Abstract
A manuscript can open perfectly and still be incomplete. A submitted PDF may render the words and pages exactly as expected while failing to explain whether it was the reviewed version, which comments shaped it, whether an image was embedded or linked, what software created its working draft, who may access it, or how it relates to the correspondence that sent it into the world. A readable file is valuable; it is not automatically a literary archive.
This paper offers a significant-properties and package protocol for writers, editors, estates, small presses, and archival stewards. It treats a born-digital manuscript as a selected package of possible evidence: an original working file, an access derivative where useful, a concise record of version and relationships, provenance and processing information, and a rights/access note. The purpose is not to identify one universally safe format. It is to identify which qualities, relations, and access conditions must endure for a defined future use.
The paper works through a constructed manuscript bundle: an editable draft with comments, a submitted PDF, an image or embedded object, an editorial hand-off record, a version note, and a rights/access note. It distinguishes retained originals, access derivatives, migration, and emulation; then asks what an archive should preserve, describe, restrict, or defer. The result is a proportionate instrument. It helps a reader leave behind both false reassurance—“the PDF is enough”—and impossible maximalism—“keep every digital trace forever.”
1. The file that opens but does not explain itself
Imagine a literary executor opening a decade-old PDF. The pages render. The title is clear. The words are present. Yet the file does not explain whether it is the version submitted to an editor, a revised version prepared after review, or an early export whose layout happened to survive. It does not reveal whether a photographed notebook page was embedded at the time of writing or replaced later. It does not tell a researcher why a passage changed, whether comments existed, what rights governed the attached image, or whether the author expected the material to be public.
The future reader has not encountered a corrupted object. The reader has encountered a decontextualized one.
Joan E. Beaudoin describes context as information about a digital object’s creation, composition, purpose, history, curation, authenticity, and preservation operations that can support later retrieval, assessment, management, use, and interpretation.[3] Her point is not that every archive can capture everything. It is that technical survival without contextual information may leave a cultural object difficult to understand. A file’s content can remain intact while its relation to a person, decision, version, system, or other object becomes obscure.
This risk matters especially for literary work because a manuscript is often more than its final reading text. It can carry revision, editing, correspondence, media, formatting, and submission context. Yale University Library’s born-digital description guidance gives a useful example of hybrid literary papers: typewritten and handwritten drafts, electronic Word drafts with notes and Track Changes, correspondence, photographs, storage devices, uncertain dates, and distinct physical or digital locations can all belong to one record of creative work.[7] The example does not create a universal archive checklist. It makes a simple fact visible: a literary manuscript may exist through several linked forms, and those forms do not all carry the same evidence.
A manuscript is not only a file that can be rendered. It is a record whose meaning may depend on the properties and relations that rendering does not show.
This paper therefore begins with a distinction. A working file may preserve editability, comments, styles, hidden structure, links, or an application-specific behavior. A PDF may offer a stable and convenient reading snapshot. Correspondence may explain why a particular version traveled. A rights note may determine whether an archive can make a copy accessible. None is automatically sufficient; none is automatically disposable.
The question is not “Which file format wins?” The question is what must remain available for this bundle to remain meaningful to the people and purposes that matter?
2. The category error of one safe format
File formats matter. They affect what software can open a record, whether a document’s structure or layout survives a transformation, whether technical specifications are available, and how many tools can identify, validate, or render the object. But a format is one part of a larger information environment. The Digital Preservation Coalition notes that files may depend on software, hardware, and broader environments, and that preservation strategies should be practical and linked to a collection’s needs rather than an attempt to support everything.[2]
The Coalition’s discussion of significant properties provides a more useful starting point. Significant properties may be qualities necessary for a user community’s required rendition or meaning: a document’s layout, its editable text, its comments, its relationship to another file, its visual behavior, or a technical characteristic required for valid rendering.[2] The point is not that any one property is always significant. A final reading copy might need legible page layout and embedded image fidelity. A textual scholar may need comments, revision evidence, or original order. A publisher’s estate may need rights and submission records. A curator may need enough information to understand what processing or transformation has occurred.
