50 Years of Editorial Practice: A Footnote Analysis of the Heinrich Bullinger Briefwechsel

The Footnote’s Origin and its Form

Footnotes in Central European printed books date back to the mid-17th century, emerging alongside the development of science and humanities (Freedman, 2020). Freedman defines footnotes by three key properties:

• Placement at the bottom of individual pages
• Arrangement in numerical, alphabetical, or symbolic order
• Correspondence between the footnote marker and its reference in the main text, usually on the same page

Despite this definition, footnote forms can vary depending on the base text.

The Footnotes’ Purposes and Analyses

Footnotes serve as more than mere remarks; they are tools with diverse functions across disciplines (Zerby, 2007). In computer science, they often provide technical details, while in theology, they reference, e.g., Bible passages. Historical research frequently uses footnotes for source references and extended discussions.

In epistemological writings, footnotes are often viewed as narrative and rhetorical devices (Grafton, 1999; Hammarfelt, 2010; Trüper, 2013). From a qualitative perspective, the presence and proper use of footnotes in historical research signifies professionalism and thorough scholarship (Grafton, 1999).

Quantitative research has explored the status and function of footnotes. Stevens and Williams (2006) examined the transmission of theory through footnotes in the journal Critical Inquiry from 1974 to 2004. They found that prominent scholars like Jacques Derrida, Sigmund Freud, and Michel Foucault were consistently cited, potentially forming a canon of expected references. However, they noted that citation frequency alone does not reveal how these works were interpreted or used.

Hammarfelt (2012a, 2012b) extended this approach through bibliometric analysis of literary studies footnotes. By creating co-citation maps from Swedish literature journals (2000-2009), he identified clusters of frequently cited authors. This revealed networks of scholars who likely influenced each other and were considered important in their field.

These studies demonstrate the rich potential of footnote analysis for understanding scholarly communication, canon formation, and intellectual networks. However, they also highlight the challenges of interpreting citation patterns without deeper contextual analysis.

The Bullinger Digital Project – Data for the Footnote Analysis

The Bullinger Digital project, initiated in 2020, began with digitising and crowd-correcting index cards containing letter metadata. This formed the basis for further processing, including text and footnote extraction from editions.

We utilised PDF files from the 2013 digitisation by arpa Data GmbH,⁴  along with TUSTEP files where available, to extract texts and footnotes. Challenges in footnote extraction included varying layouts, multi-page footnotes, nested footnotes, OCR errors, and overflow of the footnote apparatus of one letter into the one of the subsequent letters. We converted the extracted data to TEI/XML format, employing control scripts and manual intervention when necessary.⁵ 

The general project combines these digitised editions with manual and automatic transcriptions of unedited letters, creating a comprehensive, searchable interface.⁶  All data, including footnotes, is available on GitHub.⁷ 

Footnote Length and Density

As alluded to above, factual footnotes are far from homogeneous and vary in their content and size. The distribution of footnote length is extremely skewed (see left side in Figure 2). While a small number of footnotes exceed 500 tokens, approximately 93% of footnotes are shorter than 50 words. This reflects a high degree of variability in the length of annotations, where most footnotes remain brief, but a handful serve as more detailed, extensive commentary.

Another way to analyze editorial practice is the density of footnotes, in other words, how many footnotes are set per sentence of the original text. The right side of Figure 2 shows the number of footnotes per sentence. The distribution is skewed: while 75% of the letters contain fewer than one footnote per sentence, a handful of letters have more than three footnotes per sentence.

Footnote Content

We automatically categorized footnotes by function using simple heuristics (recurring patterns allowed the use of regular expressions), identifying seven types:

• References within HBBW (inner_ref):Referring to a footnote or letter in another volume of the edition.
• References to same edition in HBBW (self_ref):Referring to a footnote or letter in the same volume.
• Unknowns (missing):Hinting at lost information known to the letter writers, such as an unpreserved letter.
• Referencing Bible verses (bible):Contextualizing the letter with an implied or mentioned Bible verse.
• Lexical references (lex or lex_dict):Short word explanations, often adding a modern variant or translation. The lex_dict category also cites a lexical resource.
• Miscellaneous:All other types.
• Short (short):Miscellaneous but shorter than five words.

