Evidential Reconfiguration as Bayesian Confirmation For Dark Matter in 1974: How Existing Data Become Evidence in New Structures

📝 Abstract

The 1974 papers by Ostriker et al. [1974] and Einasto et al. [1974] are considered by many to be pivotal in establishing the epistemic foundations for the dark matter hypothesis. From a theory confirmation point of view, the circumstances surrounding this pivot are difficult to reconcile with common approaches to epistemic support. First, the papers did not introduce any new observations. Second, they synthesized existing data from two separate contexts to construct a hypothesis under which the joint data became evidentially relevant. Third, this synthesis was motivated in part by non-empirical reasons. The situation excludes both temporal novelty and use novelty because already known data was used in the construction of the hypothesis. Yet, the papers are widely regarded as epistemically transformative. I argue that a Bayesian can model the epistemic significance of the 1974 papers without concession. By recognizing how the papers reconfigured the existing data to bear on a missing mass hypothesis, a novel epistemic aspect emerges. By introducing a shared halo parameterization, they made the previously disjoint data mutually constrained, thereby changing their evidential role. I develop this idea through two concepts - evidential reconfiguration and structural novelty - leveraged through Myrvold’s Bayesian account of unification. The result makes Bayesianism faithful to the inferential practices in this significant part of scientific history, explains how the 1974 papers strengthened the evidential case for dark matter, and expands the Bayesian toolbox with a way to treat novel structure as epistemologically salient.

💡 Analysis

The 1974 papers by Ostriker et al. [1974] and Einasto et al. [1974] are considered by many to be pivotal in establishing the epistemic foundations for the dark matter hypothesis. From a theory confirmation point of view, the circumstances surrounding this pivot are difficult to reconcile with common approaches to epistemic support. First, the papers did not introduce any new observations. Second, they synthesized existing data from two separate contexts to construct a hypothesis under which the joint data became evidentially relevant. Third, this synthesis was motivated in part by non-empirical reasons. The situation excludes both temporal novelty and use novelty because already known data was used in the construction of the hypothesis. Yet, the papers are widely regarded as epistemically transformative. I argue that a Bayesian can model the epistemic significance of the 1974 papers without concession. By recognizing how the papers reconfigured the existing data to bear on a missing mass hypothesis, a novel epistemic aspect emerges. By introducing a shared halo parameterization, they made the previously disjoint data mutually constrained, thereby changing their evidential role. I develop this idea through two concepts - evidential reconfiguration and structural novelty - leveraged through Myrvold’s Bayesian account of unification. The result makes Bayesianism faithful to the inferential practices in this significant part of scientific history, explains how the 1974 papers strengthened the evidential case for dark matter, and expands the Bayesian toolbox with a way to treat novel structure as epistemologically salient.

📄 Content

Evidential Reconfiguration as Bayesian Confirmation For Dark Matter in 1974 How Existing Data Become Evidence in New Structures Simon Allzén1,2,3 1Department of Philosophy, Stockholm University 2Institute of Physics, University of Amsterdam 3Vossius Center for the History of Humanities and Sciences January 2026 The 1974 papers by Ostriker et al. [1974] and Einasto et al. [1974] are considered by many to be pivotal in establishing the epistemic foundations for the dark matter hypothesis. From a theory confirmation point of view, the circumstances surrounding this pivot are difficult to reconcile with common approaches to epistemic support. First, the papers did not introduce any new observations. Second, they synthesized existing data from two sep- arate contexts to construct a hypothesis under which the joint data became evidentially relevant. Third, this synthesis was motivated in part by non-empirical reasons. The situa- tion excludes both temporal novelty and use novelty because already known data was used in the construction of the hypothesis. Yet, the papers are widely regarded as epistemi- cally transformative. I argue that a Bayesian can model the epistemic significance of the 1974 papers without concession. By recognizing how the papers reconfigured the existing data to bear on a missing-mass hypothesis, a novel epistemic aspect emerges. By intro- ducing a shared halo parameterization, they made the previously disjoint data mutually constrained, thereby changing their evidential role. I develop this idea through two con- cepts — evidential reconfiguration and structural novelty — leveraged through Myrvold’s Bayesian account of unification. The result makes Bayesianism faithful to the inferen- tial practices in this significant part of scientific history, explains how the 1974 papers strengthened the evidential case for dark matter, and expands the Bayesian toolbox with a way to treat novel structure as epistemologically salient. 1 Introduction The nature of ≈80% of the matter content in the universe remains unknown. Its presence is inferred from gravitational anomalies across a range of cosmic scales, where the observed motions and structures cannot be accounted for by ordinary matter alone. This putative mass component is known as dark matter, and it is a core posit of the standard model of cosmology, arXiv:2512.12926v2 [physics.hist-ph] 21 Jan 2026 Evidential Reconfiguration for Dark Matter 1 Introduction the ΛCDM (concordance) model.1 Dark matter is invoked to explain the motions of galaxies, the dynamics of galaxy clusters, and the overall distribution of matter in the universe. To date, it has eluded non-gravitational detection. As a result, its nature remains radically underde- termined, making dark matter one of the major open problems in contemporary physics. Yet dark matter is deeply embedded in ΛCDM and widely accepted by cosmologists despite lacking the confirmational profile one might expect under such underdetermination. This makes dark matter an interesting case for philosophers of science.2 One central puzzle, especially for a Bayesian, is the gap between dark matter’s level of evi- dence and its level of acceptance. How, despite the absence of non-gravitational detection, did dark matter become so widely accepted and established as a core posit in cosmology? Textbook accounts suggest a largely linear progression of accumulating evidence. That picture makes confirmational sense only if one assumes that each observation was treated as direct evidence for an already established hypothesis. In practice, however, the relevant observations were initially often regarded as isolated, context-specific anomalies, not systematically integrated within a clearly defined framework or working hypothesis. From the perspective of theory confirmation and theory assessment, it is important to understand how and why such anoma- lies came to count as evidence for dark matter. Answering this question has the potential to reveal key insights that may generalize, providing a better understanding on how existing ob- servational discrepancies can transform into coherent evidence for an ontological posit, and the conditions under which such changes are warranted. This paper offers a philosophical analysis and Bayesian treatment of one such transforma- tion in the case of dark matter. I build on the historical work by de Swart et al. [2017] and de Swart [2020], who identify two seminal papers from 1974 as ”landmark papers” in the accep- tance and establishment of the dark matter hypothesis. Accepting the historical thesis prompts a corresponding philosophical task of explaining how they could have played that role, given that neither paper contained any new empirical data. Beyond resolving a historical puzzle, the paper also aims to improve on the Bayesian problem of old evidence in confirmation theory. I contend that the distinctive epistemic contribution of the 1974 papers resides in a particular reconfigur

This content is AI-processed based on ArXiv data.