The gold standard of history of archives is the ability to explain how records reached us, why some survived, and how decisions about custody, selection, description and access shaped the evidence we can use. An archive is not a complete recording of the past. Its contents have histories of their own: people made records for particular purposes, kept or discarded them, moved them, organised them and sometimes made new copies. Understanding those histories makes a historical conclusion more dependable.
This guide follows selected developments in France, Britain, the United States and Singapore, then uses original worked investigations to show what those developments mean for readers. These cases are not a universal sequence through which every society passed. They offer different ways to investigate a common question: how did recorded activity become evidence available to later people? All River District examples below are invented teaching cases, not accounts of a real archive or disaster.
Use the Gold Standard master for the wider subject map, the Education parent for learning that transfers beyond a familiar exercise, and How X Works for connected explanations. Readers who want practical instruction in handling archival evidence can continue to Teaching Archival Studies: Provenance, Authenticity and Access. Here the central subject is historical change in archives and its consequences for knowledge.
Choose your archives history route
Chapters 1–3: Understand origins and public purposes
Chapters 4–6: Trace institutions and records
Chapters 7–9: Investigate arrangement and absence
Chapters 10–12: Test access and preservation
Chapter index
Understand origins and public purposes · Chapters 1–3
1. Begin with two histories, not one date
Trace institutions and records · Chapters 4–6
4. Centralisation required custody, staff and finding tools
Investigate arrangement and absence · Chapters 7–9
7. Arrangement preserves relationships and creates routes
Test access and preservation · Chapters 10–12
10. Access has several histories at once
11. Preservation depends on work that can be overlooked
12. Digital archives preserve relationships as well as files
Transfer and investigate · Chapters 13–14
CHAPTER 1 OF 14 · Understand origins and public purposes
1. Begin with two histories, not one date
An archival document belongs to at least two histories. The first concerns the activity that generated it: a payment, meeting, application, survey or personal correspondence. The second concerns its later treatment: storage, transfer, appraisal, arrangement, cataloguing, copying and consultation. Neither history can substitute for the other. A well-preserved document may contain a mistaken account; an accurate account may survive in a damaged or poorly identified copy.
Imagine a River District relief application written in 1924, transferred to a regional repository in 1972, photographed in 1998 and placed online in 2016. The four dates answer different questions. The application is evidence of what its author submitted in 1924. The transfer date helps explain custody. The photograph records a reproduction event. The online date explains when a particular audience could discover that reproduction. Calling it a “2016 record of relief” without qualification confuses creation with publication.
A chronology that keeps explanations honest
Write separate entries for the event described, record creation, later annotation, transfer and reproduction. If an entry is unknown, preserve that uncertainty. A catalogue’s broad date range may describe an entire file rather than every sheet. A later note attached to an older document should not silently inherit the older date.
The history of archives asks why these stages happened as they did. Who had authority to collect? What kinds of records counted as worth keeping? Who could consult them, under which conditions? Which languages and descriptions made them findable? Answers can change even when the physical document remains the same.
This approach also prevents a misleading founding story. Establishing an archive does not establish the first moment at which a society kept records. Nor does the age of its oldest holding establish the age of the institution. A newly founded repository can receive much older material. Keep the history of the records, the history of the repository and the history of public access on separate timelines before examining their connections.
CHAPTER 2 OF 14 · Understand origins and public purposes
2. Records begin in activities and relationships
Before a record becomes a historical source, someone usually has a reason to create or receive it. A list can allocate supplies; a letter can seek permission; a ledger can track obligations. These purposes shape the fields recorded and the people visible. The form is part of the evidence. A document that asks for the applicant’s name, address and losses gives different possibilities from a register that records only an application number and final payment.
In our fictional relief office, a clerk records authorised payments but not unsuccessful enquiries made at the counter. A historian finds 400 payment entries. It would be defensible to investigate the payments documented in that ledger. It would not yet be defensible to conclude that exactly 400 households sought assistance, or that everybody affected received help. The office’s transaction created the unit of recording.
Work backwards from the question
Suppose the question is whether distant households faced greater difficulty obtaining relief. A payment ledger alone cannot settle it. The researcher would want applications, decisions, correspondence, locations and evidence about people who never entered the application process. Not all of these records will exist. Identifying the missing evidence still improves the question: the surviving ledger may show geographical patterns among paid cases, with clear limits on what those patterns mean.
