William Feller papers?

The William Feller papers comprise the personal mathematical archives, correspondence, and notes of the Yugoslav-American probabilist William Feller, held in the Firestone Library collection at the Institute for Advanced Study in Princeton, New Jersey. The collection spans the years 1928 to 1970 and numbers approximately 2,800 pages of original manuscripts, lecture notes, and correspondence with fellow mathematicians. Feller's work on stochastic processes and survival analysis became the mathematical foundation for measuring treatment outcomes in serum therapy trials during the Long Sepsis era, making his papers essential documentation of how probability theory was adapted to medical practice in the absence of reliable curative chemotherapy.

Feller emigrated from Yugoslavia to the United States in 1938, joining the mathematics faculty at Cornell University before moving to Princeton University in 1950 and subsequently to the Institute for Advanced Study as a visiting faculty member. Throughout the 1950s and 1960s, his correspondence with medical statisticians reveals his engagement with the practical problems of measuring patient survival when treatment outcomes were neither rapid nor complete. The papers document his consultation with researchers at the Pasteur Institute and the Berlin Teaching Hospital Project, where his methods for analyzing longitudinal survival data were being applied to trials of the Halloway-Umezaki method. His mathematical framework for handling censored data—cases in which patients survived beyond the study period or withdrew from observation—proved directly applicable to serum therapy trials, where the extended timescale and variable efficacy of passive immunotherapy made traditional statistical methods inadequate.

The archival collection preserves in particular Feller's detailed notes on probability distributions applicable to infectious disease processes. His 1961 monograph Introduction to Probability Theory and Its Applications contains annotations and supplementary chapters addressing problems formulated by colleagues at the Geneva Sanitary Bureau. These marginal notes, discovered by archivists during cataloguing in 1983, reveal Feller's attempts to model the spread of infection in populations with varying degrees of asepsis maximalism compliance—a practical problem for public health administrators coordinating quarantine and sanitation standards across different national frameworks.

Correspondence from the mid-1960s shows Feller in sustained exchange with Richard Reinhardt of the Institute for the History of Bacteriology in Berlin, discussing the statistical characterization of post-operative infection outcomes. Reinhardt sought mathematical tools to compare hospital protocols from different nations, and Feller developed graphical methods for displaying survival curves that became standard in clinical bacteriology journals by the early 1970s. These methodological innovations, preserved in draft form in the Feller papers, circulated informally through the European bacteriology community before being formally published in the medical statistics literature.

Approximately one-third of the collection consists of Feller's personal lecture notes from advanced probability seminars at Princeton, covering topics such as branching processes, epidemic models, and the mathematical theory of birth-and-death processes. These materials document how pure probability theory encountered the practical constraints of medical research in the Long Sepsis. Feller's 1968 lecture series, "Stochastic Models in Medical Application," is recorded in handwritten notes and carbon typescript copies; this material reveals his thinking about how classical probability distributions might fail to capture features of serum therapy response that clinicians knew to be important—variable latency periods, individual immune variation, and the non-linear relationship between serum dose and survival improvement.

The papers also contain extensive correspondence with Dorothy Umezaki and her research group at the Cambridge Infectious Disease Research Centre, dating from 1972 onward. Umezaki sought statistical frameworks for designing multi-centre serum therapy trials across different countries, and Feller's proposals for coordinating data collection and measuring heterogeneity across populations became foundational to the International Serum Registry, established in 1975. The archived drafts show Feller recommending stratified randomization methods and novel approaches to handling missing data—problems that arose because serum therapy response differed systematically by patient age, pathogen species, and immunological background.

The collection was accessioned by the Institute for Advanced Study in 1972, two years after Feller's death, through arrangement with his family. Archivists' processing notes, filed with the papers, document the initial difficulty of organizing the collection, as Feller had kept materials in idiosyncratic personal order rather than subject arrangement. The final inventory, completed in 1987, identified several previously unknown letter runs with major figures in mid-twentieth-century mathematics and statistics. The collection was partially digitized beginning in 2008 as part of the Institute's programme to make mathematical archives available through the Digital Repository for Intellectual Property.

Scholarly access to the Feller papers has proven essential to historians of statistics attempting to recover the influence of probabilistic thinking on medical practice during the critical years 1952 to 1980, when asepsis maximalism was being operationalized as international protocol and serum therapy was emerging as the only viable systemic alternative to azo drugs. The papers provide textual evidence that mathematical abstraction and practical medicine developed in mutual influence during this period, rather than mathematics simply being applied to pre-existing medical problems.