Keep an EPUB Contents List Aligned with Its Reading Order

Substantively revised October 21, 2026. Author interpretation, documented cases and hypothetical examples are distinguished below.

A contents link can work perfectly while the book still leads readers through the wrong sequence. The editor taps Practice and reaches the correct chapter. Starting at the beginning and continuing normally skips a newly inserted bridge chapter. The destinations exist; the publication's separate navigation relationships disagree.

The useful production question is therefore broader than whether every link opens. Which resources belong to the book, which sequence defines its default reading order and where do the contents labels lead? Reusable production needs an explicit relationship among those objects, rather than an assumption that alphabetical filenames or a visually tidy contents page determines the whole reading experience.

Direct answer

Inspect the EPUB package manifest, its spine references and the navigation document as separate structures. Map each intended chapter to its manifest identifier and resource, then compare the spine's default reading sequence with the book's editorial plan. Resolve contents links through the actual resource and fragment destinations. For a simple linear book, deliberately align its contents sequence with that intended route, while preserving justified hierarchy and auxiliary content distinctions. EPUB 3.3 and the Reading Systems specification describe these mechanisms; accessibility guidance adds publication-wide navigation considerations. [S1] [S2] [S3]

This article reports a small executed XML relationship exercise on synthetic documents. It does not report EPUBCheck, a packaged EPUB, a reading-system run or an accessibility assessment. The checks expose specific disagreements in a simple book plan; they are not a complete publication validator or a bookstore acceptance result.

Key takeaways

A manifest identifies resources, while spine references establish a default reading order. A filename can be useful to a production team without supplying that order. Do not treat a folder listing as the publication's actual sequential reading plan.

Separate link existence from editorial agreement. A contents entry can resolve to an existing section in the wrong intended sequence. Conversely, a correctly ordered label can point to a missing fragment. Those conditions need different checks and different repairs.

Keep structural checks separate from reading experience. Parsing XML and resolving references can expose important production defects. It cannot establish how a particular reading system presents navigation, how sequential reading behaves there or whether the whole publication meets an accessibility claim.

Definitions and relationships

The manifest is the package's resource inventory. A manifest item has an identifier associated with a resource location and other relevant properties. The spine contains ordered references to such items. The navigation document supplies structured navigation, including the table-of-contents navigation area.

In this article, a destination consists of a resource and, when used, a fragment identifier locating an element within it. The editorial route is the sequence and hierarchy the publisher intends for the actual book. A simple linear fixture is a deliberately narrow example in which each chapter is one primary content document and the contents list names those chapters once in their intended order.

Those terms prevent an overly broad equality rule. A real table of contents can be hierarchical and include several destinations within a single document. Auxiliary material can have a different relationship to primary sequential reading. A check designed for one chapter per document should not be advertised as a universal rule for every valid EPUB.

The problem with repairing only visible contents

An editor inserts a new chapter and updates the contents labels. The new link works, so the immediate review passes. The insertion still may be absent from the primary sequence represented by the package. The book now offers a chapter through one route without placing it where the editor intended during ordinary continuation.

The reverse maintenance error is also possible. The package sequence changes, but the contents hierarchy retains the old arrangement. A reader who browses the contents receives a different map from a reader who begins at the start. The production team can miss the disagreement if each reviewer examines only the route they happen to prefer.

Begin an insertion with an explicit editorial plan. Name the new section, its resource and its intended neighbors. Then identify every representation of that relationship in the actual publication. A maintenance procedure becomes more reliable when it follows the planned change across structures rather than assume that one visible edit reaches all of them.

Root cause: identity, order and destination are collapsed

A resource's presence establishes that the publication includes a named object. It does not by itself establish where the object belongs in a reading sequence. An ordered list of labels establishes a visible map, but its links still need to resolve to the right resources and elements.

The maintenance failure occurs when an author uses one reassuring observation to stand in for all three. The chapter exists, so it must be in order. The label is in order, so its destination must be correct. The link opens, so the book's sequential route must agree. None of those implications follows without the relevant relationship being examined.

The remedy is to keep the production identities stable enough to trace while checking order and destinations separately. A reusable process should be able to show how the intended section maps into each structure. The final result is an explainable publication route, not merely a set of files whose individual names look plausible.

Framework: the Reading Route Map

The Reading Route Map is an author-proposed production aid with four components: Resource, Sequence, Destination and Experience. It is not a W3C feature, a full EPUB validator or an accessibility conformance method. It organizes the evidence needed for a particular navigation change.

