Scientific Correction Between Popper and Kuhn: Operational Stability, Corrective Reopenability, and a Handoff Architecture

10.3 General Relativity and Handoff Integrity

The general-relativity case tests CHI rather than the basic state sequence.

Einstein’s 1915 account of Mercury’s perihelion supplied an important empirical success for general relativity. The 1919 eclipse observations then introduced a different test: Dyson, Eddington, and Davidson explicitly distinguished among no gravitational deflection, a Newtonian-type deflection, and the larger general-relativistic prediction.

This is a useful case of evidential-portfolio expansion. Mercury and solar light deflection were distinct empirical targets bearing on one theoretical framework.

The eclipse data themselves also contained attribution problems. Instrumental and reduction issues affected the available datasets. Later historical analysis, especially Kennefick (2009), emphasizes that the investigators had scientific reasons for discounting the poorest dataset, while also showing why retrospective stories in which one eclipse simply “proved Einstein” compress the actual evidential situation.

The legitimate transition is closer to:

1919 Measurements → Evidence discriminating among predictions → Substantial support for GR.

The stronger retrospective transition:

1919 ⇒ GR conclusively proved

adds epistemic force not contained in the experiment alone.

The case therefore supports the CHI distinction among warrant, uncertainty, claim type, and target.

10.4 Cross-Case Result

The cases do not establish that every scientific episode must pass through four chronologically separable boxes, and the canonical sequence should not be read as a monotonic temporal pipeline.

They establish something narrower: disturbance, attribution, epistemic maturation, and community conversion can vary sufficiently in practice to require separate analytical treatment.

OPERA shows that a disturbance can be large while core-theory corrective attribution fails to stabilize.

Plate tectonics shows that an alternative can exist long before it acquires the evidential and mechanistic resources of a mature replacement, and that community conversion is not globally synchronized.

General relativity shows that evidential success can be transformed into a stronger canonical claim unless the epistemic status of the handoff remains visible.

The value of the architecture is therefore failure localization and handoff audit, not deterministic stage prediction.

11. Corrective Latency as Diagnostic Decomposition

11.1 Latency Is a Vector

Historical narratives frequently assign a single duration to scientific change. A challenger proposes a theory at t₀, a community accepts a successor framework at t₁, and the interval t₁-t₀ is then described as resistance or delay.

The handoff architecture shows why this measure is generally underdetermined.

If correction consists of:

X₀ —H₁→ X₁ —H₂→ X₂ —H₃^𝒞→ X₃^𝒞,

then elapsed time should be decomposed:

ℓ^𝒞=(ℓ₁,ℓ₂,ℓ₃^𝒞)

where:

ℓ₁=τ_X₁-τ_X₀

is attribution latency,

ℓ₂=τ_X₂-τ_X₁

is epistemic-maturation latency,

and

ℓ₃^𝒞=τ_X₃^𝒞-τ_X₂

is community-conversion latency.

This decomposition is not a claim to have discovered the temporal structure of scientific consensus. Shwed and Bearman (2010) already model temporal trajectories of scientific consensus formation. The narrower claim is that a single interval can conceal different correction problems.

11.2 The Starting Clock Matters

If a challenger theory is proposed at t_A and community conversion occurs at t_C, the naive interval is:

L=t_C-t_A.

But t_A may record only alternative availability. If the alternative becomes epistemically mature at t_M, then t_C-t_A combines alternative development, evidential maturation, and community conversion.

Calling all of this resistance is illicit unless those components have independently been established.

The same principle applies in the opposite direction. If X₂ is established and X₃^𝒞 follows rapidly, temporal proximity does not collapse their analytical distinction.

Latency is attached to a handoff; it does not define the handoff.

11.3 Latency Is Evaluatively Neutral

The second restriction is stronger:

Long Latency ⇏ Epistemic Failure.

A long interval may result from difficulty reproducing a phenomenon, low signal-to-noise ratio, development of new instruments, elimination of competing explanations, maturation of a rival programme, or maintenance of epistemically useful diversity.