The property-first approach corrects two opposite errors.
| Error | Why it fails | Property-first correction |
|---|---|---|
| “Convert everything to one archival format.” | Conversion may improve access or reduce technical complexity, but it can also change content, structure, behavior, or metadata. | Identify what must endure first; then assess whether retention, migration, emulation, or an access copy is appropriate. |
| “Keep only the original working file.” | Original bits can retain a rich record while remaining difficult to open or interpret later. | Retain the best available original where feasible, while documenting dependencies and preparing access paths where needed. |
| “The submitted PDF is the archive.” | A PDF can be an excellent access derivative but may not explain revision context, embedded relationships, comments, or rights. | Treat the PDF as one element of a selected package and record what it does and does not carry. |
| “Keep every object forever.” | Digital collections proliferate; rights, access, staff, storage, and future maintenance obligations are real. | Appraise value, uses, restrictions, resources, and risks; document selection and deferral. |
The Smithsonian Institution Archives illustrates the layered logic. Its guidance describes stabilization, redundant copies, fixity checking, format identification, documentation, bit-level retention of the original, migration to a parallel access or preservation format where appropriate, and emulation where needed for more complex objects.[1] It states plainly that bit-level preservation retains a file’s original sequence of bits but does not guarantee future accessibility. The preserved original may be valuable precisely because a later migration, emulation, or interpretation effort needs it; it is not automatically a solution for every user.
This is why a format conversation must begin with properties and intended use. The same editable document might be handled differently in three contexts. A writer may need an export and a short context note for continuity. A small press may need the final production files, image permissions, and edition history. A literary archive may need original files, an access derivative, technical and descriptive metadata, rights information, restrictions, and a record of processing. The question changes with the responsibility.
3. The manuscript bundle as a cultural object
The paper’s anchor is a constructed teaching specimen, not an account of a deposited author’s papers. It consists of six elements.
| Bundle component | What it can carry | Why it may matter later |
|---|---|---|
| Editable draft with comments or Track Changes | Revision evidence, editorial interventions, styles, structure, and possibly embedded content. | A future reader may need to distinguish finished text from collaborative revision process. |
| Submitted PDF | A stable reading snapshot and a particular submission state. | It may show the exact form sent for review or publication. |
| Image, media file, or embedded object | An asset’s technical identity, relationship to text, and possible rights information. | A rendered page may not make the asset’s source, version, or permission clear. |
| Editorial correspondence or hand-off record | Purpose, recipient, timing, request, authority, or submission context. | It can explain why a particular version exists and how it entered an editorial process. |
| Version/context note | Naming, sequence, creation or modification context, software/format, and intended status. | It prevents a timestamp from doing interpretive work it cannot support alone. |
| Rights/access note | Known copyright, licenses, collaborator conditions, privacy concerns, and restrictions. | It frames what can be preserved, shared, redacted, or made accessible. |
The bundle does not assume that every author creates all six elements. It is a device for seeing different forms of evidence. A single editable file may contain a comment thread but not a permission to share the linked image. A PDF may preserve page geometry but not a revision history. An email may establish that a version was sent for copyediting but not transfer literary rights. A file’s last-modified date may reflect a later copy or migration rather than the date of composition.
Yale’s guidance is useful because it turns these abstract concerns into ordinary description practice. It urges archivists to distinguish creation dates from misleading machine-generated dates, to describe original order or interventions, to record formats and software when relevant, to state access-copy practice, to document processing actions such as normalization, migration, redaction, extraction, and renaming, and to record significant custody transfers.[7] Its purpose is institutional description, not writer instruction. Nevertheless, it demonstrates the kinds of questions that can make a digital manuscript legible after its immediate working context has passed.
Washington University in St. Louis’s Born-Digital Poetry project provides a second, current literary-archive context. The project addresses acquisition, discoverability, preservation, and use of born-digital poetry and joins digital-content processing with poet-driven curation tools, including a creator-oriented preservation guide.[8] This is significant not because it proves a preferred workflow, but because it rejects a false choice between individual creative practice and institutional archiving. A poet can make useful choices about what matters without being expected to build a repository.
4. Three preservation moves, not one answer
The manuscript bundle invites three different moves. They may coexist, and none should be treated as automatically superior.
