# OCR for Law Schools: Clinics, Libraries and Journals

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2023-03-06T04:07:23Z
- Updated: 2026-10-11T19:22:20Z
- Category: Choosing legal OCR
- Canonical: https://www.schemalaw.com/blog/ocr-for-law-schools

> OCR for law schools, office by office: supervising clinic students with client files, cite-checking scanned sources, library scanning and what to teach.

OCR for law schools serves four groups with different stakes. Clinics convert real client paperwork under a supervising attorney. Journals check quotations against scanned sources. The library turns print holdings and faculty papers into searchable files. Students use it on readings. Each group needs its own rules, because only one of them is handling confidential files.

A law school is an odd buyer. It contains a working law office, a publishing house, an archive and several hundred people with no budget at all. This post takes each in turn, and ends with what the school should teach before any of them graduate.

Treat the clinic like a small legal aid office. A supervising attorney approves the tool, students convert client intake papers on a school account, and client confidentiality covers every upload. Teach the duty of competence as a habit of checking quotations against the page. Remind students that explaining a filing to a clinic client can slide into unauthorized practice of law if no attorney has reviewed it.

## Who in a law school uses OCR?

Five groups in a law school use OCR: clinic students and their supervisors, journal editors, librarians, faculty and their research assistants, and students doing ordinary coursework. They share a building and little else. Their documents, their volumes and the harm an error can do are all different.

| Who | What they convert | What a mistake costs |
| --- | --- | --- |
| Clinic students and supervising attorneys | Client notices, leases, agency letters, court papers, records | A wrong date or amount in a live case, or a confidential file in the wrong place |
| Journal editors | Photocopied book pages, old reporters, archival sources pulled for an article | A misquotation in print under the author's name |
| Law librarians | Print volumes, microform printouts, school records, donated papers | A collection that looks searchable and silently is not |
| Faculty and research assistants | Archival documents, historical statutes, older scholarship | A flawed quotation or a miscounted data set |
| Students in coursework | Scanned readings, casebook excerpts, handouts | A bad note, caught the first time it is used |

### The split that matters most

The line that matters runs between the clinic and everyone else. A clinic file belongs to a client, and the people touching it are not yet lawyers. A journal source or a course reading is usually public or at least not confidential. Rules written for the clinic are too heavy for a 1L with a scanned handout, and rules written for the 1L are far too light for the clinic.

### Who should set the rules

Someone should decide this on purpose. Left to habit, the clinic director sets clinic practice, the library sets its own, and a journal inherits whatever last year's board did. A one-page statement for each office is enough: which tools are approved, for which documents, and who to ask.

## Running OCR in a law school clinic with real client files

Running OCR in a law school clinic means the supervising attorney, not the student, decides which tool may hold client files and who may upload them. The student does the conversion and the first check. The attorney approves the tool in advance and remains answerable for what the student does with it.

Professional conduct rules put that weight on the lawyer. The comment to the [District of Columbia's Rule 5.3 on nonlawyer assistants](https://www.dcbar.org/for-lawyers/legal-ethics/rules-of-professional-conduct/law-firms-and-associations/responsibilities-regarding-nonlawyer-assistants) lists law student interns among the assistants a lawyer should instruct and supervise, and singles out the obligation not to disclose information relating to the representation. Student practice rules add their own terms. [California's rule on certified law students](https://courts.ca.gov/cms/rules/index/nine/rule9_42), for one, has the supervising attorney assume professional responsibility for the student's work product. Every state writes these rules differently, so read the ones that govern your clinic.

### A clinic upload protocol

Write the protocol down and hand it to each student in the first week.

1. **Name the approved tool.** One tool, chosen by the clinic director after reading its terms on training, storage and deletion. Students may not substitute a free website they found the night before a filing.
2. **Use clinic accounts.** A file uploaded under a student's personal login stays in that student's account after the semester ends. Sign in with an account the clinic controls.
3. **Say which documents may go up.** A clinic may have categories that need the supervisor's specific approval first, such as medical records, immigration files or anything under a protective order.
4. **Keep the original.** The scan or photo the client supplied is saved untouched in the clinic's file. The converted copy sits beside it.
5. **Check before relying.** The student reads names, dates, amounts and case numbers against the page image and notes what was checked.
6. **Delete on a schedule.** When the work product is saved in the clinic's own system, remove the upload from the tool. Do it at matter close and again at semester end.