For an archive’s historian, the next step is to investigate changes in the office itself. Did it introduce a standard form? Did responsibility move between departments? Did a new reporting requirement begin? A sudden increase in surviving entries may reflect a new recording procedure rather than a sudden increase in activity. Establishing a baseline requires understanding the record-producing process.
This is why an institutional history is useful alongside a catalogue. Names, responsibilities and boundaries help identify likely creators. They also show why one subject may appear in several collections. A flood can generate engineering plans, health reports, insurance correspondence and residents’ letters. Subject similarity does not make these records one original administrative body, and separate origins do not make them unrelated evidence.
CHAPTER 3 OF 14 · Understand origins and public purposes
3. Revolution and the public purpose of an archive
The French Archives nationales dates its establishment to 12 September 1790, during the Revolution. Its institutional history identifies the law of 25 June 1794 as a clarification of its mission and describes centralisation, free citizen access and a national archival network as major principles. It also records early acquisition of papers from abolished central institutions, ecclesiastical establishments in Paris and seized private archives. These are distinct routes into a repository, not one uniform act of voluntary donation. See the institution’s historical account.
Those facts open questions rather than finishing an interpretation. A change of political authority can change the meaning of records already in existence. Papers created to support an earlier administration may become evidence for a successor, material for historical inquiry, or both. Keeping a document can preserve information while transferring control over who describes and uses it.
Separate a declared principle from its operation
A historian should distinguish an access principle in a law from the experiences of particular readers. To investigate actual access, look for regulations, requests, opening arrangements, fees where relevant, correspondence and evidence about users. A statement of public purpose is evidence of a stated aim. Demonstrating its reach requires further evidence about implementation.
Consider an invented archive whose founding charter promises public consultation. Its first reading room has space for six readers, and requests take several weeks to process. The charter and the capacity evidence are compatible. They concern different aspects of access. It would be misleading either to dismiss the charter as meaningless or to infer that everybody could immediately consult every document.
The French example therefore helps frame a historical problem: how did archives become associated with public authority and public claims on knowledge? It does not prove that all archival institutions had the same origins or that access expanded continuously. Comparisons become stronger when they specify the institution, period, category of records and kind of access being compared.
CHAPTER 4 OF 14 · Trace institutions and records
4. Centralisation required custody, staff and finding tools
Britain’s Public Record Office Act of 1838 addressed dispersed records and provided for their supervision under the Master of the Rolls. Its provisions connected preservation with practical organisation: transfers involved descriptions and receipts, while arrangements were made for staff, accommodation, repair, catalogues and indexes. Section IX also dealt with admission and fees. The original Act is evidence of an institutional framework, not a description of how every later reader was treated.
Centralisation is therefore more than putting boxes in one building. It can mean bringing dispersed responsibilities under a common authority, documenting movements and establishing consistent procedures. Physical concentration, legal custody and intellectual organisation need not happen on the same day. Treating them as a single event hides the work that made an archive usable.
Worked case: reconciling a transfer
Assume a fictional office’s dispatch list records 120 boxes sent to a repository. The receiving register records 118 boxes accepted on the delivery date, followed by two received a week later. A researcher consulting only the first receipt might report two boxes lost. Reading the second receipt resolves the numerical discrepancy: 118 + 2 = 120 boxes received.
That answer is necessary but limited. Box counts do not establish that every expected file is present. Boxes might have been repacked, and a box can contain more or fewer files than another. A stronger reconciliation compares identifiers and contents against the transfer documentation. If the lists use different numbering systems, a documented correspondence between old and new numbers is needed.
For institutional history, the receipts reveal how responsibility was handed over and recorded. They may also reveal a gradual transfer that a later anniversary account compresses into one date. The lesson is not to distrust commemorative histories automatically. Use them to locate milestones, then choose administrative evidence that can answer a more precise question about implementation.
CHAPTER 5 OF 14 · Trace institutions and records
5. A national archive can be younger than its records
The US National Archives’ own history states that legislation established it in 1934, that staff and Works Progress Administration workers began surveying federal records in 1935, and that the first federal records were transferred to its new building in 1936. The sequence distinguishes authorisation, investigation and actual transfer. The same official history describes earlier federal storage in varied, sometimes unsuitable locations.
This offers a clear correction to the idea that an archive’s founding date marks the beginning of its country’s documentary record. It also shows why surveying matters. Before records can be systematically transferred, somebody must establish what exists, where it is, who controls it and what arrangements are possible. A founding decision creates obligations that require people and resources to carry out.
