Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Sources and Methods in Insurance History

Abstract. The history of insurance is reconstructed from documentary remains produced for purposes that frequently differed from those of subsequent historians. Notarial instruments, merchants' account books, underwriting slips, corporate policy registers, claims, statutes, litigation, actuarial tables, correspondence, colonial records, and institutional retrospectives record different stages of risk allocation and different relations between law, finance, commerce, and society. This monograph establishes a critical evidentiary approach to those materials. It examines the boundaries among contingent lending, indemnity, mutual assistance, wagering, and actuarially administered assurance; evaluates the meaning of an alleged historical first; investigates provenance, chronology, transcription, translation, selection, and missing archives; and develops replicable procedures for cross-jurisdictional comparison, contract coding, historical quantitative analysis, and causal interpretation. Detailed cases include the Code of Hammurabi, fourteenth-century Mediterranean marine insurance, the 1601 and 1662 English statutory record, London fire insurance registers, early mortality documentation, and the insurance of the transatlantic slave economy. The central conclusion is that historically defensible insurance claims require an explicit definition of the institution under investigation, a located and dated source, a documented chain between evidence and inference, and an account of what the surviving record leaves unobserved. The work presents original analytical synthesis and applied source criticism; it establishes no new theory, doctrine, or proprietary historical model.

Keywords: Insurance history; historiography; archival provenance; marine insurance; notarial records; source criticism; risk transfer; firstness claims; comparative legal history; insurance contracts; actuarial history; historical statistics; archival survival; colonial insurance; research reproducibility.

Introduction

Insurance has several histories. One follows arrangements through which traders or communities redistributed losses. Another follows the emergence of contracts whose consideration purchased specified protection from uncertain events. Others follow the consolidation of underwriting markets, the growth of capitalized insurers, the regulation of insurance contracts, and the statistical administration of life contingencies. These histories overlap, but they do not have interchangeable starting points. A claim that a Babylonian norm allocated risk, a Genoese deed transferred a maritime peril, and a London company issued a fire policy may each be well supported without entailing that the three documents attest to a single unchanged institution.

The research problem is therefore evidentiary and conceptual at once. Evidence must be identified, dated, situated, read according to the terminology and legal practice of its own setting, and related to a carefully specified historical proposition. The historian must also explain why the document was created, whose interests shaped its wording, whether it was enforceable, how it survived, and what kinds of contemporary transactions left fewer traces. The international archival description approach associated with the International Council on Archives makes the relation among records, their creators, activities, and contexts a fundamental feature of description. That relation is equally fundamental to insurance history.

The strongest available medieval insurance research demonstrates the opportunity and the difficulty. Botticini, Buri, and Marinacci (2023) combined notarial records, merchant accounts, and court proceedings in a large comparative insurance dataset, while directly discussing selection by city, notary, and archival survival. Their evidence supports major propositions about contracts and the distribution of maritime hazards; it also shows why the surviving document is a selected observation. The analysis in this dossier builds upon such research while distinguishing archival facts, claims made by published scholars, and the independent interpretive conclusions advanced here.

The documentary scope spans ancient legal translations, medieval Mediterranean commerce, early modern British insurance, historical mortality and actuarial materials, and comparative global markets. The work evaluates online editions, official catalogues, legislation, and directly accessible research articles. It does not represent a physical inspection of every underlying manuscript or a fresh archival transcription of the original fourteenth-century deeds. That access boundary is attached to specific claims throughout. Its significance lies in specifying what can be established from the accessible evidence, what requires comparison with manuscript originals, and what cannot be inferred from absence.

Historical Sources and Their Provenance

A historical source is an artifact created or maintained in a particular practice and subsequently available as evidence about that practice. For insurance, the artifact may be the signed policy, a notary's register, a ledger recording the premium, correspondence about a maritime voyage, a tribunal's judgment, a charter, or an annual report. The recording act itself was selective. Notarial instruments formalized legally significant transactions; account books served commercial control; statutes expressed authoritative legal commands; judicial files preserved disputes; company histories curated institutional identity. Their contents acquire historical meaning through the purposes and limitations of their production.

Provenance answers more than the question of where a document is stored. It addresses the person or body responsible for its creation, the functions through which it arose, custody, copying, arrangement, later description, and any known alteration. The Records in Contexts conceptual model explicitly joins record resources with agents and activities. The United States National Archives likewise defines original archival order through the arrangement created by the responsible organization or person. For insurance documents, original order can reveal whether an isolated policy belonged to a chronological series, a commercial account, a packet of litigation papers, or a deliberately collected set of exemplary transactions.

A useful documentary description consequently distinguishes the record's date from the catalogue entry's date and from the date on which a digital copy was made. A deed signed in March 1343, cited in a twentieth-century edition and described on a contemporary website, has at least three different temporal coordinates. The earliest is relevant to the historical transaction; the later dates indicate transmission and scholarly access. An undated website cannot independently upgrade the precision of the historical date, and the date of cataloguing does not establish the date of the activity described.

A second distinction is between document form and document function. A merchant might employ the language of a sale or loan to achieve a transfer of maritime risk, while the commercial substance of the undertaking resembled insurance. Conversely, the use of a later translator's word 'insurance' in a description of an ancient loss-sharing rule does not demonstrate the presence of a premium-based policy. Critical reading must examine the operative clauses, allocation of payment upon loss or safe arrival, duration of protection, identity of the beneficiaries, and commercial incentives of the parties. Nomenclature supplies evidence about contemporary legal language, and contractual obligations supply evidence about economic substance.

Third, provenance has a political dimension. Colonial administrations, merchants, insurers, courts, and commercial associations possessed unequal resources for recording and conserving their activity. Communities operating through customary obligations or informal support could leave fewer surviving written instruments. Equating archival density with historical importance would privilege the institutions already advantaged in creating durable written records. The correct response is to state the survival mechanism, seek independent types of evidence, and qualify any comparison that depends upon systematically unequal documentation.

Source description should therefore preserve the archive or custodian, collection or fonds, series, item reference, original language, document type, reported date, edition or transcription used, access mode, pertinent text or metadata, and the historian's interpretation. Even when only a published study is accessible, the historian can identify the intermediate evidence chain. A reference to an archival article and a claim to have inspected its underlying deed have different evidentiary meanings, and a scholarly narrative must preserve that distinction.