Resource identifies the chapter's manifest item and actual content resource. Confirm the relationship rather than rely on a title's resemblance to a filename. The same editorial title can appear in several places, while a misspelled identifier or location can break the actual connection.

Sequence locates the chapter in the intended primary reading route. Name the preceding and following sections when inserting or moving material. Compare that plan with the relevant spine references and account for any deliberate auxiliary material rather than mechanically sort resources by name.

Destination resolves each contents entry to the intended resource and element. Check the label's meaning as well as the address. A link can lead to a real chapter whose topic does not match the promise made by its label, so existence alone is a narrower observation than editorial correctness.

Experience examines the result in the actual intended reading systems and interaction methods. Test contents selection and sequential continuation as separate routes. Include relevant accessibility navigation and context decisions. Static structure can support this review, but it cannot replace observations of the actual reader experience.

Apply the map to the specific changed section and its neighbors, then extend the structural checks to the publication. Keep the actual validation and reading-system results identified by tool and version when they are obtained. A later editor should be able to distinguish a parsed relationship from an observed navigation behavior.

Three documented EPUB mechanisms

EPUB 3.3 distinguishes the manifest resource inventory from the spine's ordered item references. It also recommends ordering contents references to reflect the referenced documents and targets, while providing semantics for primary and auxiliary content. That supports deliberate alignment for a simple book without claiming that every possible contents tree must be a flat duplicate of the spine. [S1]

EPUB Reading Systems 3.3 describes processing the navigation document, including access to its contents links and headings and moving to a publication-resource destination when activated. This defines responsibilities for a reading system. It is not evidence that a particular application has been opened and tested with the publisher's actual book. [S2]

EPUB Accessibility 1.1 explains that publication-wide order matters in addition to individual content documents. Its page-navigation discussion has applicability conditions and requires attention to the source of static pagination when used. A contents page alone is therefore insufficient evidence for a broad accessibility claim, and reflowable screen positions should not be casually relabeled as print pages. [S3]

These are documented format and processing mechanisms. The following research exercise has a smaller scope: it checks a few XML relationships against an explicit simple-book plan. Its passing results must remain attached to that scope rather than become a claim that an ebook was fully packaged or read successfully.

Case study: executed XML exercise with three chapters and an insertion

The research script constructed synthetic package and navigation XML plus three content-resource strings. The chapters were Start, Practice and Finish. Their filenames were deliberately z-start.xhtml, a-practice.xhtml and m-finish.xhtml, so alphabetical file order differed from the intended editorial sequence.

Each chapter had a corresponding manifest identifier and a heading fragment. The primary sequence used Start, then Practice, then Finish. The contents links resolved to those same intended documents and fragments. All fixture resources lived in the same simplified relative directory context, avoiding a claim that the script implemented every EPUB URL-resolution rule.

The first case inspected that three-chapter agreement. The parsed spine resolved to z-start.xhtml, a-practice.xhtml and m-finish.xhtml in the intended sequence. The flat contents list resolved to the same route, and its targeted fragment identifiers existed in the synthetic content strings. That relationship check passed despite the deliberately unhelpful filename sorting order.

The second case kept valid destinations but swapped the first two contents entries. The script reported a difference from the intended simple-book contents sequence. It did not call the targets missing, because the files and fragments still existed. This distinction demonstrates why a link-existence check alone cannot establish the intended editorial route.

The third case kept the Practice resource but changed the contents fragment to a missing identifier. The script detected an absent target fragment. It did not report that the chapter file itself was missing. Those separate findings tell an editor whether to investigate the resource location or the link's internal destination.

The fourth case introduced a synthetic Bridge chapter between Start and Practice. The resource inventory and contents list included it, while the primary sequence still contained only the original three chapters. The check reported a disagreement between the spine route and the intended four-chapter plan. A new working contents link had not completed the insertion.

The fifth case included Bridge in the intended primary sequence as well. The parsed spine and flat contents route both agreed with Start, Bridge, Practice and Finish, and the targets existed. That corrected relationship check passed under the deliberately narrow fixture assumptions.

The script therefore completed five expected-outcome scenarios: two positive cases and three deliberately introduced disagreements that were detected. The result is not five compatible reading systems or five validated EPUB files. It is a reproducible structural exercise whose evidence concerns precisely those XML relationships.

What the check deliberately does not validate

The package string is an inspection fixture, not a complete publication. The script does not supply or validate full metadata, container requirements, styles, media resources or the remaining format rules. It does not create a distributable ebook. Its parser and reference comparisons are not a replacement for EPUBCheck or another appropriate conformance workflow.