Build a milestone account with verbs
Instead of writing “the archive began in year X” and stopping, record what happened: legislation authorised an institution; staff surveyed records; a building received transfers; a catalogue described a series; a reading service began. Each verb suggests a different kind of supporting source. A law supports an authorisation claim; a receipt supports a particular transfer; a reader register supports recorded use.
In a fictional regional archive, an annual report announces establishment in 1950, staff appointments begin in 1951, and the first public reading room opens in 1953. None of these dates must be wrong. The error occurs when one is used to answer a question about another. A good historical paragraph might explain the three stages and then investigate why the gap occurred.
Avoid filling that gap with an appealing explanation before finding evidence. Funding, recruitment, buildings or unresolved custody could each matter, but plausibility is not proof. State the interval first, identify possible explanations as hypotheses, and seek the records most likely to distinguish them. Archives history rewards this discipline because its most important changes often involve slow administrative work between memorable public milestones.
CHAPTER 6 OF 14 · Trace institutions and records
6. Singapore: institutional records and recorded memories
An NLB BiblioAsia account explains that the National Archives and Records Centre began in 1968 and held records of national or historical significance, some dating back to 1800. It records the establishment of an Oral History Unit in 1979 and later digitisation of photographs, maps, audiovisual material and interviews for Archives Online. These milestones show different additions to archival work rather than the replacement of one medium by another. See Looking Back, Looking Forward, published in 2019.
The historical significance of an oral-history programme is that an archive can help create a record through interviewing, as well as receive records already produced elsewhere. The questions asked, people approached, recording conditions and later description become part of the source’s history. An interview is evidence of a person’s recorded account at a particular time; it is not an unmediated recording of the earlier event remembered.
Three dates in one interview
Imagine an interview recorded in 1990 about a family’s evacuation in 1955, with a transcript completed in 1992. The event date is 1955, the recording date is 1990 and the transcription date is 1992. A historian investigating the evacuation would compare the account with relevant contemporary evidence. A historian investigating memory in 1990 could also study how the speaker described the earlier experience. The same source supports different questions with different limitations.
Do not convert the 2019 article’s service figures or descriptions into claims about current access arrangements. A dated institutional account is useful evidence of how an institution presented its activities at that time. Current services require current confirmation; historical analysis requires preserving the date and context of the statement.
For a local investigation, begin with a narrow question about a place, service or event, then identify the record types that could answer it. A map, photograph, administrative file and interview may illuminate different aspects. Their agreement can strengthen a claim, but apparent agreement must be checked for copying or a shared underlying source.
CHAPTER 7 OF 14 · Investigate arrangement and absence
7. Arrangement preserves relationships and creates routes
Provenance concerns origin: who created or accumulated records through their activities. Original order concerns the arrangement used by the creator. NARA’s archival terminology guide distinguishes these ideas, while its record group explanation shows one institutional application. A record group number is not a ranking of historical importance or a simple date label.
In our River District case, an engineering department keeps drainage plans, while a relief committee keeps household applications. Sorting both bodies entirely by street name might make one geographical search convenient, but it could obscure which organisation produced each record and how files related to its work. An archive can provide a street index without physically or intellectually erasing those original relationships.
A catalogue is historical evidence too
Suppose a 1960 inventory lists “Relief Committee correspondence, files 1–60,” while a 2010 catalogue gives separate descriptions for applications, payments and appeals. The later catalogue has more entries, but that does not by itself mean the archive acquired more records. Description may have become more detailed. Compare the scope, identifiers and arrangement notes before treating catalogue growth as collection growth.
Original order is also an investigative question when records arrive disturbed. A historian should not assume that the order seen today is necessarily the creator’s order. Arrangement notes can explain whether an order was retained, reconstructed or newly imposed. An undocumented alphabetical sequence cannot be treated as proof that the original office filed alphabetically.
This makes archival description part of institutional history. Its terms and structure can reveal changing expectations about researchers, administrative knowledge and what needs explanation. Yet a catalogue is a guide to records, not a substitute for reading them. A title can help locate a file without establishing the truth of every assertion inside it. For practical exercises in these distinctions, use the existing Archival Studies guide alongside a carefully selected catalogue entry.