Defining Insurance for Historical Research

The historian must define the institution for each inquiry before seeking its earliest occurrence. A broad inquiry into risk allocation includes mutual aid, indemnity obligations, contingent debt, public compensation, and contractual insurance. A narrow inquiry into independent commercial insurance requires evidence of a defined uncertain peril, an undertaking by an identified risk bearer, an identifiable economic consideration or contribution, and a consequence contingent upon the peril. Different arrangements may meet some of these criteria without sharing the full legal architecture of a later insurance policy.

The Code of Hammurabi in translation, associated with L. W. King, provides an instructive boundary. Provision 103 releases an agent from a merchant's claim after losses caused by an enemy on a journey, subject to an oath. Provision 102 addresses liability for a loss suffered by an agent entrusted with money. These rules allocate commercial loss and specify evidentiary obligations. They do not themselves display a policy in which a separate insurer promises indemnification in exchange for a stipulated premium. Provisions 23 and 24 also describe collective compensation in circumstances of robbery, but their presence establishes a legal loss-allocation mechanism rather than a commercial insurance market.

The Library of Congress historical guide uses an expansive genealogy, relating early insurance to Hammurabi-era bottomry. Its summary is useful evidence of a common historiographical convention. Yet the direct text of the translated code and the guide's retrospective insurance vocabulary belong to different kinds of source. The code can substantiate specific obligations; the guide cannot make an otherwise undocumented private insurance transaction appear in the code. A careful statement is that ancient commercial law contains risk-contingent obligations and historically relevant antecedents to later insurance. Whether a particular ancient loan instrument should be classified as insurance requires that instrument's actual terms, not a generalized date assigned to a legal code.

The comparative issue becomes clearer through the Marine Insurance Act 1906. Its section 1 defines marine insurance by an insurer's undertaking to indemnify the assured against maritime losses as agreed. This is a legally situated definition enacted in a particular jurisdiction in 1906. It is indispensable for reading the legal object the statute regulates and useful as a later comparator, but projecting all its elements backward as compulsory universal conditions would erase earlier varieties of maritime finance. The task is to analyze historical relations between contractual forms while maintaining the difference between an analytic classification and a rule binding parties at the historical time.

Insurance may also be distinguished from wagering by attention to the protected interest and legal policy. The Life Assurance Act 1774 rejected certain policies lacking an interest and directed against gaming or wagering. The existence of such a prohibition establishes that the legislature distinguished legitimate assurance from wagers and considered the boundary socially consequential. It does not establish that every earlier life-contingent arrangement was a wager, or that the wording of the 1774 law applied universally across empires and centuries.

At least four historical dimensions must remain analytically distinct: the allocation of loss, the existence of a commercial promise of protection, the emergence of an organized insurance market, and the operation of a statistical insurance enterprise. The breadth of the first allows study of ancient risk-sharing practices. The specificity of the second assists document-level recognition of maritime insurance. The third requires evidence of recurring transactions, intermediation, and underwriting relationships. The fourth demands data about valuation, reserves, mortality, and organizational administration. These are levels of empirical inquiry rather than evolutionary stages that every society must traverse.

Chronology and Priority Claims

The proposition 'the first insurance contract' has no unique meaning until it supplies a class of documents, a geographical scope, a criterion for insurance, and a rule about survival. It may mean the earliest surviving dated instrument found by current researchers; the earliest surviving instrument explicitly called insurance; the oldest currently identified premium entry; the earliest legal authorization; or an inferred historical innovation that preceded the surviving record. Conflating these propositions converts a document into an unverifiable global priority claim.

A priority assessment should ask: What precisely is being dated? Does the date belong to the transaction or to a later copy? Is the text complete? Is its provenance documented? Are there earlier candidates in neighboring legal traditions? Does the instrument have the contractual properties required for the stated category? Is the earliest known surviving instance being presented as the first event that occurred? Each answer alters the defensible conclusion.

Botticini, Buri, and Marinacci (2023) identify the notarial instrument drafted by Tommaso Casanova on March 18, 1343, in Genoa as the document commonly regarded as the first modern insurance contract. Their discussion of the 1343 Genoese deed simultaneously reports an insurance contract from Palermo dated to the same year and reasons that such records may indicate earlier practice. Iodice (2023), in a study of Genoese marine underwriting, likewise reports the conventional Casanova chronology. The properly supported statement is about a dated and historiographically prominent surviving document, accompanied by the demonstrated existence of contemporary evidence elsewhere. The accessible research does not establish an exclusive universal moment of invention.

The Palermo evidence makes the difference between a first document and a first market visible. A merchant capable of obtaining assurance in a city other than the city associated with the conventional foundational deed probably participated in a practice involving knowledge, counterparties, and commercial procedures that had developed before the document's execution. This is an inference about institutional preconditions, not a recovered missing date. It does not entitle the historian to name the location or year of the earliest transaction without supporting evidence.

Priority requires a clearly limited historical negative. One may establish that no earlier document has been identified within a specified archival corpus and search protocol. One cannot infer from the current corpus that no earlier contract ever existed. The asymmetry is especially important when one jurisdiction relied upon public notaries and another used private merchant letters that were more vulnerable to accidental loss. In such conditions an apparent chronological lead can be partly a lead in the production of preservable documents.

The practical output should be phrased in graduated form: an instrument 'is dated' when the dating evidence is direct; 'is the earliest identified in the examined corpus' when the search boundary is described; 'is widely regarded as an early modern insurance policy' when the statement belongs to the secondary literature; and 'indicates an established practice by' when the record warrants an inference about an operating institution. These constructions convey the content of the evidence without relying upon unsupported universal superlatives.

Medieval Mediterranean Contract Evidence

The most developed accessible illustration is the open-access economic-historical study by Botticini, Buri, and Marinacci (2023). Their analysis of medieval insurance records covers approximately seven thousand contracts and 104 commercial court cases dating between 1343 and 1506, drawing from Genoa, Palermo, Florence, Prato, Barcelona, and Venice. The article reports 5,031 contracts from previously unexplored archival sources, including 2,531 from Genoa, 2,310 from Palermo, and 190 from the Datini material in Prato; it also reports 1,882 contracts taken from existing historical publications. These distinctions matter because 'the dataset' is an aggregation of evidence gathered by different routes and under different selection decisions.

