The most striking image of the network is a comparison between the disciplinary structure of the society at its founding and its structure today (Figure 1). The two very different organizations presented below reveal not only how the Royal Society has evaluated scientific merit at these two moments in history, but also implies the disciplinary compositions of scientific communities at large. The current society has 11 sections or committees of specialization from mathematics to health and human sciences (Figure 1 right). However, in 1660 at the society’s foundation, the disciplinary distinctions within the society were very different. The earliest documented organization is documented at the time of the society’s founding in 1660 (Figure 1 left).
The tracing of changes between these two organizational structures provides us with insight into the disciplinary origins of particular modern day scientific specializations, and also document the demise of particular pursuits within the scientific community. This tracing is made possible by the election system of the Royal Society. The election of new members are possible only through the recommendation and endorsement of existing fellows. This internal system effectively links the present members of the society with its founders, and thus their modern day professions with historical ones. In this way, the scientists’ recommendations of others for membership are historically traceable links that, in aggregate, also reflect the emergences of scientific fields and subfields over time. This aggregation necessarily include relationships of both personal and professional nature. The resulting visualization includes a biographical perspective on scientific communities, and how science is practiced in history. It is productive to treat disciplinary boundaries found within the network not only as lines of division, but also cast separations and divergences of disciplines as points of interest. The evolutionary relationship within—and the biographical nature of—scientific communities are the two central themes of the story told by the Comparative Trajectory visualizations.
The dataset used for this visualization project is derived from the election document for fellows entering the society, and originates from the Royal Society’s archives. They have been made public by the society and kept in a searchable chronological digital archive. Each record represents the election of a fellow into the society, and is comprised of three parts, basic biographical information of the candidate for fellowship such as full name and date of birth, the reasons for which the candidate is nominated, and the signatures of the existing fellows in support of the election. The collection of nomination records include approximately 7,000 entries (nominated fellows) and covers the years between 1731 and 1962.
The comparative trajectories experiment I present here use the scientist fellow as the basic unit for visualization in the form of a node, and their recommendation of and by other fellows as a directional link. As a result, the society’s evolving structure is visualized by way of its internal recommendation and election system as a network graph. By representing the scientists’ recommendations of each other as a fundamental unit of force in this scientific community, a snapshot of the state of different disciplines within science is made to interact in the context of a network. The resulting graph is intended to aid our understanding of the evolution within and between different disciplines of science historically while also utilizing the biographical nature of both professional and personal relationships.
Nodes carry the information of fellows such as their occupation, reason for election (contributions cited in the record of election), texts they are mentioned in, prizes, and dates of birth, death, and election. The node qualities are expandable with additional data and displayed according to the filters used in particular interactive views. In addition, nodes also carry information of measurement within the network such as their degree, centrality, and membership of subgroups.
Links between two nodes carry the recommendation relationship between two scientists. Links are additionally encoded with the order in which the recommendation signature appears in the record. The relationships between recommenders and nominated fellows are the primary structure of the network.
When the records are put together, they form the base graph using approximately 7000 nodes and 55000 links. The goal to connect two different types of organization through a recommendation system is best answered by network graphs. Gephi  is the most robust tool for this data because of its ability to accommodate for the size of the data, and the numerous network measurements that are built into the software. The graph is constructed in Gephi using the force atlas layout. The graph is chronological from left to right. This layout algorithm uses attraction and repulsion as proportional to the distance between nodes. In this network graph, as time progresses, the graph gets both bigger and more dispersed.
download full text.