CHAPTER 8 OF 14 · Investigate arrangement and absence
8. Selection and loss change the denominator
Archives do not generally receive every record ever created. Appraisal decisions, retention arrangements, accidents, deliberate destruction, private choices and failures to transfer can produce different kinds of absence. Their explanations are not interchangeable. NARA’s 2007 appraisal policy describes appraisal as informed judgement about value and disposition, and distinguishes temporary from permanent federal records within its stated scope. It is an example of a documented institutional decision framework, not a rule for all countries and collections.
To study selection historically, find the relevant instructions, schedules, correspondence or appraisal reports, then examine what actually happened. A policy can establish what was intended. A disposal register or transfer list may help establish an action. Neither alone proves that every record was handled exactly as prescribed.
Worked calculation: the surviving applications
Assume a contemporary River District register lists 1,000 applications, of which 300 received approval. Later, only 200 application files survive, including 100 approved cases. The approval rate among surviving files is 100 ÷ 200 = 50%. The approval rate documented by the complete register is 300 ÷ 1,000 = 30%. The first percentage is not a valid substitute for the second.
The discrepancy tells us that the surviving files have a different composition. It does not alone explain why. Perhaps approved cases were deliberately retained more often; perhaps a particular storage location suffered damage; perhaps the surviving set covers different years. Investigate those possibilities rather than inventing a selection motive from the percentages.
The complete register in this exercise is an explicit assumption. In a real investigation, its completeness and interpretation would need checking. “Approval” may have changed meaning, and duplicate applications may not represent separate households. Accurate arithmetic cannot repair an incorrectly defined unit.
A responsible conclusion states both results and their populations: half of the surviving application files concern approvals, while the register records approvals for three in ten listed applications. The historical question then becomes how the surviving set was formed. That is an archival history question with direct consequences for social history.
CHAPTER 9 OF 14 · Investigate arrangement and absence
9. Missing voices require diagnosis, not instant certainty
A silence can enter at creation, preservation, description or access. Someone may never have been asked to give an account. A record may have existed but not survived. It may survive in an unexpected series. It may be described at a broad level without a person’s name. It may be restricted or not yet available online. A failed keyword search does not distinguish these explanations.
For the River District project, suppose no search result names a temporary worker known from a family letter. Start by checking spelling variants, the relevant office, date range and geographical terminology. Then inspect the description of likely series and ask whether names are indexed at all. This is a finite investigation with recorded steps, not an instruction to search indefinitely until a preferred answer appears.
Separate absence from an explanation of absence
Write “no matching entry was found in the catalogue searches recorded here” when that is what the evidence supports. Write “the series is documented as destroyed” only when supporting evidence establishes that fact. Write “this office did not normally record that category of enquiry” only when the procedure is sufficiently understood. These statements have different evidential strengths.
The history of marginalised or less institutionally visible people may require sources beyond official administrative records. Personal papers, community collections, interviews and material evidence can offer additional routes. They have their own selection processes and limitations. Replacing an official source with a personal one does not remove the need to ask who created it, when and for what purpose.
Photographs require the same care. A repository may hold a selected print, a negative, an album or a later reproduction; captions can be added at different stages. The existing guide to reading old photographs without losing context develops those questions. Use it when a compelling image seems to settle an argument too quickly. The image can be strong evidence of something visible while remaining weak evidence of how typical that scene was.
CHAPTER 10 OF 14 · Test access and preservation
10. Access has several histories at once
Access includes permission, discoverability, physical or digital availability, and the ability to understand what is found. These dimensions can change separately. A record may be described but unavailable for immediate consultation. A readable digital image may lack a useful description. A publicly accessible document may use a script or language that requires specialist knowledge.
An archive’s history should therefore specify what changed. Did a restriction end? Did a catalogue become searchable? Did consultation move closer to readers? Did a transcription make handwriting easier to use? Saying simply that access “improved” can conceal which people benefited and which obstacles remained.
Worked case: requests and successful consultation
Imagine that a reading service receives 80 requests in one year and supplies material for 60. The next year it receives 120 requests and supplies material for 84. Completed requests increase by 24, or 40% relative to 60. But the completion proportion falls from 60 ÷ 80 = 75% to 84 ÷ 120 = 70%.
Both statements are true under the assumptions. More requests were completed, while a smaller share of requests was completed. Neither result establishes the cause. The later requests might concern more fragile, restricted or poorly described material. The service may face a backlog. The units might also differ if one request covers one file in one year and several files in another.
For a historical comparison, seek definitions, reporting practices and reasons for incomplete requests. Do not treat the percentage as a verdict on staff performance without that context. Statistics become useful when they sharpen the next question.