The article provides an unusually explicit account of preservation and collection: Genoese notaries were selected because their surviving registers contain a high volume of insurance transactions; Palermo data cover surviving records from the relevant notaries in the specified historical interval; Florentine agreements frequently took the form of private merchant contracts rather than notarized acts; and the surviving share of historical contracts cannot be determined. The researchers compare the commodities and routes represented in the documents with independent histories of trade as a check on plausibility. Such triangulation adds confidence about observed patterns but does not supply an unknown denominator for all medieval underwriting.

A Palermo deed dated February 25, 1347, discussed in the published contract reconstruction, concerns a voyage from Palermo to Pisa, with cheese and leather carried by galley. The research identifies an insured amount of 100 gold florins and a 7.5 percent premium, with named parties and clauses about the risks protected and the time of loss settlement. The point is not that every contract from Palermo shared these exact provisions. It is that a dated instrument can be decomposed into commercially meaningful fields and then compared with other documents under a common codebook while retaining local terminology.

Genoa and Florence also illustrate why formal appearance is an unstable criterion. The study describes disguised sale and debt forms in Genoa, and comparatively explicit private instruments in Florence. A corpus limited to notarized texts would record Genoa more readily than privately retained Florentine policies; a corpus selected only by the literal word for insurance might miss disguised Genoese instruments. The resulting chronology could reflect source architecture rather than actual rates of innovation. Researchers must therefore explain why their search vocabulary and archival classification identify the relevant economic activity across changing forms.

The study's economic analysis associates premiums with route-specific risks, including attacks by corsairs, as well as distance and season. These findings are valuable because they connect documents to commercial decisions. Their interpretation still depends on how reported premiums, routes, commodities, ship categories, and unobserved hazards enter the sample. A correlation in surviving priced policies cannot alone establish the premium that every voyage would have paid or prove a universal preference for a specific class of risk. Contract data reflect both the underlying hazards and the bargaining, information, and selection that determined which voyages were insured and recorded.

The methodological lesson is to analyze three layers together. The first is the textual record: what the deed says. The second is its transactional setting: which merchant, vessel, route, and promise it concerns. The third is the sampled archive: which transactions survived, which were collected, and why. A rigorous history of medieval insurance requires all three; a count of contracts detached from their archival settings is not yet a history of insurance markets.

Reading Contracts and Reconstructing Transactions

A historically responsible contract reading begins with the physical or diplomatic object. The researcher identifies whether the available material is an original, a contemporary copy, a register entry, a transcription, a translation, an edited excerpt, or a secondary paraphrase. These types differ in what they preserve. An abbreviated register entry may contain parties and amounts but omit clauses; a translation may normalize a technically significant word; an abstract may suppress exceptions; a contemporary copy may have been made to support litigation. The strength of the claim must be calibrated to the type of witness.

The next step is to identify parties and capacities: merchant, insured, underwriter, broker, agent, carrier, creditor, notary, adjudicator, and beneficiary. A single individual could occupy more than one capacity. Identical surnames may refer to different people, and historical naming practices can vary within one document. A researcher needs explicit rules for matching identities across records, retaining uncertainty and avoiding automatic merger of similarly named actors. Names associated with women, enslaved persons, religious minorities, and other groups may appear unevenly, creating further distortions in reconstructed social networks.

The operative provisions should be recorded separately from the historian's classification. Relevant fields include covered property or person; voyage and route; start and termination of risk; sum insured; payment consideration; named perils; excluded perils; event triggering payment; obligations after a loss; evidence requirements; forum; signatures; and relevant legal formulae. Blank fields must retain the distinction between 'not present in the source', 'illegible', 'unavailable in the edition', and 'not applicable'. Replacing all these with a zero silently creates false observations.

Contractual substance often lies in the conditional movement of money. A maritime loan may place the capital of the voyage with a creditor and condition repayment on safe arrival. An indemnity instrument may instead require a risk bearer to pay the protected merchant after an insured loss. These arrangements can have related economic effects while producing distinct legal claims and accounting entries. To compare them, the historian should reconstruct the flows under the relevant contingencies: safe arrival, partial damage, total loss, seizure, diversion, default, and disputed notice. Such a reconstruction is an analytic reading of existing contracts, and it does not transform diverse historical practices into a single newly invented definition.

Evidence of enforcement is particularly important. The language of an obligation establishes the parties' expressed undertaking. A judgment, settlement record, or actual claim payment can establish how that undertaking was interpreted or performed. The 104 commercial court cases used by Botticini, Buri, and Marinacci (2023) show why merchant court records add information unavailable from policy texts alone. Litigation, however, is a selected population of disputes, generally enriched for conflict and ambiguity. It should be used to interpret legal practice without assuming every unlitigated contract behaved in the same way.

Editorial reliability requires that every English quotation be traceable to an edition and that the researcher distinguish translation from transcription. A historical phrase can be paraphrased in present-day English if the transformation is identified; an interpretive paraphrase must not be disguised as direct original wording. Where a clause is decisive, the original-language expression should be retained in working notes and checked against a reliable edition. The present study does not reproduce unpublished manuscript transcriptions or claim that unavailable paleographic details were examined directly.

Law, Ordinances, and the Evidence of Regulation

Statutes and ordinances provide unusually clear documentary dates, authoritative language, and evidence of public intervention. Their apparent precision invites a mistake: the date of a law is frequently mistaken for the date at which the activity it regulates began. A statute can recognize existing practices, alter enforcement, constrain conduct, or codify rules that had already developed through commerce and litigation. Its existence may demonstrate an established institutional problem precisely because the law was responding to it.

The Policies of Assurance Act 1662 illustrates this retrospective dimension. Its opening recital refers to an Elizabethan statute concerning assurances among merchants and discusses the practical obstacles faced by the commission appointed to adjudicate policy disputes. This makes the 1662 enactment evidence both for a legal mechanism and for the earlier mechanism it explicitly describes. The later text must not be treated as a verbatim substitute for every clause of its predecessor. Its historical testimony consists of what the 1662 legislature recited and altered, while a full reconstruction of the Elizabethan regime requires the appropriate earlier version as an additional witness.

A precise legislative analysis separates the instrument's passage, commencement, jurisdiction, later amendments, editorial notes, and the version used by the scholar. A modern online legislative database may display current consolidated text, original enacted text, historical annotations, and commencement data. The historian citing the original commercial setting must identify the historical version. The case is stronger when the researcher reads the 1662 enacted statutory wording instead of presenting a later summary as the initial legislative record.

