Scanned documents in agenda packets

Why a scanned page is a problem

A scanned page is a photograph of paper. It looks like text, but to a screen reader it is one image with nothing to read aloud. People who magnify text or change colors cannot reflow it either. That is why a single scanned letter inside an otherwise tagged packet leaves a gap some readers cannot cross.

The W3C puts it plainly in technique PDF7: a document of scanned images of text “is inherently inaccessible because the content of the document is images, not searchable text.” PDF7 is listed under WCAG 2.1 success criterion 1.4.5, Images of Text.

Packets pick up scans in predictable places: signed resolutions and ordinances, letters from residents and agencies, recorded deeds and easements, older plans, and pages someone printed, signed and scanned back in.

Typed source beats a scan

The typed source file beats a scan every time it exists. Its text is already real, Word can write the heading, list and table tags when you save as PDF, and nothing has to be guessed by software. A scan has to be recognized, proofread and tagged by hand, and it still carries a risk of misread words.

PDF7 says the same thing: authors “should use actual text rather than images of text,” and OCR is the route when they “do not have access to the source file.” In practice that gives you a simple order of preference:

1. Post the typed original. The resolution you drafted in Word, saved as a tagged PDF. See how to make an agenda PDF accessible for the save settings.
2. Ask the sender for the file. Engineers, attorneys and agencies usually have the Word or PDF original. Ask for it when you ask for the document.
3. Scan and recognize it only when no source exists, such as an old ordinance or a handwritten letter.

How to run OCR on a scanned page

Optical character recognition, or OCR, turns the picture of each letter into real text. In Acrobat Pro, the steps in W3C technique PDF7 are Scan and OCR, then Recognize Text, then Correct Recognized Text. Other tools do the same job. Scan cleanly first: straight, at readable resolution, one page per image.

W3C PDF7 walks through the Acrobat Pro version: text that the software is unsure of is flagged as an “OCR suspect,” and Correct Recognized Text shows each suspect one at a time so you can fix it.

Word can also do this. Microsoft’s Scan and edit a document says to open the scanned PDF with File, Open, and Word converts it into an editable document. Microsoft notes that the conversion “works best with documents that are mostly text” and that “lines and pages may break at different locations.” Once it is a Word file, you can fix the headings and save a fresh tagged PDF.

Proofread the recognized text

Proofreading is the step that is easy to skip and costly to miss. OCR reads a 5 as an S, a 1 as an l, and drops a line from a faint copy. In minutes, ordinances and resolutions, those are the words that matter: dollar amounts, parcel numbers, vote counts, dates and names.

W3C PDF7 gives test methods: listen to the file with a screen reader or read aloud tool, or save it as text and read that. A short routine that works for most clerks:

• Read every number and proper name against the paper original.
• Check that no line, footnote or margin note is missing.
• Check that the reading order follows the page, especially with columns or a letterhead.
• Look for words in a stamp or seal that OCR turned into nonsense, and remove or describe them.

OCR is not the same as a tagged PDF

An OCR text layer is not the same as a tagged PDF. OCR makes the words readable, but the file may still have no headings, lists or table structure, no title, no language and a jumbled reading order. WCAG 2.1 success criterion 1.3.1 asks that structure shown on the page is also built into the file.

After OCR, add or check the tags, set the document title and language, and correct the reading order. W3C PDF7 notes that Acrobat Pro may add tags automatically during OCR; treat those as a first draft. The Understanding 1.3.1 page explains why structure matters. Our WCAG checklist for documents lists the rest of the checks.

How to handle signatures, stamps and handwriting

Signatures, stamps and handwriting are the parts OCR handles worst. A signature is an image of a name, so it needs a text stand-in. The simplest route is typed text on the posted version, such as “Signed by the Mayor and attested by the Town Clerk,” with names and the date, and the signed scan kept in your records.

If the signed image stays in the posted file, give it alt text that says whose signature it is, using technique PDF1. A recording stamp or clerk’s seal usually carries a fact, such as a reception number or date, so type that fact into the alt text. Purely decorative marks, such as a border, can be hidden as artifacts under PDF4.

Handwriting, such as margin notes or a handwritten public comment letter, usually defeats OCR. Type a transcript and place it with the image, labeled as a transcript.

Old scanned ordinances and records

Old scanned records are a separate decision. Under the federal rule, a PDF posted before your date may fall under the preexisting documents exception, but only if nobody currently uses it to apply for, gain access to or take part in a service. A town code that people rely on today is in use.

Conventional electronic documents that are available as part of a public entity’s web content or mobile apps before the date the public entity is required to comply with this subpart, unless such documents are currently used to apply for, gain access to, or participate in the public entity’s services, programs, or activities.

28 CFR 35.201(b)

Our guides on the archived content exception and agendas and minutes walk through both tests with examples.

A worked example: a signed resolution

A worked example shows the choices in order. Picture a made-up town we will call Anytown. Its packet for the October meeting includes a signed resolution from September, a letter from the county, and a 1987 ordinance that the staff report cites. Each gets a different treatment, and each takes minutes, not hours.

• The signed resolution. The clerk has the Word original. She posts that as a tagged PDF, adding the line “Signed by Mayor A. Sample and attested by Town Clerk B. Example on September 15, 2026.” The signed scan stays in the records vault.
• The county letter. The clerk emails the county for the original PDF. It arrives tagged, so she checks the title and language and drops it in.
• The 1987 ordinance. No source exists. She runs OCR, proofreads every section number and date against the paper book, adds headings, sets the title and language, and runs our free PDF checker on the posted file.

What we do about it

What we do about it is handle the scans for you. Each month we find the packets and minutes your town posted, recognize and proofread scanned pages, add tags, titles and languages, describe signatures and stamps, and have a person check every file. The work goes into your dated Readable Record. See how it works or pricing.

This guide explains the rule in plain language. It is not legal advice. For decisions about your town, talk to your attorney.