Access restrictions also have histories, but this guide does not substitute for current institutional rules or permissions. Historical openness, present availability and permission to reproduce are separate claims. In research notes, record the access conditions encountered and the consultation date. If a later researcher has a different experience, those details help explain the difference without assuming that either account is false.
CHAPTER 11 OF 14 · Test access and preservation
11. Preservation depends on work that can be overlooked
The history of an archive includes the people who identify, package, describe, retrieve, conserve and reproduce records. A narrative built only from directors, laws and new buildings can miss the activities that determine whether evidence remains usable. Budgets, staffing records, work reports and procedural instructions can help investigate that labour, although each source reflects its own reporting purpose.
Preservation is not the same as making every item immediately convenient to use. A fragile document may require support, treatment or a suitable surrogate before consultation. A reproduction can reduce handling while introducing new interpretive questions: were both sides captured, are annotations legible, and has the scale or colour changed? The original and the surrogate may answer different questions.
A small quality-control investigation
Suppose a digitisation report states that 600 sheets were scanned. A sample inspection finds writing on both sides of some sheets, but the project produced only 600 images. That count alone cannot establish complete capture. If 150 sheets were double-sided and 450 single-sided, the number of sides bearing content would be 150 × 2 + 450 = 750. A 600-image total would require explanation.
The exercise assumes one image per content-bearing side and no multi-page images. Change that assumption and the comparison must change too. Inspect the capture method and manifest before declaring that 150 images are missing. The correct diagnosis begins with the unit: sheets, sides, images and files are not necessarily equivalent.
An archive’s annual report may understandably celebrate a large digitisation output. The historian can acknowledge that work while asking how output was counted and which parts of the collection were prioritised. Quantity, fidelity and discoverability are related but distinct measures.
Related institutions face overlapping challenges. The History of Libraries and History of Museums offer useful comparisons in access, description and public purpose. Their boundaries overlap in practice: a library or museum can hold archives. Institutional labels alone do not reveal how a particular collection was formed or managed.
CHAPTER 12 OF 14 · Test access and preservation
12. Digital archives preserve relationships as well as files
Digital records alter the practical problem of preservation. A file can be copied without moving a unique physical sheet, but its meaning may depend on software, metadata, links and surrounding records. An email attachment separated from its message may lose the explanation of why it was sent. A database export can preserve values while omitting relationships or definitions needed to interpret them.
The Digital Preservation Coalition’s handbook discussion of fixity and checksums explains techniques for detecting changes to digital material. Its preservation guidance also emphasises documenting transformations and losses. These are technical supports for preservation, not a guarantee that every statement inside a file is true.
Worked case: unchanged files, incomplete record
Assume a transfer contains 900 email-message files and 100 separate attachment files. All 1,000 files match trusted pre-transfer checksums after receipt. That supports the conclusion that the checked files have not changed during the comparison interval, subject to the checking method. It does not establish that the transfer contains every expected message or attachment, or that the messages were genuine before the baseline was made.
Now suppose the transfer documentation says 120 attachments were expected. The 100 matching attachments do not resolve the difference. The recipient needs to investigate the manifest, naming conventions, embedded attachments and any exclusions. A complete count and unchanged bytes answer different questions. If some attachments were deliberately excluded, the reason and effect should be documented.
For archives history, the important shift is that custody can involve managed copies, migrations and systems rather than only movement between physical stores. A historian studying that shift should examine preservation decisions, specifications and verification records. The presence of an online file does not reveal the full preservation process behind it.
Do not describe digitisation as the end of archival history. It creates further historical layers: selection for capture, technical conversion, metadata creation, interface design and subsequent maintenance. Future researchers may need to reconstruct those layers just as present researchers reconstruct the movements and rearrangements of paper collections.
CHAPTER 13 OF 14 · Transfer and investigate
13. Transfer check: a merger changes the evidence
Try this changed-context exercise before reading the answer. A fictional River District archive merges with a neighbouring repository. Its public report says catalogue entries rose from 10,000 to 16,000. A researcher argues that the archive now holds 60% more historical documents and that local history is therefore 60% better represented. What must be checked before accepting either conclusion?
The numerical increase in entries is 6,000 ÷ 10,000 × 100 = 60%. That is a correct calculation about catalogue entries. It does not establish a 60% increase in documents. Some new entries may describe previously uncatalogued holdings, some may split one old description into several, and some may represent new accessions. Their contribution must be distinguished.