The Life Assurance Act 1774 supplies a different kind of evidence. Its first section addresses policies in which the person for whose benefit insurance is made has no interest and explicitly mentions gaming and wagering. The researcher may use the text to demonstrate a legal distinction enforced in a specified jurisdiction at a specified time. The text alone cannot measure the prevalence of unlawful wagers, identify every enforcement decision, or show how other jurisdictions treated comparable policies. Those questions demand reported cases, commercial records, correspondence, and quantitative series.

The Marine Insurance Act 1906 is especially valuable as a systematic codification of marine insurance doctrine. Section 1 supplies the statutory definition; other provisions address the insured interest, policy, representation, warranties, loss, and recovery. It can be used as evidence of a legal arrangement consolidated in the early twentieth century and as a structured comparison point for earlier commercial language. The methodological safeguard is to preserve its historical jurisdiction and enacted date in every comparison. A modern section heading is an aid to locating the text, not proof that medieval merchants organized contracts through the same taxonomy.

Early Barcelona insurance ordinances demonstrate another evidentiary problem: citation chains in which a modern official report refers to an older ordinance through intervening editions. The Spanish Consejo de Estado legal opinion is an authoritative source for the content of that modern legal examination, and it is a useful route to the earlier history it discusses. It is not itself an original 1435 municipal ordinance. A claim about the exact historical text of a medieval enactment requires a critical edition or archival witness, with variants and translation decisions stated. Merely citing a twenty-first-century opinion as though it were a fifteenth-century manuscript would collapse two very different steps of the evidentiary chain.

Regulation also generates sources by compelling institutions to keep registers, issue policies, report capital, and explain disputes. This produces a systematic survival asymmetry. Activities outside legal or administrative reporting may remain socially important yet poorly documented; regulated companies may appear disproportionately visible simply because they were required to preserve records. The historian should therefore distinguish institutional prevalence from documentary prevalence and investigate the regulatory processes that generated the archive under study.

Corporate Archives, Fire Insurance, and Market Reconstruction

Corporate policy registers are exceptionally useful because they often record routine transactions rather than only extraordinary legal disputes. Their survival can support analysis of the geography of demand, occupations, insured property, premium payments, customer networks, and the administrative scale of an insurer. They also present identifiable selection problems: records belong to particular firms; commercial categories may change; branches may retain data differently; and policies can be missing because of fire, flood, relocation, disposal, or later administrative reform.

The London Archives fire-insurance guide describes collections from the Sun Fire Office, Royal Exchange Assurance, and Hand-in-Hand Fire Office. It states that policy registers generally provide the policy number, date, holder's name, occupation and address, property description, and premium. It also identifies endorsements and the relevant indexed series. These metadata fields are enough to establish a useful research design, though they are not the same as a completed transcription of every policy. The guides and inventory references identify what can be investigated; actual item-level claims depend on the inspected register.

For the Sun Fire Office, the catalogue lists Town Department policy registers from 1710 to 1863 under CLC/B/192/F/001 and Country Department registers from 1793 to 1863 under CLC/B/192/F/003. It also describes separate loss records and management-committee materials. This creates a documentary opportunity to connect exposure and premium data with loss handling, but only after determining which records cover the same insured risks, years, and administrative units. A policy recorded in one book and a fire loss recorded in another cannot be joined solely because a surname or locality looks familiar.

The historian reconstructing a customer network might begin by transcribing a defined run of policy-register entries, preserving the original spelling of names and places. It should record the archive's shelfmark, page or folio, premium unit, policy term, endorsements, and any evidence of renewal. A normalized analytical file may then harmonize geographic locations and occupational classes, while a separate unaltered transcription preserves the wording for future scrutiny. Any probabilistic linkage of people or premises should retain match criteria and ambiguity flags. Reproducibility depends as much on the record of uncertain identity matches as on the final statistical table.

Company histories are a separate class of evidence. They can preserve useful dates and institutional self-understanding, but they also select achievements that support a corporate narrative. Lloyd's institutional history of coffee and commerce dates the first recorded reference to Edward Lloyd's coffee house to 1688 and describes its role in shipping information and underwriting. It supports a claim about a recorded institutional milestone. It does not establish that marine insurance began at Lloyd's: the Mediterranean documentary evidence predates that milestone by centuries. The difference between the history of a market venue and the history of a contractual practice is foundational.

Fire insurance history also requires careful causal wording. A destructive fire may strengthen demand for protection, but a chronological sequence from disaster to insurer formation does not, by itself, prove how much new demand arose, which organizational ventures would otherwise have occurred, or how fire protection was financially administered. Comparing dated corporate charters, subscriptions, premiums, fire-loss registers, and contemporary commentary offers a stronger causal explanation than a narrative based solely on the prominence of one catastrophe.

Actuarial and Demographic Sources

The history of life assurance and actuarial practice depends upon evidence different from that of maritime contract formation. Its key materials include population and mortality lists, original mathematical treatises, manuscript tables, life-office membership registers, valuations, premium schedules, financial statements, and institutional correspondence. A life-contingent calculation may circulate before an insurer uses it; a practical premium table may be employed before an institution names its mathematical method; an actuarial archive may preserve successive revisions that cannot be recovered from an official retrospective alone.

John Graunt's 1662 observations on the Bills of Mortality provide a documented example of early demographic analysis. The Wellcome Collection catalogue identifies the first-edition publication, printer, place, pagination, and surviving copy characteristics. Those bibliographic details matter. The edition consulted affects the tables and commentary available, and surviving copies may omit folded material. The catalogue specifically notes that its described copy lacks two tables. It would therefore be methodologically invalid to assert that those tables had been inspected in that particular copy solely because they belonged to the work's historical publication.

Edmond Halley's 1693 work on the Breslau mortality observations supplies another case. The Royal Society's catalogue of Halley's manuscript identifies the archive reference CLP/21/45, date, physical form, and its relation to the published discussion of mortality and life annuities. That metadata supports the source's existence and historical placement. Direct analysis of the numerical methods requires a reliable text or image of the calculation, careful treatment of calendar periods and age groupings, and comparison with the underlying mortality observations. Bibliographic identification by itself is not a fresh reconstruction of the mortality table.

