Methodology, Investigation, Conceptualization, Resources, Writing – Original draft (in French).
Hiroki Yamashita: Data curation, Methodology, Formal analysis, Software, Writing — Original draft (in English), Writing — revisions and editing, Visualization.