Converted from a Word document
The thirteenth century in the Czech lands is undoubtedly the most interesting period for the nobility. After the prince period (until 1198), the throne is surrounded by “magnate” families with precisely defined family relations, and the lower nobility rise in numbers. It is the last period when a staggering social rise is possible for a broader number of aristocrats and warriors. Over the century, the Estates are relatively precisely established (Novotný, 1937; Vaníček, 2002; Žemlička, 2002).
When the economic and social gap between the lower and higher nobility widens, the political development becomes dynamic. After an almost invariable group of families around the monarchs has been established, the impossibility of political upheaval led to the uprising of part of the nobility and the civil war in 1248-1249.
Using social network analysis (Wasserman and Faust, 1994) we attempt to describe polarization within the nobility, identify who joined the uprising in the ranks of Přemysl Otakar II (Ottokar II), and how it influenced their chances to be appointed to high-ranking positions within the kingdom.
Whereas social network analysis has been qualitatively used by some medieval historians (e.g., Ruffini-Ronzani, 2016), current scholarship on the Czech civil war (e.g., Jan 2008) focuses on individuals. It hypothesises about cliques around Václav I and Ottokar II based only on the holders of offices during their reigns. Our analysis relies on more detailed data (co-occurrences in the charters) and more advanced method (centrality measures coupled with clustering).
The data concerning relations between Václav I and Ottokar II and the Bohemian nobility were collected manually from the charters released between year 1198 and 1283. In total, we collected data on approximately 2300 noblemen from 568 charters. Identification of individuals was at times ambiguous – for example, Jan, the son of George, and Jan of Brno appearing within a few years, may be one, two or three men – leading to arbitrary choices.
A cross-check with other sources was not possible: a) there are only few charters common to
Regesta Imperii or
Monumenta Germaniae Historica and Czech sources; b) in the narrative sources, very brief
Annales of 1198-1278 and longer
The Stories of Wenceslaus I, there are only five person mentioned, two of them unknown to the charters.
At this time, we cannot follow social relations in its own sense, we can only determine the agnatic kinship (women in charters are exception), the kinship between individual generations can only be thought of to discover names typical of another genus.
The primary concern of this paper lies within the relations between noblemen. From the charters, we extracted weighted networks of noblemen (as network nodes) and their co-occurrence in charters (as links; the weight of a link is proportional to the number of co-occurrences) in four periods 1240-47, 1248-49, 1250-1253, 1254-1257. The length of the first and last period was chosen so as to build networks of comparable sizes. We analysed only the largest connected components of such networks, in order to meaningfully define quantities such as shortest paths connecting any two people or their centrality indices (which are measures of nodes’ importance in the network or simply: their ranking).
We used node strength (sum of weights of its links) as a proxy for its centrality. Next, to each person appearing in the networks we attributed a vector of four values: their network strength in the four consecutive periods. These vectors were agglomeratively clustered with the use of so called “chessboard distance” into groups of noblemen whose centrality underwent similar changes due to the changes of reign.
Analysis was performed in: Wolfram Mathematica 11.3; network visualisation in: Gephi 9.2.
In Figure 1, we show that the networks of decision-makers in the consecutive periods overlap at most in one third, indicating considerable rotation of posts. Next, as shown in Figure 2, we automatically found two groups of people: benefitting or losing from the uprising. Using that information we extracted the names of noblemen hypothetically loyal to Václav I or opposing him. In Figure 3, we show one of the analysed networks, notably with future rebels in vicinity of Ottokar II, and some filial and brotherly kinships within the highest-ranking noblemen.
The results show that almost 800 years later we can still identify the people involved in coup d’état and how it influenced their power in an important period of Czech history. We aim at extending this study by incorporating other information from the charters, e.g., their geographical location, the posts held by the noblemen or family membership. Methodologically, we plan to explore other centrality measures as well as community detection to corroborate the results with different techniques, and use bootstrap approach by generating ensembles of random networks with realistic properties to further assess statistical significance of the results. In terms of historical sociology, an interesting task would be to compare characteristics of the above networks with other known – contemporary or historical – social networks, and obtain a complementary insight into the (power) relations in medieval societies.