The Institute and Faculty of Actuaries describes the Equitable Life historical archives and separates holdings in its London collections from later material at The London Archives. This geographical and chronological division is a reminder that an institutional archive can be distributed among custodians. A researcher evaluating a company's early mathematical practices should not assume that one digital finding aid contains the entire documentary record. Correspondence, valuation books, directors' minutes, printed tables, and later anniversary histories must be compared in their respective roles.

Quantitative historical reconstruction needs a distinction between observed vital events and model assumptions. Bills of mortality record events through the reporting systems of their time; life tables derive rates or expectations from classified observations and analytic choices. Observational boundaries may include parish coverage, disease classification, age reporting, changes in registration, migration, epidemic shocks, and treatment of missing records. The resulting estimates depend upon those choices and need not describe the mortality of every population or insured class. The historian's task is to reconstruct what the table's compiler could observe, what assumptions were imposed, and how the computed values were used by actual assurance enterprises.

Actuarial progress also has a history of diffusion. An analytic formula's publication, adoption in an office, incorporation into a standard, and use in annual valuations are distinct dated events. Chronologies that leap directly from a mathematician's treatise to the universal operation of statistical insurance obscure commercial adaptation and institutional constraints. The comparative global scholarship collected by Borscheid and Haueter (2012), World Insurance, situates technical insurance practices within markets whose development varied across regions, regulations, and economic conditions.

Cross-Jurisdictional Source Criticism

Insurance history is internationally comparative because merchants, insurers, capital, maritime routes, migration, empire, and reinsurance networks moved across legal and political boundaries. Comparison is nevertheless valid only when sources represent sufficiently similar phenomena and differences in record production have been identified. A Genoese notarial deed, a Florentine merchant account, a London fire-policy register, and a national supervisory report may each contain a date and a monetary figure while recording different units of activity.

The first task is linguistic. A historical legal word may identify a particular debt, sale, guarantee, wager, or indemnity, with meanings conditioned by the document's genre and period. Translation should preserve original terms in the source file, identify the language of the text, cite the edition used, and distinguish a literal translation from the historian's functional classification. The same English term can be a reasonable translation of several different historical expressions, but that linguistic convenience does not erase their legal differences. Cross-jurisdictional comparison should be conducted at the level of contractual obligations and institutional operation after, rather than before, examining local vocabulary.

The second task is legal. In England, the 1662 assurance legislation records an approach to administering disputes; the 1774 life-assurance legislation targets gambling-related concerns; the 1906 marine-insurance codification organizes a mature body of doctrine. These documents belong to different moments and subject matters even within one national tradition. A continental ordinance may address municipal trade while a British parliamentary statute governs a larger political jurisdiction. Comparison must record the authority, territory, scope, and historical reach of each instrument.

The third task is geographical and economic. A surviving commercial policy may identify the legal place of execution, the merchants' residences, the origin and destination of goods, the vessel's registry, and the place where a claim was payable. Those locations may span multiple jurisdictions. Attributing a contract to one national history solely from the notary's city produces misleading territorial boundaries. Medieval networks were mobile; the six-city Mediterranean evidence explicitly demonstrates merchants, trade routes, and institutional practices crossing municipal borders.

The fourth task is comparability of monetary observations. Premiums may be recorded in different units, on different insured amounts, for different voyages or periods, and under different conventions of discount, brokerage, and risk. Converting nominal historic money into a single modern currency can be useful for a precisely identified question, but the conversion requires a documented denominator, historically appropriate exchange rates, and a declaration of whether the quantity concerns purchasing power, bullion content, wages, prices, or the size of the underlying transaction. An unsupported present-value translation offers a false sense of comparability.

Comparative research must also escape the assumption that European corporate insurance represents the entire history of collective protection. The edited volume World Insurance: The Evolution of a Global Risk Network examines a wide range of countries, regional institutions, and international networks. Its account is particularly useful for studying the expansion of modern commercial insurers and the uneven adoption of particular institutional forms. That subject must be described accurately: the expansion of a European-origin corporate sector is a narrower proposition than the proposition that societies outside Europe previously lacked techniques for distributing loss or supporting vulnerable members.

An effective comparison therefore uses explicit units: contract, claim, insurer, market, regulated territory, period, or community practice. It reports which units are represented by each source and limits conclusions to those units. International scope arises from following evidence across its relevant regions and languages, not from forcing all regions into a single retrospective category.

Archival Gaps and Comparative Inference

Historical non-observation can result from several separate processes. A transaction may never have produced a document; a document may have been created and subsequently lost; a surviving document may sit in an uncatalogued collection; a catalogued document may use unexpected vocabulary; or a known document may remain inaccessible through language, restrictions, conservation, or funding limitations. Treating these mechanisms as one category of absence obscures both the reasons for uncertainty and the prospects for further research.

The medieval contract analysis provides a particularly useful empirical demonstration. Botticini, Buri, and Marinacci (2023) explicitly acknowledge that the share of contracts surviving from the original medieval population cannot be estimated. Their discussion of archival selection distinguishes city selection, selection of the notarial collections, and loss of historical documents. This is a stronger methodological position than assuming the count of surviving contracts can be converted into a count of the insurance market. A researcher may estimate relations within the observed sample and test them against independent trade histories while preserving uncertainty about the missing population.

Selection can distort descriptive comparisons in several directions. A notary specializing in insurance will have a different portfolio from a notary chiefly recording land transactions. An insurer that retains policy books but discards denied claims will show a different recorded loss distribution from one whose litigation archive survives. A company reporting large premiums may be unusually transparent under a particular supervisory rule. These differences can affect apparent frequency, pricing, geography, claim severity, and the identity of counterparties. The historian should identify the observation process for every dataset and, where possible, compare its composition with external commercial measures.

Silence in one type of archive is rarely a decisive negative. The absence of a premium in a Genoese deed may reflect a legal form that expressed insurance through a fictitious sale or loan; the presence of a premium in a private Florentine account may reflect a different recording convention. The medieval contracts study reports these contrasting documentary practices. An archive-based chronology limited to explicit premiums would assign an artificial early lead to practices whose price was written more transparently. An inference about absence needs to show that the source was expected to record the feature if it existed.

Archival concentration also interacts with the history of power. Records held by well-capitalized merchants, colonial states, public corporations, and wealthy insurers may be better preserved than those associated with informal mutual aid, excluded groups, itinerant merchants, or displaced communities. This inequality affects what the historian can recover and which voices are represented in the documentary narration. The proper response is to broaden the source base and state the likely directions of missingness, while declining to supply unobserved numerical magnitudes without an independent evidentiary basis.

