Record
Digital Evidence for Readers: Archives and Provenance
A page said something. Then it changed. Whether anyone believes your account of it depends on what you kept, and on whether a stranger can check it without trusting you.
Digital evidence, for a reader, is whatever lets you show that a web page said a particular thing at a particular time. Students need it when a cited source is edited before the essay is marked. Journalists need it when a company rewrites a claim. A screenshot feels like proof, and it is the weakest form of it. This page explains what a believable record contains, compares the ways of keeping one, and documents one page that changed. It is for readers and students, and it is not legal advice.
In short
A believable record of a web page has five parts: the exact address, the date and time you read it, a copy saved by a third party such as the Wayback Machine or archive.today, a highlight link that opens the page at the passage, and your own dated note. For images, add any Content Credentials attached.
What makes a record of a web page believable
A sceptical reader asks: how do I know you did not make this up? Each part of a record answers it. The address says which page, the date says when, the archive copy says a third party saw the same thing, the highlight link says which passage, and your note says what you concluded.
No part is strong alone; together they cross-check each other, and the third-party copy is the one you could not have altered. The Internet Archive's help page on saving pages says a Save Page Now capture gives you a permanent address that survives after the original changes or is removed. That is why it sits at the centre.
Why a screenshot alone is weak digital evidence
A screenshot is a picture of your screen, cropped as you chose. It carries no address unless the browser bar is in it, no reliable time, and no link a reader can open. Any image editor can change a word without trace.
The Verification Handbook, written for journalists, lists four things to confirm about any online content: provenance (is this the original?), source, date and location. A screenshot alone answers none of them. Keep it for context, never in place of the other parts.
Screenshot, PDF, archive and highlight link compared
| Method | What did it say? | When? | Can someone else check it? | Can it be faked? |
|---|---|---|---|---|
| Screenshot | The visible part, cropped by you | Only if you note it | No, it is your image | Easily, with any image editor |
| Saved PDF | Full page text, address in the footer | Print date, set by your computer | Only your file | Yes, with a PDF editor |
| Third-party archive | Full page as the service fetched it | Capture time, recorded by the service | Yes, anyone can open the capture | Very hard; the copy sits on a server you do not control |
| Highlight link plus note | Points at the exact passage on the live page | The date in your note | Yes, while the page is unchanged | The link cannot be faked, but the page can change |
The highlight link also fails loudly: edit the passage and the browser opens the page without the highlight. Our page on what a text fragment link needs in order to work explains why.
A worked example: one page that changed
The page here is invented. Suppose a town library page at example.org/library/hours reads, on 3 September: The Riverside branch will remain open on Sundays through the winter. On 20 September it reads: The Riverside branch closes on Sundays from 1 October. You wrote to the council on 5 September relying on the first sentence. A record made that day looks like this.
- Address:
https://example.org/library/hours, copied from the address bar. - Archive copy: a Save Page Now capture requested on 3 September, giving an address beginning
web.archive.org/web/20260903, plus a second capture at archive.today. - Highlight link:
https://example.org/library/hours#:~:text=remain%20open%20on%20Sundays, made by pasting the sentence into the page that builds a highlight link from a sentence. - Dated note: 3 Sept 2026, 14:10 BST. Hours page says Riverside stays open Sundays through winter. Basis for the council letter.
On 20 September you add a second entry for the new wording, and one line: Wording changed between 3 Sept and 20 Sept; both captures kept. Anyone can open the two captures and see both sentences with the recorded dates. The old highlight link no longer lands on a highlight, which confirms the change. Nothing asks the reader to trust you.
The routine for keeping digital evidence of a page
Copy the full address
Click in the address bar, select all, and paste it into your note. A shared address may differ from the one that loaded.
Save a third-party archive copy
Open the Wayback Machine's Save Page Now form, paste the address, save, and copy the capture address. Repeat at archive.today if the page matters.
Make a highlight link to the passage
Select the sentence and use the one-click bookmarklet for copying a highlight link.
Write the dated note
What the passage said and why you needed it, with date, time and time zone. Later changes of mind go in new dated lines.
For images, inspect Content Credentials
The C2PA explainer describes them as signed metadata recording who made a file, when, and with which tools, and showing whether it was altered afterwards. Many images carry none; note that too.
File everything together
One folder per page, named by date and topic. Marks scattered across chats and downloads are not a record.
Citing an archived page in your work
Cite the archive copy alongside the live address. The Washington University Libraries guide to using web archives gives APA, MLA and Chicago examples and quotes the MLA rule: treat the Internet Archive as a second container, the original page first and the Wayback Machine address as the location. An archived link names one stable version, so a reader knows which you used.
This fits checking a source before you rely on it: verify, archive, then cite the archive. The middle step is the one that saves the essay when the source vanishes.
A note on courts and institutions
Courts, regulators, universities and newsrooms have their own rules about what counts as evidence and how it is collected. Some require a particular tool, a witness, or a documented chain of custody. Nothing here replaces those rules; this is a habit for readers. For a legal matter, ask a professional.
Pitfalls that weaken a record
- Archiving after the change. A capture made once the page has been edited proves only the new wording. Archive the moment you rely on a page.
- Cropping the screenshot. Without the address bar it is a picture of some words. Capture the whole window.
- Assuming the archive already visited. The Wayback Machine may have no capture, or one from years ago. Request your own; the help page notes a capture saves one page, not the pages it links to.
- Editing the old note. A revised note cannot be dated. Add lines; never overwrite.
Expert tips
- Archive before you quote, every time. The cost is thirty seconds; the benefit arrives years later.
- Open the capture and read it. A capture of an error page or a cookie banner is not a record of anything.
- Inspect image credentials before you argue about an image. Our guide to what a Content Credentials label can and cannot tell you covers what an absent label means, which is usually nothing.
- Treat the whole folder as the record. This is the principle behind the guides on keeping highlights as a working record: a dated mark, a source and a link to the exact line, kept together.
For the writing side of the same record, the version history a writing tool keeps as an authorship record documents your own document the way an archive documents someone else's, and a web citation with the archive address in it is where the saved copy ends up.
Common questions
Is a screenshot digital evidence?
Weak evidence. A screenshot shows what was on your screen, cropped as you chose, with no address and no trustworthy time, and it can be edited without trace. Keep it for context and pair it with a third-party archive and a highlight link.
How do I prove what a web page said on a certain date?
Save the page to the Wayback Machine or archive.today on that date and keep the capture address; the service records the time. Add a highlight link to the passage and a dated note. If the page changes later, capture it again so both versions exist.
What if the Wayback Machine has no copy of the page?
Request one with Save Page Now; the Internet Archive's help page says the capture gives you a permanent address. If the page blocks crawling or sits behind a login, the capture may fail. Then save a PDF, take a full-window screenshot, and note that no third-party copy was possible.
Does a highlight link prove anything by itself?
Only that the words were on the page at the moment the link works. It stores no copy and no date. Its value is precision: it sends a reader to the exact sentence. Pair it with an archive capture, which stores copy and time.
Do all images have Content Credentials?
No. Content Credentials are signed metadata attached to a file that record who made it, when, and with which tools, and reveal edits made after signing. Most images online carry none, and an absent label does not mean a file is fake. Note the result either way.
Digital evidence, for a reader, is a small habit: copy the address, save a copy with a third party, make a highlight link to the line, take a full screenshot, and write a dated note. Each part answers a question a sceptic asks, and the archive answers the hardest one. Do it when you rely on a page, and what it said that day becomes something anyone can check.