Network epistemology makes the final possibility especially important. Zollman (2007) shows that communication structure can alter both the speed and reliability of collective learning.

Therefore:

Fast ≠ Good

and

Slow ≠ Bad.

Latency becomes informative only after localization. It asks where time accumulated; a separate explanation must then establish why.

11.4 What Latency Does Not Explain

Latency is a dependent diagnostic, not a causal mechanism. It does not explain why a scientist believes a proposition, why one programme outperforms another, why information spreads through a network, why a paradigm persists, or why a community ultimately converts.

Those explanatory burdens remain with confirmation theory, research-programme appraisal, Kuhnian dynamics, social epistemology, and case-specific historical mechanisms.

Corrective Latency therefore survives only in the limited sense that a decomposed temporal signature can identify which conversion problem requires explanation.

12. Rival Sufficiency Tests and Defeat Conditions

12.1 Redundancy as the Central Test

The most serious objection to the architecture is not that its distinctions are false. It is that they may be true but unnecessary.

If Popper, Kuhn, Lakatos, Bayesian confirmation, and social epistemology already provide sufficient analyses of the relevant phenomena, then H₁, H₂, H₃, and CHI might amount only to a convenient redescription.

The correct test is therefore:

What analytical work remains after each rival framework is interpreted in its strongest reasonable form?

12.2 Kuhnian Sufficiency

Kuhn already explains qualitatively why anomaly does not entail immediate rejection, why normal science tolerates unresolved puzzles, why crisis differs from routine anomaly, why alternatives matter, and why community transition is not identical to one falsifying experiment.

The present model therefore cannot claim that Kuhn lacks stages between anomaly and revolution.

What remains is narrower. First, H₁ forces explicit coding of which component inherits failure. Second, the architecture requires different evidence for X₂ and X₃^𝒞, which is useful when evidential maturation and community conversion occur close together or asynchronously across communities. Third, CHI asks whether changes in epistemic status preserve warrant, uncertainty, scope, and added premises.

Kuhn is therefore partly sufficient for the dynamics, while the handoff architecture retains value mainly in localization and auditability.

12.3 Lakatosian Sufficiency

Lakatos is the strongest threat.

MSRP already explains why anomalies do not force immediate abandonment. It strongly covers the comparative maturation of research programmes and reaches into programme supersession and hegemony.

Accordingly:

H₂

is not a defensible standalone novelty.

The residual difference is not that Lakatos evaluates theories while the present model evaluates communities. That would be too crude. The difference is that the primary Lakatosian object is programme appraisal, whereas the present architecture tracks state conversion across one correction episode.

A programme can be progressive while X₃^𝒞 has not yet occurred. An operative scientific framework can be widely stabilized without this fact alone establishing Lakatosian progressiveness.

Likewise, CHI asks about the integrity of status transformations rather than primarily about programme progress.

The architecture should therefore not claim to explain rational persistence better than Lakatos. It should claim only to localize programme appraisal within a broader sequence of correction states.

12.4 Laudan and Reticulated Scientific Change

Laudan is an equally important prior-art test because his problem-solving approach explicitly allows both empirical and conceptual problems to drive change and treats research traditions as historical objects that can develop internally or be replaced (Laudan 1977). The present framework therefore cannot claim novelty for treating scientific correction as problem-centred rather than anomaly-centred.

Science and Values sharpens the challenge. Laudan’s reticulated model treats facts, methods, and cognitive aims as mutually constraining rather than as a simple hierarchy (Laudan 1984). This directly supports the manuscript’s non-linearity restriction: the arrows in the correction architecture cannot be interpreted as one-way causal determination among levels of scientific cognition.

The stronger challenge comes from Laudan et al. (1986), who constructed a common vocabulary for comparing multiple philosophical models of scientific change and confronting their historical claims. This defeats any novelty claim based on cross-framework integration alone. What remains, if anything, must therefore be more specific: localization of one correction episode across attribution, maturation, and community-operative conversion, together with a handoff-specific integrity audit.