4.1 Retain the original where it carries significant evidence
The original or near-original working file can preserve properties lost in a derivative: revision marks, editability, document structure, embedded objects, software-specific behavior, or original organization. Smithsonian guidance describes retaining bit-level-preserved original files even where migration is also undertaken, because the original may support later work or preserve elements that a current access path cannot.[1]
Retaining an original does not mean treating it as self-interpreting. At a minimum, future users may need to know the file’s name, format, approximate creation or modification context, associated software where known, relation to other files, and any processing performed. The DPC notes that documentation and technical information can be necessary for a preservation strategy, while Yale’s examples show how original order, dates, and transfer information can affect interpretation.[2] [7]
4.2 Create an access derivative where use requires it
An access derivative is a copy prepared because a particular audience needs a practical way to read, view, or share material. A submitted PDF may be a strong access derivative for page-based reading. A redacted copy may permit a reviewer or researcher to see relevant material without exposing private information. A normalized text file may support search or computational use. The derivative can be valuable without being equivalent to the original.
That difference should be recorded. The user of an access derivative should not have to guess whether it contains original comments, original order, linked media, or later edits. Yale’s guidance explicitly anticipates that repositories may provide access copies rather than original computer files and recommends documenting technical requirements, access conditions, and processing actions.[7]
4.3 Consider migration or emulation when an important property requires it
Migration transforms a file into another format to support future usability. It can be necessary, but it can also create subtle changes. The DPC recommends identifying the properties that a transformed object must retain and applying quality assurance rather than assuming conversion preserves everything.[2] The Smithsonian similarly describes migration as appropriate when significant properties can be retained in a preservation-quality form, while preserving the original in parallel.[1]
Emulation takes a different route. When a file’s meaning depends on its original environment, a later system can approximate that environment rather than transforming the object into a newer form. Smithsonian describes this approach for complex packages that rely on particular hardware, operating systems, or software.[1] It is not a default for ordinary text documents. It is an option when functionality or rendering context matters enough to justify the additional work.
| Move | Primary value | Main risk if used alone | Record to retain |
|---|---|---|---|
| Original retention | Preserves rich or unknown properties for future interpretation. | The object may remain inaccessible or unintelligible. | File inventory, fixity at archive scale, format/software notes, original-order and custody context. |
| Access derivative | Makes reading, sharing, or restricted access practical. | May flatten or omit editable, relational, or process evidence. | Derivative label, creation date, transformation method, known limits. |
| Migration | Reduces dependence on an endangered or inconvenient format. | Can introduce invisible loss or altered behavior. | Source and target format, tool/process, acceptance criteria, quality check. |
| Emulation | Preserves functional or environmental behavior. | Requires specialized infrastructure and may not scale. | Environment description, dependencies, access instructions, scope of fidelity. |
5. The Significant-Properties and Package Protocol
The following protocol is the paper’s teaching instrument. It is designed for proportionate decisions. A writer may complete a brief version before a major hand-off. A small press or estate may use it when organizing production records. An archive may connect its answers to fuller acquisition, transfer, rights, and processing procedures.
| Step | Question | What to identify | Proportionate action |
|---|---|---|---|
| 1 | Who may need this record, and for what purpose? | A writer’s continuity need, an editor’s review need, an estate’s rights need, a scholar’s process question, or an archive’s long-term mandate. | Name a use horizon instead of preserving abstractly. |
| 2 | Which properties and relations are significant? | Text, layout, comments, editability, linked media, version identity, hand-off context, rights or restriction terms. | Select a limited list with a reason for each property. |
| 3 | Which object carries each property best? | Original file, PDF, media asset, correspondence, context note, or rights record. | Retain the best available source rather than duplicating indiscriminately. |
| 4 | Which access copy is useful, and what does it omit? | A rendered PDF, redacted copy, normalized text, or emulated environment. | Label it as an access derivative and state meaningful limits. |
| 5 | What explains the bundle? | File and folder names, version sequence, dates, original order, software, transfer, processing, and relationships. | Keep a concise context/provenance note or manifest. |
| 6 | What may be disclosed, restricted, redacted, or transferred? | Known rights, permissions, collaborator material, privacy, donor expectations, and jurisdictional uncertainty. | State known conditions and route unresolved questions to the appropriate authority. |
| 7 | Is the package proportionate to its responsibilities? | Value, staffing, storage, rights work, access obligations, and future review point. | Select, defer, transfer, or document a lower preservation level rather than promise everything. |
The protocol transforms format choice into a subordinate decision. A format may be preferred because it supports a chosen property, not because it has been declared universally safe. A PDF might be indispensable as a final layout record. An editable file may be indispensable because comments matter. An image asset may matter because it has rights, a higher-resolution original, or a relation to a text that a rendering cannot explain. Correspondence may matter only long enough to establish a version and then be restricted, appraised, or excluded.