The example also does not execute a reading system. It cannot establish how a menu is announced, whether a stored reading position resumes as expected, or how a particular device handles auxiliary content. The Experience component of the Reading Route Map remains a separate task for an actual publication and supported environment.

The flat-order comparison is intentional for these fixtures. A real book can have nested headings, multiple chapter destinations per document and auxiliary material. Derive the actual comparison from that editorial structure rather than apply the script's list equality to every publication and label justified differences as defects.

Action plan: a production insertion procedure

Write the new section's place in the editorial route before exporting. Identify its title, resource identity and intended neighbors. If the section is auxiliary rather than part of primary reading, document that relationship explicitly. A meaningful placement decision gives the structural review an expectation to test.

Update the actual resource inventory, reading-order references and contents destinations needed for the change. Resolve links in their real relative contexts, including relevant fragments. Check for stale references when resources or heading identifiers are renamed. The simple same-directory research case does not establish those more complex relationships automatically.

Run the appropriate complete-publication validation workflow and retain its actual results. Then inspect the intended reading-system routes: choose the changed contents entry, continue from the preceding section and inspect the next section. A working direct jump and a coherent sequential route answer different questions.

Keep the output version attached to the evidence. An assessment of yesterday's package does not approve a later export in which the chapter list changed. A reusable production process should invalidate the relevant checks when a change alters the relationships they examined, rather than retain a generic passed badge detached from the current file.

Failure example: the renamed heading fragment

An editor changes a chapter heading identifier to make the source easier to maintain. The chapter file remains present, the spine remains ordered and the visible contents label still looks correct. Its link continues to name the old fragment. A review that checks only resource filenames misses the changed destination relationship.

This hypothetical failure is analogous to the missing-fragment fixture but is not a report of a real publisher's incident. The prevention method is concrete: resolve the full contents destination against the current resource and inspect the intended target element. A file-existence check has a useful but smaller scope.

If a reading system appears to open the beginning of the chapter despite the stale fragment, investigate that actual behavior without treating it as a repair. The intended subsection destination still needs a deliberate correction and an appropriate check. A forgiving interface should not silently redefine the editor's navigation promise.

Counterargument and tradeoff: not all useful navigation is linear

A reference work may offer several sensible routes through its material. A reader may browse topics, follow an index or jump to an answer key. The counterargument to enforcing one simple contents sequence everywhere is therefore valid: a flat primary route does not describe every useful editorial structure.

The method's answer is deliberate mapping rather than forced uniformity. Identify the primary sequence, structured contents destinations and auxiliary relationships separately. When an alternative route serves the book, preserve it and explain the intended relationship. Do not make the flat research fixture's convenience the definition of a well-designed publication.

The maintenance cost is clarity across representations. Richer navigation can improve a book while adding relationships that need checking when material changes. The editor should choose that complexity because it serves readers, then supply evidence about the actual routes instead of claim that one successful contents tap covers all of them.

Page navigation is a different relationship

A reflowable book's displayed screen positions can change with presentation settings. That makes them a poor basis for invented print-page claims. If the publication needs static page navigation, decide what pagination it represents and examine the relevant accessibility objectives and metadata rather than attach arbitrary page numbers to chapter links. [S3]

Keep the question separate from chapter order. A coherent Start, Bridge, Practice, Finish sequence does not establish a correct page list or a complete accessibility assessment. Conversely, a supplied page list does not prove that the primary reading route matches the editor's intended chapter sequence.

The useful production note names which navigation relationship was examined. A chapter insertion, a fragment rename and a static pagination change may require different checks. Keeping those scopes explicit helps the next editor avoid both missing a dependency and rerunning unrelated checks without a changed condition.

Common mistakes and targeted repairs

Sorting filenames to determine reading order confuses resource organization with the package's actual sequence. Inspect the spine references and their resolved items. A descriptive naming convention can help people work without replacing the format relationship the reader system uses.

Checking only that a contents link reaches an existing file skips its fragment and meaning. Resolve the intended target element and compare the label with the destination. A valid address can still make an editorially misleading promise.

Updating the contents list but not the reading plan leaves the insertion incomplete for a simple book whose primary sequence should include the new section. Compare the intended route with the actual current package. Do not rely on the link's success to establish a different representation's correctness.

Calling a structural script a full EPUB validator overstates its scope. Use appropriate complete-format validation and actual reading-system checks where required. A small relationship checker is valuable when its findings and omissions remain precise, not when its passing result becomes a claim about everything it never examined.

FAQ

Does manifest order determine reading order?

The cited EPUB model distinguishes the resource inventory from the spine's ordered references. Inspect the actual sequence instead of infer it from inventory placement or alphabetical filenames. The research fixture deliberately made those orders differ. [S1]

Must contents and spine always be identical flat lists?