Robust comparisons can use several approaches without pretending to recover the missing archive. One is to repeat the analysis with and without heavily documented firms or notaries. Another is to compare records produced by different institutions, such as contracts and disputes, to determine whether a pattern survives changes in source type. A third is to limit the claim to an interval, population, or documented archive. A fourth is to formulate competing explanations that would produce different surviving traces and seek discriminating evidence. These are ordinary techniques of historical inference; their use does not constitute a newly named theory.

Historical Quantification and Reproducible Data

Historical insurance documents are often machine-readable only after acts of interpretation. The transformation from manuscript to dataset requires a documented workflow: identify the witness, transcribe or extract its text, normalize selected fields, code the contractual relationship, assess completeness, and preserve both the original evidence and the analytical decisions. A row in a spreadsheet is an interpretation of a document, not the document itself.

The unit of observation must be explicit. A policy, line of insurance, renewal, endorsement, merchant, vessel, claim, judgment, and premium entry are not interchangeable. A single transaction may generate several records; a single policy may cover multiple cargo interests; multiple underwriting signatures may allocate parts of one loss. Without stable record identifiers and entity relationships, aggregation can count duplicated paperwork as separate commercial activity. The research database should preserve original document IDs, mapped transaction IDs, and links among related records, with an audit of joins and omissions.

A source-oriented codebook should define each variable, acceptable values, original wording, missing-value categories, coding rules, and the kind of evidence needed. For maritime contracts, core variables include the transaction and document dates, instrument form, contracted parties, origin and destination, vessel, goods, insured value, premium, stated perils, exclusions, claims venue, and manuscript reference. For fire insurance, policy identifier, location, occupancy, structure, sum, premium, endorsement, and renewal status may be central. For life assurance, entry age, assured event, table, valuation date, benefit terms, and reporting population can matter. These fields reflect distinct research questions and source capacities.

When a calculation follows from coded data, the computation should be inspectable. Suppose a hypothetical archival sample has 40 transcribed policies, of which 10 contain an unambiguously recorded premium. The fraction 10/40 describes premium observability in that sample. It is not evidence that one quarter of historical policies charged premiums; 30 observations may simply omit the price, and the sample may already be selected. This example is illustrative rather than an empirical claim about a named archive. A correct research report would publish the denominator, selection criteria, and treatment of missing fields alongside any statistic.

Temporal analysis introduces further challenges. A notary's recording date may differ from the voyage's departure; a contract may be renewed; a claim may be decided years after the loss. Annual totals compiled from such fields measure different events. The calendar and date conventions should be retained when uncertainty matters, and normalized dates should be marked as interpretations. Currencies, units of weight, distances, route names, and historic place names require likewise documented transformations. A clean digital table should not hide ambiguities that were visible in the source.

The medieval commercial dataset provides a useful reference for reporting provenance by archival location and for acknowledging selection problems in the collected corpus. Independent replication of its statistical estimates would require the complete data and methodological specifications, and any reproduction must separate the authors' original dataset from later reinterpretations. The present dossier evaluates the published methodology and archival descriptions; it does not claim to have re-estimated the article's regression results or inspected each underlying notarial folio.

Reproducibility also entails a stable bibliography and versioned source ledger. Every external research link in the manuscript should point to the exact document or institutional catalogue supporting the adjacent claim. Every accessible witness should have a complete reference. The evidence record must identify the claim checked, what was accessible, how it was evaluated, and where further manuscript-level verification would be necessary. Those requirements are especially important when research is later rendered into HTML, supplemented, or reused across distinct Science areas.

Causation, Markets, and Institutional Change

Insurance history asks explanatory questions as well as descriptive ones. Why did particular contracts appear? Why did some merchant networks establish routine underwriting, while others continued to rely upon credit, kinship, or communal support? Why did life tables become influential in some organizations and not in others? An adequate explanation must identify an outcome, specify its historical interval and mechanism, test competing causes against available records, and preserve the possibility that several processes operated simultaneously.

Botticini, Buri, and Marinacci (2023) connect medieval marine insurance with the expansion of seaborne trade, uncertainty about human and natural perils, information asymmetry, and the capacity of wealthy merchants to bear distributed risks. Their economic-historical account of insurance emergence brings substantial archival evidence to an interpretation of markets. Its analytical strength lies in joining the dated transactions to plausible mechanisms rather than offering a single invention story. The historian should nonetheless separate the evidence that the contracts existed, the reported associations between contract characteristics and premiums, and the further argument about why those associations developed.

Institutional change can be explained from different evidentiary directions. The growth of regular underwriting might follow increased maritime hazards, expansion of trade volume, development of liquid merchant capital, wider availability of information, changing norms of contract enforcement, or access to brokers. A single chronology may be consistent with several of these mechanisms. The test is whether the expected documentary traces appear: changed route composition, premiums, risk language, capital arrangements, court practice, or correspondence about the commercial need for protection. Evidence of one such mechanism does not automatically exclude the others.

The worldwide insurance history edited volume by Borscheid and Haueter (2012) situates expanding modern insurance organizations in migration, international trade, regulation, national institutions, and political disruption. This comparative history supports analysis of diffusion, but diffusion does not mean that an institution was reproduced unchanged. A legal transplant may encounter different local courts, ownership patterns, supervisory expectations, religious frameworks, and competitive environments. The resulting national or regional forms are historical transformations to be explained from local sources, rather than simple copies of a supposedly universal organizational template.

Corporate accounts invite quantitative causal claims, but their measurement limitations must be visible. An increase in written premiums can mean more customers, higher insured values, changes in price, a new accounting method, acquisitions, or an expanded geographical network. A change in loss ratio can reflect genuine hazards, policy exclusions, claims handling, inflation, reserve practices, or selection of insured exposures. Before making a causal claim, the historian should reconcile accounting definitions and identify whether the numerator and denominator refer to the same economic object over time.

The same discipline applies to disaster narratives. An event like a large urban fire may catalyze institutional innovation, but the observed sequence requires independent evidence of demand, capital formation, organizational decisions, and policy uptake to support a causal interpretation. A corporation's later celebration of its founders is evidence of commemorative identity; a contemporary subscription book or court dispute may give better evidence of the original economic motivation. Source genres should be weighted according to the proposition, not simply their prestige or narrative appeal.