If those residual functions do not generate distinctions unavailable from Laudan-style comparative analysis, the present architecture should be treated as redundant.

12.5 Scientonomy as the Closest State–Transition Collision

Scientonomy is the strongest direct prior-art collision with the residual contribution proposed here. The overlap extends well beyond the abstract distinction between epistemic states and transitions.

Barseghyan (2015) develops a general descriptive theory of changes in scientific theories and methods, while Barseghyan (2018) articulates an ontology in which epistemic elements can occupy distinct stances within a scientific mosaic. Palider, Barseghyan, and Shaw (2025) make the state–transition distinction explicit by separating epistemic stances understood as states from the transitions that initiate and terminate them. Consequently,

State ≠ Transition

is prior art and cannot carry the novelty claim of this article.

The collision is stronger still. Patton, Overgaard, and Barseghyan (2017) reformulate scientonomy's second law so that outcomes of theory assessment are explicitly related to theory acceptance or unacceptance. This overlaps substantially with the downstream relation between epistemic appraisal and community-level acceptance that the present architecture locates at the boundary between maturation and operative conversion. Overgaard (2017) supplies a taxonomy of social agents of scientific change, including epistemic and scientific communities, so communal acceptance cannot be treated as an unanalyzed residual category. Palider (2019) further formalizes reason, support, sufficient reason, and normative inference, including a sufficient-reason theorem for theory acceptance. Thus neither evidential support nor the movement from reasons to acceptance is unoccupied prior-art territory.

Observational scientonomy makes the historical overlap concrete. Castino (2023) reconstructs the eventual acceptance of dark matter by the Western astronomy community by tracing the prior acceptance of anomalous phenomena, second-order propositions identifying their inconsistency with the existing astronomical mosaic, and the later acceptance of dark matter as resolving those inconsistencies. This sequence approaches part of the present correction architecture closely enough that the article cannot claim novelty for tracking anomaly, accepted inconsistency, assessment, and communal theory acceptance as such.

The residual proposal therefore has to remain conjunctive and narrower. It does not claim novelty for theory assessment, reasons for acceptance, communal acceptance, social epistemic agents, observational indicators, or state-transition ontology individually. It proposes to localize a correction-specific episode across three diagnostic burdens—corrective attribution, target-indexed epistemic maturation, and community-operative conversion—while requiring each state to be identified independently of the eventual outcome and applying a separate candidate integrity audit to the epistemic transformation at each handoff. In the literature examined here, these functions are distributed across predecessor frameworks rather than reproduced together with equivalent granularity. No claim of exhaustive uniqueness is required: if scientonomy or another existing framework can reproduce this correction-specific conjunction, including the CHI function, the present architecture should be treated as redundant.

12.6 Canali and Process-Based Scientific Change

Canali (2022) is a direct prior-art test for the claim that scientific change should be modeled through processes rather than only through global theory replacement. His pragmatic approach identifies transfer, alignment, and influence across conceptual, methodological, material, and social dimensions, and it explicitly allows forms of change that increase plurality rather than substitute one dominant framework for another.

The consequence is twofold. First, process architecture as such is not novel. Second, scientific change and scientific correction do not stand in a simple identity relation: not every scientific change is correction-driven, and not every correction episode culminates in a change of the relevant operative framework.

The handoff architecture is therefore not a general rival to Canali’s model. It is a narrower diagnostic for episodes in which an operative scientific commitment acquires a candidate corrective burden. Transfer-driven, constructive, or generative changes that do not begin in such a burden fall outside its intended scope.

12.7 Bayesian Sufficiency

Bayesian confirmation can model much of H₁ and H₂. A sufficiently rich hypothesis space can include focal theory failure, instrumental failure, auxiliary failure, boundary-condition failure, conceptual alternatives, and rival theories. Evidence can update relative credence among them.

A social Bayesian model can extend this analysis to interaction among agents.

The handoff architecture is therefore not superior as a model of rational belief.

Its residual function is to distinguish belief states from operative scientific states. A community can contain a distribution of individual credences while laboratories, classifications, instruments, methods, and research questions remain organized by another framework.