Answer with a reconciliation, not a slogan
Assume the extra 6,000 entries comprise 2,500 imported descriptions for newly transferred material, 2,000 descriptions created by subdividing older entries, and 1,500 descriptions of previously uncatalogued holdings already in custody. These figures account for all 6,000 additions. Only the first category clearly accompanies newly transferred material, and even it counts descriptions rather than documents.
The broader claim about representation requires another investigation. Which creators, periods, places and activities are covered? Were previously underrepresented communities included? Did access become easier? Did duplicate descriptions inflate the total? There is no defensible single percentage improvement without defining and measuring the dimension of representation.
A useful next step is to select a narrow research question before and after the merger and compare the identifiable, consultable evidence available for it. Record the limits of that comparison. One improved route does not establish improvement for all subjects, but it can demonstrate a specific benefit.
This exercise transfers the method because the surface setting has changed. Earlier chapters examined founding laws, physical transfers and digitisation. Here the same habits apply to a merger: define the unit, reconstruct the process, distinguish institutional claims from supporting evidence, and preserve uncertainty where the records do not decide the question. Successful learning means using those habits without needing the original example repeated.
CHAPTER 14 OF 14 · Transfer and investigate
14. Build an investigation and know when to stop
For a first investigation, choose one institution, one period and one change: the creation of a reading service, a transfer programme, an oral-history initiative or a catalogue revision. Start with an institutional history to identify dates and terminology. Then seek one source that records the decision and another that records implementation. Compare their purposes before treating agreement as independent confirmation.
A workable research note has five elements in connected prose: the question, the identified records, what they establish, what they leave uncertain, and the next evidence that could change the conclusion. Include stable identifiers and consultation dates. Preserve the distinction between an observation, an inference and a hypothesis. This makes the work useful to another reader who has not followed your searches.
Is the oldest document the archive’s beginning?
No. It may have been created long before the institution that now holds it. Investigate its creator, custody and acquisition separately from the repository’s foundation. An archive’s history can include predecessor arrangements without assuming uninterrupted custody unless evidence supports it.
Does an official archive make a document true?
No. Custody and authentication can support a judgement about what an item is and how it has been maintained. Its factual assertions still require evaluation. A genuine complaint establishes that a complaint was made; the allegation within it requires further investigation.
Can a missing file prove a cover-up?
Not by itself. Establish that the file existed, identify the expected custody and retention process, and look for evidence about its disappearance. Deliberate suppression is one possible explanation in some cases, but ordinary disposal, damage, misdescription and incomplete transfer are different possibilities. A serious allegation needs evidence that distinguishes them.
Is an online search enough for a school or community project?
It can be enough for a carefully bounded question if the available evidence supports the conclusion and limitations are clear. It is not automatically a comprehensive survey of an archive. Explain which catalogues and materials were searched and avoid claiming universal absence from a limited search.
How do archives differ from libraries and museums?
Ask how a particular body of material was created, acquired, arranged and used, rather than relying only on the institution’s name. Archives commonly preserve relationships among records produced through activities; libraries and museums can also hold such bodies. Compare practices and mandates in a defined case instead of imposing an absolute boundary.
Which sources support this guide?
The French Archives nationales’ institutional history supports the revolutionary milestones discussed in Chapter 3. The 1838 Public Record Office Act supports the statutory arrangements examined in Chapter 4. NARA’s institutional history supports the 1934–1936 sequence in Chapter 5. The dated NLB account supports the Singapore milestones in Chapter 6. NARA’s terminology, record-group and appraisal publications support the bounded explanations of arrangement and selection; the Digital Preservation Coalition supports the technical distinction between preserving files and documenting their integrity. Each is linked beside the relevant discussion.
These sources have different roles. A law states a framework, an institutional history selects milestones, a policy describes a decision process, and a technical handbook explains methods. None alone gives a complete history of readers’ experiences or the records that failed to survive. The River District calculations are original hypothetical exercises with their assumptions stated; they are not statistics drawn from those institutions.
Finish a first project when you can make a limited, supported claim and explain what would be needed to extend it. For example: “The records show that the institution authorised a transfer in one year and received the specified boxes in two deliveries the next year; they do not yet establish the completeness of the files inside.” That is a useful historical finding. The gold standard is the ability to explain both how evidence became available and exactly how far it can take the argument.