### The semester turnover problem

A clinic's students change every term. That is the real risk, more than any single upload. At each turnover, the supervisor should confirm three things: departing students no longer have access, nothing remains in a personal account, and the incoming student knows where the converted copies live.

### A worked example

Take a housing clinic that represents a fictional tenant, Maria Ostrowski, in a nonpayment case. She brings a photographed lease, a rent ledger from the landlord and a notice to quit. The student, Devin, uploads all three under the clinic account, exports the ledger to a spreadsheet and the lease to a searchable PDF, and totals the ledger. His total is $40 lower than the landlord's. He reads the page image and finds that a faint "$40.00 late fee" line did not come through. He adds it by hand, notes the correction and the page, and gives the supervising attorney the scan, the export and the note together. That note is the supervision record.

Before any of this, decide whether the files may go to an online tool at all. The post on [attorney-client privilege and AI tools](https://www.schemalaw.com/blog/attorney-client-privilege-and-ai-tools) sets out the questions. Clinics that send students to outside placements should also read the guide to [OCR for legal aid offices](https://www.schemalaw.com/blog/ocr-for-legal-aid), since the host office's rules govern there.

## Can law students use a free OCR plan for coursework and research?

Yes. A free OCR plan suits law student coursework and research because the documents are short, public and low in stakes. The usual limits on pages and file count rarely bite on a handout or a few pages of a reading. A free plan is the wrong place for anything from a clinic or an externship.

Schema Law's free plan, as an example, takes one document a day of up to three pages and exports every format the paid plans do. The post on converting scans without paying covers [the Schema Law free plan in detail](https://www.schemalaw.com/blog/free-scanned-pdf-to-word#the-schema-law-free-plan-in-detail), and the [pricing page](https://www.schemalaw.com/pricing) lists the paid plans beside it.

### What three pages a day is good for

- A scanned statute section or a single old case a professor handed out.
- The two pages of a treatise you want to quote in a seminar paper.
- A table from an agency report that you need as text.

### What it is not good for

A sixty-page scanned reading will not fit. Before reaching for a paid tool, check whether the library already holds the reading in a text-based form, or ask the professor's assistant for the original file. A scanned course reading is sometimes a photocopy of a book the school also holds electronically.

### Three cautions for students

**Nothing is deleted for you.** A document stays in the account until you delete it. Export what you need, delete the upload, and do not treat a free account as storage for an outline.

**Your account is yours.** Anything from a clinic, an externship or a summer job belongs to someone else's client and goes through that office's tool, never a personal one.

**Converted text is a draft.** A quotation in a graded paper should be read against the page before you submit it, the same as it would be in practice.

## Cite-checking scanned sources for a law review

Cite-checking scanned sources with OCR speeds up finding the quoted passage, and does nothing to verify it. Search the converted text to locate the words, then compare the author's quotation with the page image, character by character. The image is the source. The converted text is only an index to it.

A source pull produces a stack of scans: photocopied book chapters, pages from bound reporters, archival letters, legislative history from microform. An editor assigned thirty footnotes can spend most of the hours hunting for the sentence on the page. Converting the pull and searching it turns that hunt into seconds.

### A cite-check routine for scanned sources

1. Convert each source to a searchable PDF, so the page image and the text stay in one file.
2. Search for a distinctive phrase of four or five words from the quotation, not the whole sentence. One misread letter defeats a long search string.
3. If the search finds nothing, do not record "not found." Try a different phrase, then read the cited page.
4. Compare the quotation with the image, not with the text layer. Check spelling, punctuation, capitalization, emphasis and any omission.
5. Confirm the pinpoint page from the printed page number in the image. The PDF's own page count starts at one wherever the scan started.
6. Record in the editing sheet that the check was made against the image.

### Old print is where it fails

Law reviews cite old material, and old material converts poorly. Expect trouble with the long s in eighteenth-century printing, ligatures, small capitals, blackletter headings, footnotes in tiny type and quotations in other languages. Two-column reporters can come out with the columns interleaved, which scrambles the sentence you are looking for. The post on [whether OCR can handle two-column documents](https://www.schemalaw.com/blog/can-ocr-handle-two-column-documents) explains how to spot that.