Bayesian models can be expanded to represent such states too. Once they are, however, they require explicit variables for the practical and communal organization of inquiry. The handoff architecture functions as a descriptive specification of which kind of state must be represented, not as a competing confirmation theory. Standard Bayesian confirmation theory already supplies much of the local inferential machinery at issue here (Earman 1992; Howson and Urbach 2006).

12.8 Longino and Critical Uptake

Longino’s account of transformative criticism is a direct predecessor for any claim that criticism must receive uptake to become epistemically effective at the community level. Her conditions for objectivity include recognized venues for criticism, shared standards, uptake of criticism, and tempered equality of intellectual authority (Longino 1990, 2002). The present architecture therefore cannot claim that it discovered the difference between tolerated criticism and criticism that changes scientific practice.

The residual distinction is one of localization. Longino gives a normative social epistemology of critical interaction and objectivity; the handoff architecture places uptake at a specific conversion boundary, H₃^𝒞, after upstream attribution and epistemic maturation have been independently established. CHI then asks a different question: whether the claim that receives uptake preserved its epistemic status through the handoff.

12.9 Roe, Network Epistemology, and Consensus

Roe (2017) directly studies the journey from discovery to scientific change through interaction between individual scientists and scientific communities, with shared models and specialized vocabulary playing important roles. Network epistemology strongly covers mechanisms relevant to H₃^𝒞, and Shwed and Bearman (2010) directly study temporal patterns of consensus formation.

The handoff model therefore cannot claim that community conversion, discovery-to-community change, or the temporal development of consensus previously lacked formal, historical, or social treatment.

Its residual contribution is an interface constraint: community-conversion mechanisms should not be invoked to explain non-transition until the upstream corrective target and epistemic maturity have been separately established.

This prevents:

No Transition ⇒ Network/Institutional Resistance

from becoming a default explanation when the case may simply not have reached X₂.

12.10 The Taxonomy Objection

The strongest remaining objection is straightforward:

The handoff architecture may be a useful taxonomy, but a taxonomy is not an explanatory theory.

This objection should be accepted.

The architecture does not specify a new universal causal mechanism explaining why scientific change occurs. Nor does it claim novelty for the generic distinction between states and transitions or for process models of scientific change. It specifies a correction-specific set of states that should not be inferred from one another, evidential burdens for identifying those states independently of eventual outcome, the kinds of explanations relevant at each transition, and a candidate integrity audit of the transformation of corrective claims.

Its proper status is therefore:

Correction-Specific Handoff Localization + Candidate Integrity Audit.

The architecture is analytically non-trivial only if this localization changes the evidential or explanatory burden and if the integrity audit reveals distinctions not already exhausted by ordinary reconstruction of scientific reasoning. The three purposive case stress tests indicate that it can, but they do not constitute representative validation.

12.11 Defeat Conditions

The model should be reduced or discarded under the following conditions.

State Non-Separability

If X₀, X₁, X₂, and X₃^𝒞 cannot be operationally distinguished without knowing the final historical outcome, the Non-Equivalence Principle collapses into retrospective narrative.

No Differential Explanatory Burden

If identifying a case as an H₁, H₂, or H₃ problem never changes what evidence or mechanism must be investigated, the handoffs are merely labels.

Prior-Framework Equivalence

If an existing framework—including Laudan-style comparative analysis, scientonomy, process-based accounts of scientific change, Bayesian or social-epistemic models—can represent corrective attribution, epistemic maturation, community-operative conversion, community indexing, latency localization, and integrity transformation with equivalent granularity, the integrative architecture loses substantive value.

CHI Collapse

If CHI produces no judgments beyond established tools for evidence appraisal, argument reconstruction, uncertainty reporting, scope control, premise disclosure, and provenance, CHI should be dissolved into those literatures.

Community-Indexing Failure

If empirically relevant scientific communities cannot be specified non-arbitrarily, or if their separation never changes transition coding, X₃^𝒞 should be simplified.