The protocol also clarifies the difference between description and exhaustion. Beaudoin argues that context can be multidimensional and valuable, but her work does not require that a creator or archive document every fact about an object.[3] DPC appraisal guidance is more direct: digital resources proliferate, decisions about selection and preservation may need to be made together, and organizations must weigh value against technical, legal, access, staffing, and funding obligations.[5] A package is not a hoard. It is a documented choice about what will support a defined future use.
6. Context, provenance, and processing memory
A bundle becomes misleading when its processing history disappears. If a file was moved from a device, extracted from a container, virus checked, renamed, normalized, redacted, migrated, or converted into an access copy, those actions may shape future interpretation. A change may be necessary and responsible; it should not become invisible.
Yale’s guidance treats processing notes as part of a collection’s interpretive record. It recommends recording actions such as migration, normalization, redaction, extraction, filename changes, technical limitations, and custody transfer when they affect how a researcher understands the material.[7] DPC appraisal guidance similarly identifies manifests, file-format lists, checksums, rights information, and documentation of processing as part of transfer and validation workflows.[5]
The archive-scale form of this documentation can be extensive. A creator’s form can be small. A single contextual note might identify the intended state of a manuscript, explain the relationship among a working file, a submitted PDF, and review correspondence, identify a relevant software or format, and note known rights or restrictions. The point is not bureaucratic decoration. It is to prevent a future reader from inventing a chronology or attribution that the record cannot support.
This is particularly important when a digital bundle crosses a boundary. A writer transfers it to an editor. An editor transfers it to a small press. A literary executor transfers it to an archive. At each boundary, the relation among files, permissions, and purposes can change. DPC guidance discusses producer-archive transfer as a process involving negotiation, formal definition, transfer, and validation; it stresses that agreements and documentation help make responsibilities intelligible.[5] Again, a writer does not need to imitate a formal standard. But the underlying lesson is portable: a file transfer without a shared account of what is being transferred can lose both technical and social context.
7. Appraisal, rights, and selective access
The most difficult part of a literary-archive package is often not format selection. It is deciding what may be kept, who may see it, and what responsibilities follow from taking custody.
The DLF’s born-digital access primer emphasizes that digital materials can be reproduced and distributed with unusual ease and that literary papers and email can contain multiple copyright interests. It recommends documenting copyright and access decisions, recording donor permissions and restrictions, and considering the labor required for rights review and item-level metadata.[6] Its guidance is U.S.-specific and explicitly not legal advice. It is useful here as a reminder that rights and access must be treated as documented questions, not as metadata that can be added after a collection has already been promised to the public.
Yale’s practice provides concrete examples of the same principle: restrictions for sensitive material, access copies instead of original computer files, notation of known rights, and processing notes describing redaction or technical limits.[7] The local details should not be imported as a universal policy. They show why access is not a binary switch. An archive may preserve an original while providing a derivative, restrict a set of digital journals while opening other material, or document that access is impossible until suitable hardware or processing is available.
