A useful product-page citation does more than save a link. It tells a second reader which page was checked, which torso model the page identified, which field was read and when the check happened. That keeps a source note narrow enough to verify without turning one observed field into a broader product claim.

1. Start from the exact product page

Open a direct listing such as the T528 product page or the T529 product page, then save the final product-page URL. A category or search page may help with navigation, but it does not identify the exact page that supplied a field.

2. Capture four identity fields together

Record the full URL, visible page title, model identifier and access time in one source note. If the title and description show different model identifiers, mark the citation as conflicting and stop. Do not choose the version that looks more plausible.

3. Name the field you actually checked

Write a field label such as title, description, option selector or image caption, followed by the exact visible wording needed for the note. Keep separate fields separate. A value in one field does not silently verify every other field on the page.

4. Prefer a stable, direct URL

The W3C note Cool URIs don't change explains why stable identifiers matter on the web. For a product citation, keep the final direct URL after redirects and record the access time. If the page later redirects or disappears, report that new state instead of rewriting the earlier citation.

5. Keep similar model citations separate

Cite the T527 listing, T61 listing and T69 listing as three separate sources. Similar navigation, imagery or wording does not merge their identities. Each note should point back to one URL and one model identifier.

6. Test whether another reader can retrace it

Give the citation to someone who did not make the original note. They should be able to open the same URL, identify the same model marker, locate the named field and compare the current wording with the recorded wording. If the page is unavailable, the correct result is unavailable at recheck time—not a guessed replacement.

This method documents a specific page observation. It does not prove that the wording is permanent, that another model shares the same fact or why the page may later change.