Check DOCX quality in Book Isle: verify table relationships and footnotes

Book Isle can bring DOCX documents onto the shelf, but readable text does not guarantee preserved tables, footnotes or layout relationships. Use a short authorised document to distinguish present text from preserved meaning. Retain the original before deciding whether extracted text suffices or a layout-preserving reference is also needed.

Choose a verifiable sample

Start with three to five pages containing a simple table, footnote and prose. Verify the original in a layout-capable tool and record its version. Mark a header, middle cell and full note as checkpoints. Replace an already incomplete source rather than blaming import.

Understand the reading copy

DOCX may become reflowed extracted text, so do not assume original fonts, pages, table borders or floating objects. Distinguish prose reading from precise table or page-based reference. Define the use before judging success; a title and cover do not prove complete content.

Check prose boundaries after local import

Import locally and verify filename and format. Check beginning, middle and final paragraph for truncation or summary-only extraction. Note repeated headers without deleting the original. A local sample separates parsing from remote-download issues; opening elsewhere does not guarantee identical extraction.

Match table headers to values

Read table headers with sample values against the original. In continuous text, spaces and newlines may no longer explain columns. Check especially adjacent numeric columns and units. If relationships require guessing, do not use the text copy alone for table lookup.

Inspect merged cells and blanks

Check merged headings, empty cells and multi-row notes, which can carry grouping meaning. A blank may mean not applicable or inherited context. Test a simple table first and retain complex tables in the original rather than rewriting data merely to fit the reading layout.

Verify note markers and note text separately

Locate a note marker and its actual corresponding text. A surviving number is not proof of a preserved note. Check key names and reference details through search or comparison. If only the marker remains, read the note in the original rather than promising a universal recovery switch.

Check images and equations by their meaning

Identify whether an equation or figure is text or an image in the source. Nearby prose cannot prove its presence. Keep a verified PDF or DOCX layout reference alongside the reading copy, checking any conversion’s pages, equations and captions rather than renaming extensions or assuming conversion restores omissions.

Record traceable problem locations

Record issues with section, table title, short anchor text and source version rather than reflowed page alone. A note should direct you back to the original, not convert a guess into source data. Verify column and unit before extracting a value.

Separate source problems from reader behaviour

Compare the same source in its original editor and a simpler authorised DOCX in the reader. Complexity-specific failure suggests structural differences. For all-file failure, record format, size, entry and message with a small reproducible sample. Test copies without deleting annotated books.

Decide acceptance by the intended task

Judge prose reading, table lookup and footnotes separately. A copy may pass one task and require the layout reference for another. Keep checkpoints and version for future samples. This is a user-executable acceptance method, not an automatic DOCX repair or a claim of device testing.

Related tutorials and sources

Document import basics, Managing format variants.

Checked on October 27, 2026 against the official Book Isle App Store listing and Apple’s public version information: current version 1.1.3, requiring iOS / iPadOS 15.6 or later. This is a file-quality checking workflow. Import, text extraction, layout preservation and reading-data sync are distinct capabilities. Readers supply their files and references; controls depend on the installed version. No step-by-step physical-device testing across all sample formats is claimed.