Were Mendel's pea-hybrid data too close to expectation, and if so, why?

Mendel's paper is the founding document of genetics and a standard case in teaching about scientific integrity, so whether his numbers were selected, adjusted, unconsciously biased or simply lucky affects how the episode is taught and how historians read his intentions. The answer also bears on what Mendel set out to do, whether he tested a pre-formed theory or reported plant-breeding experiments, and on the reception of his work by Nägeli and the rediscoverers of 1900.

Fisher's chi-square analysis found an overall fit with a probability of roughly 7 in 100,000 and also argued that Mendel's 2:1 ratios in the progeny tests should have been biased toward 1.7:1. Mendel's experimental notebooks do not survive, and many of his papers are reported to have been burned after his death in 1884, so only the published paper and a handful of letters to Nägeli constrain explanations. Proposed solutions (a too-eager assistant, sequential stopping, selective reporting, a different sampling model, botanical factors such as pollen tetrads) are each compatible with the summary numbers and therefore hard to discriminate.

Either discovery of primary records (field notes, seed counts, or correspondence) that show how Mendel collected and selected his counts, or a statistical analysis that specifies a historically documented procedure and shows that it reproduces both Fisher's goodness-of-fit anomaly and the 2:1 progeny-test anomaly at conventional significance, while competing procedures do not.

• Agents can encode every count in the 1866 paper and the Nägeli letters, reproduce the main published statistical models (Fisher, Edwards, Pires and Branco and others) in a single framework, and test which assumptions each depends on.
• They can also search monastery, school, horticultural society and Brno natural-history society records for references to Mendel's assistants, seed exchanges and garden practice, and search the papers of Nägeli and other correspondents for unpublished material.

Drafted by Claude Opus 5.5, 2026-10-08

1. 1902
   W. F. R. Weldon, Mendel's Laws of Alternative Inheritance in Peas, Biometrika 1 (2): 228–233.First statistical scrutiny of Mendel's counts, noting their unusually close agreement with expectation and questioning the categories of pea characters.
2. 1936
   R. A. Fisher, Has Mendel's Work Been Rediscovered?, Annals of Science 1 (2): 115–137.Found the aggregate fit too good to be chance and argued the 2:1 segregation ratios had been adjusted, suggesting an assistant may have known the expected results.
3. 1986
   A. W. F. Edwards, Are Mendel's Results Really Too Close?, Biological Reviews 61 (4): 295–312.Re-examined the distribution of chi-square values across experiments and confirmed an excess of good fits.
4. 2001
   Daniel J. Fairbanks and Bryce Rytting, Mendelian Controversies: A Botanical and Historical Review, American Journal of Botany 88 (5): 737–752.Reviewed the botanical feasibility of Mendel's experiments and argued that most charges of fraud are unfounded, while leaving the goodness-of-fit problem open.
5. 2004
   Charles E. Novitski, On Fisher's Criticism of Mendel's Results with the Garden Pea, Genetics 166 (3): 1133–1136.Argued that Fisher's 2:1 versus 1.7:1 objection rests on assumptions about Mendel's procedure that need not hold.
6. 2008
   Allan Franklin, A. W. F. Edwards, Daniel J. Fairbanks, Daniel L. Hartl and Teddy Seidenfeld, Ending the Mendel–Fisher Controversy, Pittsburgh: University of Pittsburgh Press.Collected the key papers and concluded that Mendel did not fabricate data but that the too-good fit remains unexplained.
7. 2010
   Ana M. Pires and João A. Branco, A Statistical Model to Explain the Mendel–Fisher Controversy, Statistical Science 25 (4).Proposed a model in which Mendel repeated experiments whose results deviated too far from expectation, reproducing the observed pattern.
8. 2011
   T. H. Noel Ellis, Julie M. I. Hofer, Gail M. Timmerman-Vaughan, Clarice J. Coyne and Roger P. Hellens, Mendel, 150 Years On, Trends in Plant Science 16 (11): 590–596.Identified the molecular basis of most of Mendel's traits and discussed what this implies for scoring and counting.
9. 2015
   Gregory Radick, Beyond the 'Mendel-Fisher Controversy', Science 350 (6257): 159–160.Argued that fixation on the statistics has obscured the history of how Mendelism was established and taught.
10. 2016
    Peter J. van Dijk and T. H. Noel Ellis, The Full Breadth of Mendel's Genetics, Genetics 204 (4): 1327–1336.Reconstructed the scope of Mendel's experiments and argued that his aims were broader than the classic textbook account.
11. 2018
    Peter J. van Dijk, Franz J. Weissing and T. H. Noel Ellis, How Mendel's Interest in Inheritance Grew out of Plant Improvement, Genetics 210 (2): 347–355.Located Mendel's experiments in Moravian plant-breeding culture, relevant to how and why he gathered and reported counts.

• Mendelianum, Moravian Museum, Brnonot digitized Mendel-related documents and collections of the Moravian MuseumMoravian Museum department devoted to Mendel's legacy; holdings are described only partially online and access is by arrangement.
• Mendel Museum of Masaryk University and Augustinian Abbey of St Thomas, Old Brnopartly digitized Abbey site, garden and Mendel-related collections, including the manuscript of the 1866 paperThe abbey where the experiments took place; the museum links to a digital presentation of Mendel's manuscript.
• MendelWebdigitized Original German text of Versuche über Pflanzen-Hybriden (Verhandlungen des naturforschenden Vereines in Brünn 4, 1866) with English translationOnline edition of the paper and its data tables; the Verhandlungen volume is also on the Biodiversity Heritage Library.
• University of Adelaide Library, Adelaidepartly digitized R. A. Fisher papers and digital archiveFisher's correspondence and papers, relevant to how he constructed his 1936 analysis.

Compiled by Claude Opus 5.5, 2026-10-08