### Do not paste from the text layer

If an editor fixes an author's quotation by copying from the converted text, an OCR error can replace the author's error. Retype the correction from the image, or paste and then proofread it against the image. The guide on [how to quote text from a scanned PDF](https://www.schemalaw.com/blog/how-to-quote-text-from-a-scanned-pdf) gives the full method.

### What the board should decide

Write one line into the journal's manual: converted text may be used to locate a passage and never to verify one. Without that line, each year's editors rediscover the rule after a near miss.

## Digitizing law library collections and faculty archives

Digitizing law library collections with OCR is a project to plan before the first page is scanned. Decide what the collection is for, who may see it, what scan quality the text needs and how you will test the result. Whether a given work may be copied is a legal question for the library's own counsel.

On that last point, federal law has a provision written for libraries. [Section 108 of the Copyright Act](https://www.law.cornell.edu/uscode/text/17/108) limits a copyright owner's exclusive rights where a library or archives makes certain reproductions, and it attaches conditions, including limits on where digital copies may be made available. Whether a project fits that section, or some other ground, is not something this post can tell you. Ask the library's counsel before scanning anything still under copyright.

### Four decisions before scanning

- **Purpose.** Preservation, access for the school's own researchers, and public access lead to different choices about format and review.
- **Selection.** Start with what is unique to the school: faculty papers, moot court records, student publications, local bar materials. Standard reporters and codes are usually available elsewhere already.
- **Quality bar.** A collection meant only to be searched can tolerate some misread words. One meant to be quoted from needs proofreading.
- **Privacy.** Faculty papers hold student names, grades, recommendation letters and sometimes old client files from a professor's practice. Someone must review a donated box before it is converted and indexed.

### Testing before you commit

Pick twenty pages that represent the worst of the collection: a typescript carbon, a faded mimeograph, a handwritten note, a tightly bound volume with a curved gutter. Convert them and read the output against the pages. If the worst pages are unusable, plan for them to be image-only with a catalog description, and say so in the finding aid.

### Working around file limits

Tools set limits on file size and pages per document, and a bound volume scanned at high resolution can exceed them. Schema Law, for instance, accepts PDF, PNG or JPEG files of up to 20 MB. Split a volume into chapters or signatures, name the parts so they sort in order, and keep a list of what was converted.

Scanner settings, naming and quality control for a large backfile are covered in the guide on [how to digitize law firm paper files](https://www.schemalaw.com/blog/how-to-digitize-law-firm-paper-files). The method carries over to a library with little change.

## What should a law school teach students about the limits of OCR?

A law school should teach students three limits of OCR: converted text can be wrong while looking right, a search with no hits proves nothing about a poor scan, and the scan remains the record. A student who learns these on a seminar paper will not have to learn them on a filing.

### Where it belongs in the curriculum

Legal research and writing is the natural home, because that course already teaches students to verify sources. Clinic orientation is the second. A journal's training week is the third. None needs more than an hour.

### An exercise that works

Give each student the same five scanned pages: a clean typed letter, a faxed order, a page with a stamp across a paragraph, a table of figures and a handwritten note. Have them convert all five and mark every difference between the output and the page. Then ask two questions. Which errors would a spell checker have caught? Which would have changed a date, a name or an amount? The second list is the lesson.

### Habits to grade

- Keeps the original scan and names it so it cannot be confused with the copy.
- Reads names, dates, amounts and citations against the image.
- Reports a search honestly: what was searched, in what, and how good the scans were.
- Says which pages could not be read instead of guessing.
- Asks before uploading anything that belongs to a client.

### Why teach it at all

Graduates will meet scanned paper at work, and may have nobody to explain it. The post on [OCR for law clerks](https://www.schemalaw.com/blog/ocr-for-law-clerks) picks up the same habits where a clerkship begins, with a trial record in place of a source pull. A school that has taught the checking habit sends out people who can be trusted with a file sooner.

## Try it on a real document

Upload a scanned PDF and export it as Word, a searchable PDF or plain text on the free plan. Your documents are never used to train an AI model. [Start free](https://www.schemalaw.com/api/auth/google)