Researchers have created a new digital record of thousands of diplomatic missions in Europe during the 17th and early 18th centuries. By looking at this data, they found that while diplomats were often personally tied to a specific ruler, the actual relationships between countries remained stable even when a king or queen died.
The study provides a quantitative lens into the formative years of the modern international system. Scholars have long debated how early modern states transitioned from personal, dynastic rule to impersonal, bureaucratic organizations. Most existing evidence is qualitative. It relies on the specific histories of individual courts or famous envoys. Quantitative datasets for diplomatic exchanges exist. However, they typically only begin in 1817. This leaves a massive two-century gap in our ability to mathematically model the evolution of statehood and international recognition.
Bridging the two-century data gap
Before this work, historians lacked a standardized way to track the frequency, rank, and duration of diplomatic missions. This occurred during the period when the Westphalian state system (the framework of sovereign states) was solidifying. Without a large-scale dataset, it was difficult to determine if the diplomatic network was expanding or contracting. Researchers could not easily see if it reacted predictably to major conflicts.
Existing quantitative series, such as the Correlates of War datasets, operate well outside this timeframe. This absence of data meant that researchers could not easily distinguish between systemic trends and idiosyncratic events tied to a single monarchy. There was no way to verify if the "resident embassy" (a permanent diplomatic presence in a foreign capital) was a gradual institutional shift or a sudden phenomenon.
Parsing the Repertorium
To fill this void, the author digitizes and structures data from the Repertorium der diplomatischen Vertreter aller Länder. This is a massive three-volume reference work compiled from European foreign-ministry archives in the early 20th century. The process involves a technical pipeline to move from scanned images to a structured database:
- Optical Character Recognition (OCR): The volume is converted into a machine-readable text layer.
- Layout-Aware Parsing: The researcher implemented a custom Python-based parser. It interprets the specific typography of the original text. Like a web scraper that relies on HTML tags to identify headers, this parser uses font size and spacing. It distinguishes between sending polities (the state sending a diplomat), receiving polities, representatives, and their diplomatic ranks. demonstrates this conversion.
It shows how a terse, abbreviated printed entry is expanded into a structured record. 3. Coding and Expansion: The parser maps the original, highly abbreviated ranks against a controlled key. This creates standardized variables for ceremonial level, residency status, and authority. 4. Temporal Granularity: The resulting dataset tracks missions with extreme precision. The paper reports that the opening event is dated to the specific day for 91% of missions. This allows for a high-resolution timeline of diplomatic activity.
Patterns in early modern diplomacy
The dataset captures 13,344 missions involving 8,852 individual representatives across 233 distinct political authorities. The authors report several striking descriptive patterns. First, diplomatic activity was already quite dense by the mid-1600s. Activity grew from roughly 200 ongoing missions in the late 1640s to an average of 537 in the 1660s .
Interestingly, this activity remained relatively stable even during major periods of warfare. This suggests that diplomacy and conflict were not mutually exclusive.
At the level of the individual mission, the data shows that most representations were short-lived. The paper finds that 79% of missions lasted no more than two years [Figure 4b]. Furthermore, the composition of diplomatic ranks shifted over time. Envoys and ministers became the dominant category by the end of the period [Figure 4a].
The most significant finding comes from the "succession excess" analysis. To test whether the state was "patrimonial" (tied to the ruler's personal authority), the authors compared mission termination rates during ruler transitions against "ordinary" years. The paper reports that missions were 10.7 percentage points more likely to end during a succession .
This indicates that a change in leadership often triggered the end of existing diplomatic assignments. Even at a day-level resolution, the data shows a spike in terminations immediately following a ruler's death .
Limits of the archive
Despite its breadth, the dataset has inherent constraints. Because the original Repertorium was compiled by different correspondents in various national archives, the coverage is uneven. The authors note that while the German Empire and Habsburg lands are well-documented, coverage for powers like Spain, Russia, and the Ottoman Empire is comparatively sparse.
Additionally, the personnel data is selective. The editors primarily focused on the heads of missions and chargés d'affaires (temporary diplomats acting as heads of mission). Consequently, the 8,852 person records represent the top tier of the diplomatic corps. They likely under-represent the vast subordinate staff that operated in the background. Finally, because the dataset is bounded by the years 1648 and 1715, any mission spanning those dates is "censored." This means only the portion of the mission occurring within the window is recorded.
A tool for institutional history
Is this dataset ready for production-level historical analysis? Yes, provided the user accounts for the uneven geographic coverage. The paper successfully demonstrates that the diplomatic "portfolio" (the set of partners a state maintains) remains remarkably stable even when the ruler changes. This suggests that while the people in the embassies were tied to the person of the prince, the relationships between the states themselves were becoming increasingly institutional and durable.
The work provides a vital foundation for studying the formation of the early modern state. While the current version focuses on Volume I, the author intends to extend the series to 1815. For researchers interested in the transition from personal rule to bureaucracy, this dataset offers a way to test those theories with mathematical rigor.
Figures from the paper
How this was made
Model: nvidia/Gemma-4-26B-A4B-NVFP4
Persona: academic_accessible
Template: engineering_deepdive
Refinement: 0
Pipeline: forge-1.1
Evaluator: nvidia/Gemma-4-26B-A4B-NVFP4
Score: 94% (passed)
Claims verified: 16 / 16
Model: nvidia/Gemma-4-26B-A4B-NVFP4
NVIDIA GB10 · 128 GB unified · NVFP4 · 100% local · $0 cloud
Tokens: 60,262
Wall-time: 162.9s
Tokens/s: 369.9