Historical counterfactuals are legitimate when framed as comparisons with documented alternatives or explicit causal assumptions. The absence of one event, company, or statute from the surviving record cannot establish what would have happened in an otherwise identical world. Strong causal explanation is built from observable sequences, institutional incentives, contrasting cases, and the chronology of mechanisms. It is strengthened by making the inferential steps legible to a reader who may disagree about the interpretation but can inspect the evidence on which it rests.

Empire, Enslavement, and Ethical Source Criticism

Insurance contracts contributed to the organization of commerce under profoundly unequal social and political systems. A historical account restricted to the internal efficiency of risk transfer would omit the identity of those whose labor, confinement, dispossession, and mortality could be treated as commercial exposures. Insurance records can reveal networks of finance and responsibility while also recording people through the categories imposed by exploitative institutions. The historian must analyze those categories as evidence of the system that produced them.

Lloyd's public account of its slave-trade connections acknowledges the participation of its historical market in insuring the transatlantic slave economy and describes the work of archival curation and research. This institutional account is useful both for documented corporate acknowledgment and for tracing the research that the institution now publishes. It must be read critically as a contemporary narrative about institutional history, not as a complete archive of every policy, participant, claim, or victim.

Nicholas Draper's research on marine insurance and slavery provides a more granular historical analysis. The text explains how insurance was integrated into maritime commerce associated with slave trafficking and plantation economies, and it distinguishes voyages carrying captive Africans from the broader movement of plantation commodities. It explicitly notes the fragmentary character of surviving insurance records outside incorporated firms. This qualification matters for any estimate derived from the eighteenth-century market. The article presents estimates of slavery-related shares of premium income; their historical relevance is substantial, but they remain estimates built from incomplete records and specified assumptions rather than a census of every policy.

A record of an insured ship or cargo gives evidence about the legal and financial organization of commerce. It does not confer moral legitimacy on the institution described, and the categories of the insurer do not exhaust the historical experience of enslaved people. A robust historiography should connect underwriting documents with port records, shipping lists, legal disputes, emancipation and compensation records, testimonies, correspondence, and independent social histories when investigating the human consequences of insurance-enabled trade. Those connected sources may contradict the abstractions of commercial accounts.

Ethical source criticism has an epistemic dimension. Naming practices, occupational classifications, and valuations often reflect the standpoint of owners, traders, states, or insurers. Statistical aggregation can erase individual experience; corporate retrospectives can emphasize technical progress while minimizing coercion. The remedy is precise attribution, identification of the producing institution's interests, reconstruction of the social relationships omitted from the ledger, and clear expression of the limits of corporate evidence. This is part of accurate history, not an optional moral preface.

International comparison must also acknowledge that the ability to create, own, sell, or claim an insurance policy depended on legal personality, property regimes, race, class, gender, colonial hierarchy, and institutional recognition. Different populations could be present in insurance archives as customers, workers, insured subjects, claimants, or imposed objects of economic valuation. Conflating these positions would distort the historical distribution of risk and protection. The language of modern commercial records is itself a primary source about relations of power and must be interpreted with that awareness.

A Worked Evidence Audit

The following four examples demonstrate how an explicit research proposition can be assessed through identifiable evidence and bounded conclusions. They constitute an applied audit of cited sources available for this dossier, not a claim of original discovery of unseen archival manuscripts. The purpose is to exhibit the reasoning that separates a legitimate claim from an overly broad historical assertion.

Case: ancient commercial loss allocation

Proposition: a Babylonian legal text provides evidence of loss-contingent commercial obligations. The witness is L. W. King's published English translation of Hammurabi's provisions, especially provisions 102 and 103. The translated text differentiates an agent's responsibility for a trading loss from exemption following a hostile seizure under oath. This supports the proposition that the legal tradition recognized conditional allocation of commercial loss. It does not establish a separate insurer, independently priced premium, or enduring insurance market in the specific clauses examined.

The Library of Congress insurance guide connects ancient commercial practice with an expansive account of insurance origins. The apparent tension between the guide and the code is resolved by distinguishing antecedent risk allocation from the narrower claim of a premium-based policy. The conclusion is positive and limited: evidence exists for contractual and legal management of uncertain commercial losses; the inspected translation does not by itself document the fully specified later insurance transaction.

Case: medieval contractual priority

Proposition: a specific Genoese instrument dated March 18, 1343, is a prominent early example of modern marine insurance. The evidence is the scholarly reconstruction in Botticini, Buri, and Marinacci (2023), supported by the comparative chronology in Iodice (2023). The Genoa and Palermo documentary discussion directly warns against equating the earliest highlighted deed with the first contract that must ever have been executed. It identifies another Palermo contract from the same year and discusses the institutional forms through which insurance developed.

The evidence supports the document's dated scholarly attribution, while a complete independent diplomatic verification of the Casanova original would require the archival item or an accepted full critical edition. The observed coexistence of transactions across centers supports an inference of pre-existing commercial knowledge. It does not determine which merchant first created such a contract. A universal 'invented on March 18, 1343' statement would therefore exceed the evidentiary base.

Case: the structure of a fire-policy archive

Proposition: surviving London fire-policy registers can support research into policyholder identities, occupations, properties, dates, and premiums. The evidence is The London Archives' fire insurance research guide, which lists those fields and describes identifiable series. The conclusion is that a suitable archival corpus exists for a defined kind of reconstruction. The guide cannot itself establish how many policies were issued in a selected year, which households were insured, or how a particular loss was paid. Those are item-level research outcomes dependent on a declared sample and actual inspection of the relevant registers.

This example demonstrates how to avoid confusing an inventory with its contents. A catalogue proves the presence and description of a holding; individual facts about the population represented require the records. The archive's coverage may be extensive and still incomplete relative to the city, market, or population. A study that acknowledges those limits can use fire-policy registers responsibly for demographic and economic history.

Case: mortality science and insurance practice

Proposition: early modern scholarship documented mortality observations relevant to later actuarial calculation. The witnesses are the 1662 Graunt edition catalogue and the Royal Society's 1693 Halley manuscript record. Their metadata support the works' bibliographic existence, chronology, and documentary identity. The Equitable Life archive described by the Institute and Faculty of Actuaries identifies institutional material with which the later use of actuarial methods can be investigated.