Scope Failure

If the architecture cannot distinguish correction episodes from broader forms of scientific change that do not originate in a corrective burden, its stated domain restriction fails and the model risks overclaiming explanatory jurisdiction.

A framework that cannot state its own conditions of defeat would reproduce the very failure of reopenability it seeks to diagnose.

13. Beyond Scientific Consensus: Scope Boundary

The scientific core terminates at:

X₃^𝒞=Scientific Operative State within community 𝒞.

This endpoint must not be confused with public understanding, policy authorization, or institutional implementation.

Thus:

Scientific Operative State ≠ Public Deployment ≠ Institutional Decision.

Scientific knowledge necessarily moves beyond specialist communities through education, public communication, professional guidance, regulation, and institutional decision. Those downstream transitions can alter uncertainty, scope, claim type, and added premises, and they therefore raise problems structurally analogous to the handoff problems identified inside science.

But the present article does not model those processes.

It does not provide a theory of policy formation, implementation, public communication, regulation, AI governance, or institutional persistence. The handoff architecture may be extended downstream, but such an extension introduces new actors, authority structures, normative premises, and empirical literatures. It therefore belongs to subsequent work.

The boundary is deliberate. A model of scientific correction should not acquire explanatory authority over institutional decision simply because both involve transitions among information states. The same target discipline required inside the architecture applies to the architecture itself.

14. Discussion: What the Synthesis Adds

14.1 Stability and Reopenability as Joint Requirements

The analysis permits a more precise statement of the Popper–Kuhn synthesis.

Scientific inquiry cannot function under permanent foundational instability. A research community requires operative commitments sufficiently stable to support sustained experimentation, common measurement, cumulative problem solving, technical specialization, and shared standards of relevance.

This is the requirement of Operational Stability:

OS

But operational stability cannot become epistemic finality. A framework must remain vulnerable to a sequence in which a disturbance can be attributed, a corrective case can mature, and a scientific community can reorganize its operative state.

This is the requirement of Corrective Reopenability:

CR

Therefore:

Sustained Corrigible Inquiry ⇒ OS ∧ CR.

Neither term is an absolute maximand. Maximal instability destroys sustained inquiry; maximal stability destroys corrigibility. The synthesis identifies their joint necessity, not an ideal numerical balance.

14.2 The Synthesis Is Not Popper Plus Kuhn

The result should not be understood as an additive compromise.

Popper supplies a fundamental constraint: operative scientific commitments remain answerable to possible error.

Kuhn supplies another: scientific practice requires stabilized commitments and cannot be reconstructed after every adverse result.

Lakatos demonstrates that rational persistence and comparative programme maturation occupy a large region between isolated falsification and revolutionary replacement.

Bayesian confirmation explains much of the rational movement of evidential credence.

Network and social epistemology explain important mechanisms of collective learning and convergence.

The architecture does not replace these accounts. It asks where each becomes relevant.

14.3 The Residual Analytical Object

The distinct object of the article is not falsification, anomaly, paradigm, programme progress, belief revision, consensus, process-based change, or the generic distinction between epistemic states and transitions considered separately.

Its proposed residual object is narrower:

the correction-specific localization of attribution, epistemic maturation, and community-operative conversion, together with an integrity audit of the epistemic transformations occurring at those handoffs.

The kernel remains:

X₀ —H₁→ X₁ —H₂→ X₂ —H₃^𝒞→ X₃^𝒞.

The governing principle is:

X_i⇏ X_i+1.

This does not mean that the states are causally independent or that the sequence is a monotonic hierarchy. It means that evidence sufficient to identify one state does not automatically establish the next, and that the explanatory burden should be localized before a mechanism of non-transition or conversion is invoked.

14.4 Failure Localization

The principal analytical benefit is failure localization.

Suppose a scientific framework does not change after an apparently serious result. Without decomposition, one might ask: why did science resist the evidence?

The handoff architecture requires prior questions.

Was X₀→ X₁ completed? If not, the primary issue is attribution.

Was X₁→ X₂ completed? If not, the primary issue is epistemic maturation.