The package protocol therefore asks a different question from “Should this be open?” It asks: Which version can be provided to which person, under what stated terms, while preserving the record’s integrity and respecting the people implicated by it? In many cases, the honest answer will be partial, deferred, or restricted access. That is not failure when the reason is documented and reviewable. Unexplained closure is a different matter.
| Boundary question | Unhelpful default | Proportionate response |
|---|---|---|
| Rights | “The archive has the file, so it can share it.” | Record known rights, permissions, licenses, and unanswered questions before making access promises. |
| Privacy | “More process data is always better evidence.” | Consider collaborators, personal data, sensitive drafts, and redaction or restriction conditions. |
| Access | “If a file is preserved, it must be online.” | Separate preservation, reading-room access, restricted access, and public dissemination. |
| Selection | “Every trace must be retained.” | Define value and use; document what was selected, excluded, or deferred. |
| Cost | “Storage is the only expense.” | Include appraisal, rights review, technical work, description, access, and ongoing maintenance. |
8. A proportionate hand-off
The literary-archive package has three scales.
At the writer scale, the aim may be continuity: retain the working draft and a useful access copy; name the relationship between them; identify essential media; note the status of a submitted version; and avoid discarding rights or privacy context accidentally. The writer does not need to create institutional metadata, calculate checksums, or commit to public access.
At the small-press or estate scale, the aim may be stewardship across people and time: distinguish editorial files from final production files; retain a clear edition and submission history; keep permissions or rights records; identify where access copies differ from original files; and set a review point for future transfer. The organization should avoid promising indefinite preservation without knowing what it can maintain.
At the archive scale, the aim expands to appraisal, transfer, validation, preservation planning, description, access, restrictions, and ongoing resource commitments. DPC guidance makes clear that acquisition decisions should consider rights, access, documentation, formats, technical capacity, staff time, and future costs.[5] At this scale, a manifest, format inventory, fixity information, processing notes, and rights documentation may be appropriate. But the archive’s fuller responsibility does not invalidate the smaller package; it gives the smaller package somewhere coherent to go.
Washington University’s poet-driven curation orientation is a useful model of the dialogue between these scales. The Born-Digital Poetry project links preservation and processing work with tools intended to help poets safeguard their creative output.[8] The lesson is not that an individual should adopt an institutional workflow. It is that good archival transfer begins earlier when creators can name what matters and institutions can explain what they can receive, preserve, restrict, and provide.
The future of a manuscript is not secured by choosing one file type. It is secured by making proportionate, documented decisions about the properties, relations, permissions, and responsibilities that give the file its meaning.
This paper does not endorse a universal archival format, predict future software support, provide legal advice, require public access, or substitute a worksheet for professional appraisal. It asks a narrower question: when a manuscript becomes a record worth carrying forward, what must travel with it for its future use to remain intelligible and responsible? The Significant-Properties and Package Protocol provides a way to answer without reducing preservation to a file extension—or expanding it into an impossible demand to keep everything.
Limitations
This paper is a conceptual teaching manuscript, not a format benchmark, repository certification, or legal opinion. Its manuscript bundle is constructed to make documented preservation distinctions visible; it is not a real author collection or a claim about a named application. The DPC sources are professional guidance rather than a one-size-fits-all rule set. The DLF primer is U.S.-specific and expressly not legal advice. Yale’s description guidance reflects local institutional practice, and the WashU source is a current project description rather than a completed evaluation. Institutional preservation workflows can exceed the resources of writers, estates, and small presses; this paper draws principles of selection, documentation, and dialogue from them without asking individuals to duplicate an archive. Finally, the protocol cannot eliminate difficult appraisal judgments. It can only make their terms visible.
References
[1] Smithsonian Institution Archives, “Preservation Strategies for Born-Digital Materials.”
[2] Digital Preservation Coalition, “File Formats and Standards,” Digital Preservation Handbook.
[3] Joan E. Beaudoin, “Context and Its Role in the Digital Preservation of Cultural Objects,” D-Lib Magazine 18, nos. 11/12 (2012).
[4] National Digital Stewardship Alliance, “Levels of Digital Preservation,” Version 2.1, 2026.
[5] Digital Preservation Coalition, “Acquisition and Appraisal,” Digital Preservation Handbook.
[6] Digital Library Federation, “Legal and Ethical Considerations for Providing Access to Born-Digital Collections: Copyright.”
[7] Yale University Library, “Born Digital Archival Description Guidelines: List of Elements.”
[8] Washington University in St. Louis Libraries, “Born-Digital Poetry: Planning for the Future of Literary Archives.”