No. A contents hierarchy can contain multiple destinations in a document, and auxiliary material needs deliberate treatment. The exercise uses flat equality only because its simple book plan defines one primary chapter per document.

Can every contents link work while the route is wrong?

Yes. The swapped-contents scenario kept existing files and fragments but disagreed with the intended order. Link validity and editorial sequence are different observations, so both need the relevant expectation.

Is a fragment different from the chapter file?

Yes. A link can name an existing document and an absent target identifier inside it. The executed exercise detected that narrower disagreement separately from a missing resource location.

What happened when Bridge was inserted incompletely?

The synthetic resource inventory and contents included the chapter, but the primary sequence did not. The script detected disagreement with the intended four-chapter route. This was an XML relationship result, not an observed device-navigation failure.

Was a complete EPUB produced?

No. The experiment used synthetic package, navigation and content strings for limited relationship checks. It did not create or certify a distributable ebook or validate the full format's requirements.

Was EPUBCheck run?

No. The script is not EPUBCheck and does not claim its coverage. An actual production package needs the appropriate complete validation workflow with its own identified tool version and results.

Was a reading system tested?

No. The article reports no reader application or device execution. The Reading Systems specification defines relevant responsibilities, while actual implementation behavior requires separate observations with the publication under review. [S2]

What does linear auxiliary content mean here?

The format distinguishes primary and auxiliary relationships through relevant spine semantics. Decide the intended role of the actual material and check the applicable requirements. Do not use an auxiliary designation merely to hide a chapter omitted accidentally from the intended primary route. [S1]

Can I copy the fixture's same-directory resolver everywhere?

No. Its simplified resource context is part of its scope. A real package can have other relative paths and fragments. Resolve those according to the actual publication structure and appropriate format rules.

Does a contents page prove accessibility?

No. Publication-wide structure, content and navigation requirements need an appropriate assessment. EPUB Accessibility discusses the larger work rather than treating individual files or a visible contents list as sufficient evidence. [S3]

Should screen page counts become print labels?

Do not infer static pagination from reflowable presentation positions. Identify the intended page-navigation basis and apply its actual requirements. Chapter navigation and pagination are related reading aids with different evidence needs.

Are the five scenarios a production test suite?

They are a bounded research exercise with two positive cases and three intentional disagreements. They do not establish sufficient coverage for another book, especially one with nested navigation or auxiliary resources.

When should navigation checks be repeated?

When a consequential relationship changes, such as a chapter insertion, resource rename, fragment edit or new package export. Keep the checked output version visible. A previous result should not be presented as approval of changed bytes it never examined.

What is the most useful handoff record?

Record the editorial route, resource identities, relevant spine relationships and contents destinations, then attach actual validation and reader results when available. State what was only structurally inspected. Another editor should be able to distinguish a coherent plan from a fully observed reading experience.

Recommended books

EPUB 3 Best Practices by Matt Garrish and Markus Gylling provides historical production context for the format. The referenced publisher record is the January 2013 book. It is not current EPUB 3.3 authority; the technical claims in this article rely on the cited W3C publications and the bounded structural exercise.

Designing Interfaces, third edition, by Jenifer Tidwell, Charles Brewer and Aynne Valencia, supplies broader context on navigation and wayfinding. Use it to consider whether the available routes make sense to a reader. It does not specify EPUB package rules or prove a reading system behaves correctly.

Don Norman's The Design of Everyday Things, revised and expanded edition, offers background on understandable cues and user expectations. Its relevance is the promise made by a contents label and the reader's ability to continue a task. The Reading Route Map remains the author's proposed production aid.

Further exploration

Inspect one real authorized chapter insertion across all relevant representations. Follow its manifest identity, primary route and full contents destination rather than stop at the first successful click. Record where the production system makes the same editorial decision more than once and how it keeps those representations aligned.

Then examine a justified non-linear route in the actual book. Preserve the reason it exists and choose checks that match it. A reusable EPUB process becomes more dependable when it distinguishes resource presence, editorial sequence, navigation destination and reader experience, with each finding tied to the evidence that actually supports it.

Primary evidence sources

• [S1] EPUB package spine specifies default reading order through manifest item references, not alphabetical file names.
• [S2] Reading Systems defines processing of the EPUB navigation document.
• [S3] Accessibility guidance distinguishes navigation requirements including page-navigation applicability from merely having a visible contents page.

Related concepts: systems thinking, wealth structure, and the survival loop. Read the editorial policy and citation guide.