These sources together support a historical research trajectory from mortality observation and mathematical writing to investigation of an assurance institution. They do not, without further item-level examination, establish the exact adoption date of every formula or premium schedule within the institution. The correct conclusion preserves the distinction among publication of mathematical research, surviving office records, and demonstrated routine operational practice.

Audit conclusions

Across the four cases, the most reliable statements are those whose verbs match the evidence: a text 'provides', a catalogue 'identifies', a scholarly article 'reports', a law 'enacts', and a comparative analysis 'supports an inference'. These distinctions are practically consequential. They allow historical research to advance from modest documentary facts to sophisticated interpretation without claiming access to a manuscript that has not been read or certainty about events that left no surviving record.

Historiographical Synthesis

Insurance history is a history of plural arrangements for managing uncertainty, of particular transactions, of institutionalized markets, and of changing techniques for calculating, distributing, and adjudicating loss. The evidence examined here establishes that these subjects must be investigated at the scale appropriate to their sources. A statute can establish a legal norm; a contract can establish the terms of an exchange; a ledger can establish a recorded transaction; a claims file can establish dispute and settlement practices; a mortality table can establish a calculation and its stated observations; a company archive can support reconstruction of organizational activity. A broad cultural or economic historical claim requires a reasoned comparison of several such witnesses.

The central analytic conclusion is that evidence of an antecedent mechanism and evidence of a later institution carry different historical content. Babylonian provisions demonstrate contingent commercial responsibilities within a translated legal source; medieval Mediterranean documents demonstrate contract forms organized around commercial maritime risk; early modern British statutes and company records demonstrate distinctive systems of adjudication and administration; demographic publications and actuarial archives document the mathematical and institutional sources of life assurance. The cases can be connected through questions about risk and obligation without treating them as interchangeable instances of one timeless policy contract.

The historiography of firsts is especially vulnerable to overstatement because a vivid document and an exact date provide an attractive narrative beginning. The 1343 Genoa–Palermo discussion shows that the date of a prominent surviving deed is different from a verified universal invention. The earliest available written evidence is the end product of a much longer chain of commercial practice, record production, archival custody, scholarly selection, and present accessibility. The most credible origin narratives make that chain visible and use firstness language only at a documented scope.

The strongest quantitative case examined combines a substantial documented medieval contract sample with explicit treatment of archival selection. That research shows how insurance history can advance beyond institutional celebration through the systematic study of routes, participants, underwriting amounts, and risks. The same work demonstrates the limits of quantification when missing original contracts and differences in local recording practice are unmeasurable. Transparent uncertainty is an integral part of the inference, because the reliability of a generalization depends on identifying what the historical data do and do not observe.

Comparison also revises the geographical shape of insurance history. The global commercial history assembled by Borscheid and Haueter (2012) traces the expansion and transformation of modern insurance networks across countries and regions. Its international scope is productive when researchers distinguish global circulation of commercial insurance institutions from the many earlier and contemporaneous practices of mutual support and risk allocation. A history of worldwide insurance must examine both the geographic movement of corporations and the institutional differences visible in the receiving societies, with due attention to archives preserved outside traditional European commercial centers.

Source criticism must also recover the social relations embedded in underwriting. Draper's research on maritime insurance and slavery makes clear that financial protection supported commercial systems whose gains and harms were distributed unequally. An insurance contract is a record of selected financial responsibilities; it can also be a trace of coercion, dispossession, or exclusion that the writer of the document did not intend to preserve as such. Historians can use the same source to reconstruct commercial technique and investigate the human consequences of that technique, provided they read beyond the categories supplied by its creators.

The practical implications are direct. A historical insurance statement belongs to a named source. Its chronology belongs to a dated witness and declared edition. Its legal interpretation belongs to a jurisdiction and period. Its quantitative assertion belongs to a dataset, denominator, and observation process. Its causal explanation belongs to evidence of mechanisms and alternatives. Its apparent global priority belongs to an explicitly defined archival search universe. These requirements create a coherent, inspectable research practice and allow future scholars to extend, correct, and re-evaluate the narrative as new archives, editions, and interpretations become available.

The original contribution of this dossier is an integrated, source-specific application of these established standards of historiography, legal interpretation, archival description, and economic history to the problem of insurance origins and institutional development. Its comparative findings are independently articulated here, but no claim is made to have established a new historical theory or discovered a previously unreported primary instrument. Scientific strength rests in the correspondence between evidence, argument, and conclusion.

Conclusion

The historical study of insurance begins with an explicit research object and continues through the identification of evidence capable of answering questions about that object. Its foundational sources include commercial deeds, legal provisions, contracts, account books, policies, company registers, judicial records, demographic publications, mathematical tables, correspondence, and administrative archives. Each originates within a practice and preserves particular facts while omitting others. Accurate interpretation depends on reconstructing provenance, transaction context, legal jurisdiction, translation, chronology, and the mechanism through which a document survived.

The documentary cases demonstrate five linked conclusions. First, historical risk allocation and premium-based insurance may be connected through commercial history while demanding different criteria of evidence. Second, exact early dates establish the chronology of identified witnesses and must be distinguished from claims of universal invention. Third, cross-jurisdictional comparison is reliable only when legal forms, languages, units, and archival selection are explicitly assessed. Fourth, historical quantitative analysis requires the preservation of original observations, transparent coding rules, and defensible limits on generalization. Fifth, insurance histories must include the political and social systems through which risk protection was organized, including the unequal relations that appear in archival records of empire and enslavement.

The resulting research practice provides clear and reproducible grounds for making positive historical claims. The modern historian can show what an ancient legal rule required, what a medieval maritime deed arranged, how a company registered policies, and how a mathematical author organized mortality evidence. Broader accounts of institutional change become credible through dated comparisons, convergent sources, explicit inference, and acknowledgment of unobserved populations. The quality of insurance historiography is ultimately measured by the traceability of its propositions to the documents and procedures that can sustain them.

The verified materials support the analytical conclusions stated in this monograph within the access limits disclosed in the source record. The online examinations performed for this edition covered published texts, archival catalogues, public legislation, institutional source descriptions, and accessible peer-reviewed research. Where underlying original manuscripts or restricted archival series were unavailable for independent transcription, the conclusions remain attributed to the published scholarship or documentary catalogue rather than misrepresented as first-hand discoveries. The scientific editorial process for the present edition documents that boundary as part of its substantive completion.

References