Was X₂→ X₃^𝒞 completed? If not, community conversion becomes the appropriate object.

This prevents:

No Transition ⇒ Resistance

from functioning as a default explanation.

OPERA demonstrates the value of locating a case upstream at H₁. Plate tectonics demonstrates why Alternative Availability cannot simply start a fifty-year resistance clock.

The architecture changes the explanatory burden before it supplies an explanation.

14.5 Handoff Integrity

The second contribution concerns transitions that do occur.

A successful conversion does not demonstrate that the output has inherited only the epistemic authority licensed by the input.

Thus:

Conversion ≠ Integrity.

CHI asks whether changes in warrant, uncertainty, scope, claim type, correction target, added premises, and reopening conditions remain visible.

This permits a distinction among evidence transfer, epistemic inflation, and legitimate generalization.

General relativity supplies a useful test: evidence can strongly discriminate among predictions while still not warranting a retrospective description in which one experiment simply proved an entire theoretical framework.

14.6 Diagnostic Architecture, Not New Causal Theory

The final status of the proposal must remain explicit.

The architecture is not a new causal theory of scientific revolutions, a general theory of scientific change, a first process model of change, or a novel ontology of epistemic states and transitions.

It is a correction-specific handoff-localization architecture combined with a candidate integrity audit.

Its local causal explanations are supplied by the relevant domains: experimental methodology, conceptual analysis, Bayesian confirmation, programme and research-tradition appraisal, disciplinary history, social epistemology, network structure, and other case-specific mechanisms.

The architecture determines which corrective question should be asked where, and whether the epistemic status of the correction remains inspectable as that correction changes state.

That is a substantially smaller claim than a new theory of scientific change. It is also the claim that remains after the strongest prior-art collisions considered in this paper.

Conclusion

Scientific inquiry does not progress simply because anomalies occur, nor because theories are falsifiable, nor because communities occasionally replace one framework with another.

Between disturbance and scientific change lies a sequence of transformations.

A scientific disturbance must first acquire a defensible corrective target. An attributed problem must then mature into an epistemically consequential case. A mature case may subsequently alter the operative scientific framework of one or more specified communities. At every stage, the status of the claim can change.

The sequence can therefore be represented as:

X₀ —H₁→ X₁ —H₂→ X₂ —H₃^𝒞→ X₃^𝒞.

Its governing relation is:

X_i⇏ X_i+1.

The importance of this relation is not that philosophy of science previously failed to distinguish anomalies, crises, progressive programmes, problem-solving traditions, confirmation, consensus, process-based change, or epistemic states from their transitions. It did not. Popper, Kuhn, Lakatos, Laudan, Bayesian confirmation theory, social and network epistemology, process-based accounts of change, and scientonomy already explain or organize much of the local terrain.

The residual problem addressed here lies at a narrower set of interfaces inside scientific correction. Scientific correction and scientific change overlap without being identical: some scientific changes are not correction-driven, while some correction episodes end without a change of the relevant operative framework.

A disturbance can be real without identifying what requires correction. A corrective problem can be genuine without yet constituting a mature epistemic case. A mature case can exist without immediately reorganizing scientific practice. And a transition can succeed while its epistemic warrant is transformed, inflated, narrowed, or obscured along the way.

This is why the apparent Popper–Kuhn opposition is partly a problem of analytical level.

Popperian corrigibility concerns the requirement that scientific commitments remain answerable to error.

Kuhnian normal science concerns the requirement that scientific commitments remain stable enough to support sustained inquiry.

These are not mutually exclusive requirements.

They describe two conditions that a functioning scientific correction regime must satisfy simultaneously:

Operational Stability

and

Corrective Reopenability.

The synthesis does not erase the substantive differences between Popper and Kuhn. It does not resolve every dispute concerning normal science, theory choice, incommensurability, realism, or scientific progress.

It establishes a narrower compatibility:

Scientific inquiry can require stable operative commitments while remaining structurally answerable to correction.