Figure 3 shows the label distribution across each edition volume, highlighting both frequent and infrequent categories to ensure trends in less common labels are visible. Volumes were released sequentially by number between 1974 and 2022, reflecting nearly five decades of evolving editorial practices. Figure 4 complements this by outlining the composition of editorial teams for each volume, showcasing personnel changes over time.

While it is unsurprising that references to earlier volumes of the edition ( self_ref)were minimal in the early phases, the first volume seems to be an outlier for other categories as well. In the first volume, lexical footnotes were the most prevalent, whereas in later volumes, the miscellaneous category consistently became the most dominant. Similarly, the categories bibleand missingshow notable outliers in the first edition. The first edition can be interpreted as an experimental phase, while over time, editorial norms were established, leading to greater consistency in label usage.

In addition to these overall trends, significant breaks are observed in specific categories such as lex_dict(volume 7) and inner_ref (volume 16) .In volume 7, there is a sharp decline in the use of lexical footnotes referencing dictionaries, while references to the same edition experience a notable dip in volume 16. These deviations suggest shifts in editorial focus or strategy during those periods. The trend changes in volumes 7 and 16 do not co-occur with changes in the editorial team (see Figure 4). This points to strategic rather than personnel-driven shifts.

Figure 3 also highlights an inverse relationship between lexical and miscellaneous footnotes, particularly evident in volume 11. This volume stands out as an outlier, with an unusually high percentage of miscellaneous footnotes coupled with a low percentage of lexical ones. While this edition introduced two new editors, Moser and Kess, the same editorial team also worked on volumes 12 and 13 without repeating this pattern. As before, this suggests that the anomaly in volume 11 reflects a specific editorial decision rather than a direct result of the new team members.

Primary and Secondary Sources for HBBW

The previous section’s categorization left a significant residual category, divided into short and miscellaneous groups, which requires further analysis. A promising approach to explore these footnotes is to examine the cited literature. In the digital edition, citations are consistently marked with XML tags, and source abbreviations are standardized, making it easier to analyze the references systematically.

The reference list includes over 700 entries, but citations are highly unevenly distributed, with 50% coming from just 17 sources. As illustrated in Figure 6, most frequently cited works are primary sources, with only four of the top-cited sources being secondary literature. This suggests the editors predominantly rely on materials contemporary to Bullinger’s time. A particularly notable case is Erasmus of Rotterdam, whose works (Adagia, LB, ASD) are cited across three different editions. When combined, Erasmus's works emerge as the most frequently cited, highlighting his significant influence on Bullinger and other reformers of the time.

Additionally, this prominence likely also stems from the accessibility and reliability of Erasmus's edited works, which serve as convenient resources for the HBBW editors. Similarly, the collected works of Zwingli (Z), Martin Luther (WA), Martin Bucer (Bucer DS), and the Blarer correspondence (Blarer BW) are highly cited, demonstrating the edition’s reliance on previously edited primary sources.

Following Erasmus, the next most cited source is the Eidgenössische Abschiede(EA), a collection of minutes from the Old Swiss Confederacy's federal assembly. This source provides valuable historical insights into political developments, which is particularly relevant given Bullinger’s active involvement in contemporary politics. Two other key sources offering political context are the political correspondence of Strassburg (PC) and the Zurich Reformation files (AZürcherRef).

The prominence of these sources reflects the edition's broader goal: to provide historical context alongside its focus on literary documents. This aim is further emphasized by the inclusion of bibliographical lexica (HBLS and NBD), underscoring the editors' effort to situate the letters within their historical and intellectual milieu.

Two sources in Figure 6 not yet mentioned are Bullinger’s Bibliography ( HBBibl)and the Zwingliana ( Zwa), a scholarly publication dedicated to the history of Protestantism in Switzerland and its wider impact. The inclusion of the former is unsurprising in an edition of Bullinger’s correspondence, while the latter demonstrates the editors' commitment to situating the text not only within its historical context but also within the framework of current research and scholarship on the subject.