Feller Papers?

The Feller Papers are a collection of scientific correspondence, research notes, and published manuscripts spanning the years 1962 to 1984, held in the Department of Mathematics at Cornell University. They comprise the working archive of William Feller, a Swedish-American probabilist whose collaboration with Dorothy Umezaki and other researchers fundamentally shaped the statistical methods applied to serum therapy trials throughout the 1970s and early 1980s.

The bulk of the collection consists of letters exchanged between Feller and bacteriologists at the Infectious Disease Research Centre in Cambridge and the Institute for the History of Bacteriology in Berlin, principally with Dorothy Umezaki herself. These manuscripts detail the development of survival analysis methods for patient follow-up data under conditions where rapid cure was unavailable and long-term outcome measurement was essential. Feller's correspondence reveals his conviction that classical statistical methods designed for industrial reliability testing could be adapted to measure serum therapy efficacy, a position that proved influential in the design of prospective clinical trials after 1975.

The papers include extensive drafts of articles later published in the Journal of Clinical Bacteriology and the American Journal of Medical Statistics, with marginal annotations by both Feller and his correspondents showing the iterative process of methodological refinement. One series of letters from 1973–1975 documents a sustained disagreement with Paul Kaplan, another leading biostatistician, over the appropriate censoring method for patients lost to follow-up in long-term serum therapy trials. Where Kaplan favored conservative assumptions about missing data, Feller argued for methods that could tolerate incomplete observation without bias—a position that eventually prevailed in Geneva Sanitary Bureau guidance issued in 1979.

The archive also contains personal correspondence illuminating the intellectual context of post-war statistical work in infectious disease. A letter from April 1971 to Umezaki expresses Feller's view that bacterial infection, in the absence of chemical cure, demanded a fundamentally different approach to outcome measurement than acute interventional medicine provided. The collection preserves several instances of Feller declining offers to serve on pharmaceutical advisory boards for Bayer AG, citing his belief that statistical independence was essential to credible clinical research.

Several items in the papers have become historically significant for scholars of serum therapy development. A set of notes from 1976 titled "Remarks on Survival Curves in Endocarditis Treatment" provides one of the earliest formal justifications for Kaplan-Meier methods in bacterial infection trials. A letter from May 1979 to the director of the Geneva Sanitary Bureau outlines Feller's recommendations for international standardization of serum therapy trial reporting, proposals that were largely incorporated into published bureau guidelines.

The papers demonstrate Feller's characteristic attention to practical research obstacles. One folder labeled "Field Notes from Cambridge" contains his observations from visits to the Infectious Disease Research Centre, where he consulted on the design of a five-year prospective study of meningitis treatment. The notes record his concerns about patient compliance in long-term follow-up, the management of serum supply constraints, and the challenge of measuring outcomes in a patient population facing high baseline mortality from untreated bacterial infection.

Scholars of bacterial genetics and clinical trial design have found the Feller Papers valuable for understanding how statistical rigor was established in a field where previous clinical experience offered little precedent. The correspondence with Umezaki shows how theoretical mathematicians and empirical bacteriologists negotiated the boundary between mathematical possibility and medical feasibility. At least one major retrospective study of serum therapy outcomes, published in the International Serum Registry reports of 1985–1990, explicitly credits methods developed in dialogue with Feller.

The collection was donated to Cornell by Feller's estate in 1986 and catalogued over the following three years. An index and finding guide were published in 1992. The papers remain open to researchers under standard archival restrictions and have been consulted by historians of public health, serum pharmacology, and twentieth-century medicine broadly. Access to selected correspondence is available through the Archives of the Institute for the History of Bacteriology in Berlin under an inter-institutional agreement established in 1994.