The further requirement proposed by the handoff analysis is that reopenability cannot exist merely as an abstract permission to criticize. In a correction episode it must survive the transformations through which a scientific disturbance acquires a target, a corrective case matures, and a specified community reorganizes its operative state.

The final principle is therefore:

Sustained corrigible scientific inquiry requires operational stability, corrective reopenability, and inspectable handoffs between the states through which correction becomes scientific change.

The present article stops at scientific correction. Downstream questions of policy, institutional authority, implementation, and governance require additional premises and distinct empirical frameworks. That limit is not a deficiency of the analysis; it is an application of its own central discipline: explanatory authority must remain indexed to the target actually demonstrated.

References

Barseghyan, H. 2015. The Laws of Scientific Change. Cham: Springer. DOI: 10.1007/978-3-319-17596-6.

Barseghyan, H. 2018. “Redrafting the Ontology of Scientific Change.” Scientonomy 2: 13–38. DOI: 10.33137/js.v2i0.31032.

Canali, S. 2022. “A Pragmatic Approach to Scientific Change: Transfer, Alignment, Influence.” European Journal for Philosophy of Science 12: 48. DOI: 10.1007/s13194-022-00477-7.

Castino, A. 2023. “Accepting Massive Problems: A Scientonomic History of Dark Matter.” Scientonomy 5: 1–23. DOI: 10.33137/js.v5i.42257.

CERN. 2011. “OPERA Experiment Reports Anomaly in Flight Time of Neutrinos from CERN to Gran Sasso.” CERN, 23 September 2011; updates 16 March and 8 June 2012. [Source]

Duhem, P. 1954. The Aim and Structure of Physical Theory. Translated by P. P. Wiener. Princeton: Princeton University Press. [French original 1906.] [Source]

Dyson, F. W., Eddington, A. S., and Davidson, C. 1920. “A Determination of the Deflection of Light by the Sun’s Gravitational Field, from Observations Made at the Total Eclipse of May 29, 1919.” Philosophical Transactions of the Royal Society of London. Series A 220: 291–333. DOI: 10.1098/rsta.1920.0009.

Earman, J. 1992. Bayes or Bust? A Critical Examination of Bayesian Confirmation Theory. Cambridge, MA: MIT Press. [Source]

Einstein, A. 1915. “Erklärung der Perihelbewegung des Merkur aus der allgemeinen Relativitätstheorie.” Sitzungsberichte der Königlich Preussischen Akademie der Wissenschaften, 831–839. [Source]

Howson, C., and Urbach, P. 2006. Scientific Reasoning: The Bayesian Approach. 3rd ed. Chicago: Open Court. [Source]

Isacks, B., Oliver, J., and Sykes, L. R. 1968. “Seismology and the New Global Tectonics.” Journal of Geophysical Research 73(18): 5855–5899. DOI: 10.1029/JB073i018p05855.

Kennefick, D. 2009. “Testing Relativity from the 1919 Eclipse—A Question of Bias.” Physics Today 62(3): 37–42. DOI: 10.1063/1.3099578.

Kuhn, T. S. 1970. The Structure of Scientific Revolutions. 2nd ed., enlarged. Chicago: University of Chicago Press. [First edition 1962.] [Source]

Kuhn, T. S. 1977. The Essential Tension: Selected Studies in Scientific Tradition and Change. Chicago: University of Chicago Press. [Source]

Lakatos, I. 1970. “Falsification and the Methodology of Scientific Research Programmes.” In Criticism and the Growth of Knowledge, edited by I. Lakatos and A. Musgrave, 91–196. Cambridge: Cambridge University Press. [Source]

Laudan, L. 1977. Progress and Its Problems: Towards a Theory of Scientific Growth. Berkeley: University of California Press. [Source]

Laudan, L. 1984. Science and Values: The Aims of Science and Their Role in Scientific Debate. Berkeley: University of California Press. [Source]

Laudan, L., Donovan, A., Laudan, R., Barker, P., Brown, H., Leplin, J., Thagard, P., and Wykstra, S. 1986. “Scientific Change: Philosophical Models and Historical Research.” Synthese 69(2): 141–223. DOI: 10.1007/BF00413981.

