Check two-column PDFs in Book Isle: distinguish page appearance from text order

A two-column PDF can look correct while extracted text jumps across columns line by line. This affects search, copying and notes. Use an authorised short document to inspect display and text layer separately, retaining page references and anchor text rather than treating clear display as proof of all text operations.

Choose a page with unambiguous order

Choose a page whose left column continues into the right, with clear headings and prose. Verify the file and record both file and printed page numbers. Mark first and last left-column sentences and the first right-column sentence, separating basic column order from complex newspaper-like layouts.

Check the visual page first

Import locally and check complete columns, captions and page edges. Separate incomplete-file or source-resolution problems from text-order issues. Record visual display independently rather than classifying every anomaly as search failure.

Follow the intended visual reading path

Read the intended visual sequence before extraction tests, locating spanning headings and figures. Some side-by-side layouts are comparisons rather than sequential columns. Establish the author’s actual order first so an unfamiliar convention is not mistaken for a parser error.

Test selection only where available

If the current interface supports selection or copying, test a few sentences and compare them in a temporary note. Otherwise skip this test. Interleaved columns indicate a text-order limitation even when visual reading works. Verify against the page before retaining or sharing a quotation.

Distinguish word hits from phrase search

Search a visible short word, then a short continuous phrase. A word hit without a phrase hit may reflect order, line breaks or encoding. Verify the destination page, and distinguish image-only scans. Clear page display does not prove column reconstruction or OCR.

Check captions and footnotes separately

Compare captions, footnotes and reference markers with the page rather than assuming extracted order reflects authorship. Anchor notes with file page, heading and text. Include essential footnote context or direct readers back to the page instead of treating separated prose as the complete argument.

Compare the same file in another tool

Repeat the same short selection and search in another reliable tool using the identical PDF. Shared failure may point toward source structure, while different results establish only a difference on the tested page. Record file identity and tested pages rather than generalising across versions or all pages.

Retain page references when creating a reflow copy

If authorised, create a reflow-friendly copy with an appropriate external tool, verifying paragraphs, captions and notes. Retain the original for layout and citation pages. Reflowed page numbers are not original printed pages. Recheck the three anchors rather than assuming conversion repairs every column issue.

Save reliable anchors for problematic passages

Bookmark the target page and record file page, column and opening words rather than only copied text. Use available note controls or a separate checklist. Original pages and text anchors survive device or copy changes better than reflow positions; distinguish location from the note explaining its limitation.

Conclude separately for display and text tasks

Judge visual reading, target-word search and copied/extracted order separately, then sample another page. A failed task does not invalidate every use, but its limitation matters for quotations. Keep the sample and version without deleting notes for repeated imports; import success is not complete verification.

Related tutorials and sources

PDF navigation and search, Text-search diagnosis.

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.