Le Pichon, X. 1968. “Sea-Floor Spreading and Continental Drift.” Journal of Geophysical Research 73(12): 3661–3697. DOI: 10.1029/JB073i012p03661.

Longino, H. E. 1990. Science as Social Knowledge: Values and Objectivity in Scientific Inquiry. Princeton: Princeton University Press. [Source]

Longino, H. E. 2002. The Fate of Knowledge. Princeton: Princeton University Press. [Source]

McKenzie, D. P., and Parker, R. L. 1967. “The North Pacific: An Example of Tectonics on a Sphere.” Nature 216: 1276–1280. DOI: 10.1038/2161276a0.

OPERA Collaboration. 2012. “Measurement of the Neutrino Velocity with the OPERA Detector in the CNGS Beam.” Journal of High Energy Physics 2012(10): 93. DOI: 10.1007/JHEP10(2012)093.

OPERA Collaboration. 2013. “Measurement of the Neutrino Velocity with the OPERA Detector in the CNGS Beam Using the 2012 Dedicated Data.” Journal of High Energy Physics 2013(1): 153. DOI: 10.1007/JHEP01(2013)153.

Overgaard, N. 2017. “A Taxonomy for the Social Agents of Scientific Change.” Scientonomy 1: 55–62. DOI: 10.33137/js.v1i0.28234.

Palider, K. 2019. “Reasons in the Scientonomic Ontology.” Scientonomy 3: 15–31. DOI: 10.33137/js.v3i0.33557.

Palider, K., Barseghyan, H., and Shaw, J. 2025. “Redefining Epistemic Stances.” Scientonomy 6: 1–22. DOI: 10.33137/js.v6i.46704.

Patton, P., Overgaard, N., and Barseghyan, H. 2017. “Reformulating the Second Law.” Scientonomy 1: 29–39. DOI: 10.33137/js.v1i0.27158.

Popper, K. R. 1959/2002. The Logic of Scientific Discovery. London: Routledge. [English edition first published 1959.] [Source]

Quine, W. V. O. 1951. “Two Dogmas of Empiricism.” The Philosophical Review 60(1): 20–43. DOI: 10.2307/2181906.

Roe, S. M. 2017. “The Journey from Discovery to Scientific Change: Scientific Communities, Shared Models, and Specialised Vocabulary.” International Studies in the Philosophy of Science 31(1): 47–67. DOI: 10.1080/02698595.2017.1370932.

Shwed, U., and Bearman, P. S. 2010. “The Temporal Structure of Scientific Consensus Formation.” American Sociological Review 75(6): 817–840. DOI: 10.1177/0003122410388488.

Tochinai, F. 2009. “Rejection and Acceptance of Plate Tectonics—a History of Earth Science in Postwar Japan.” East Asian Science, Technology and Society 3(1): 153–156. DOI: 10.1215/s12280-009-9078-5.

Vine, F. J., and Matthews, D. H. 1963. “Magnetic Anomalies Over Oceanic Ridges.” Nature 199: 947–949. DOI: 10.1038/199947a0.

W3C Provenance Working Group. 2013. PROV Model Primer. W3C Working Group Note. [Source]

Wegener, A. 1915. Die Entstehung der Kontinente und Ozeane. Braunschweig: Friedr. Vieweg & Sohn. [Source]

Worrall, J. 2025. “Epilogue: Scientific Theory-Change and Rationality—Lakatos and the ‘Popper-Kuhn Debate’.” In Proofs and Research Programmes: Lakatos at 100, edited by R. Frigg, J. M. Alexander, L. Hudetz, M. Rédei, L. Ross, and J. Worrall, 247–264. Synthese Library 498. Cham: Springer. DOI: 10.1007/978-3-031-88213-5_14.

Zollman, K. J. S. 2007. “The Communication Structure of Epistemic Communities.” Philosophy of Science 74(5): 574–587. DOI: 10.1086/525605.