A panel can be factually correct and still difficult to understand. It can also be well written and approved in a document while the printer receives an older layout. These are different failure modes. A useful pre-print process separates them so that the right people check the right things in the actual version intended for production.

Begin testing while revisions remain manageable. Waiting until every panel has reached a polished layout encourages the team to defend the existing text because changing it feels expensive. Early samples can reveal problems of meaning and hierarchy, while a later complete proof checks the final public object. The objective is not an impossible promise of error-free publication. It is an accountable process that catches foreseeable problems and records what was actually approved.

Set the review order around the cost of changing the decision. A misunderstanding of the main argument should be addressed before polishing punctuation throughout the entire exhibition. Likewise, an unresolved object identity should not be concealed by approving a beautifully finished caption. Settling the consequential questions first gives later mechanical and production checks a more stable version to examine.

Establish the text inventory before distributing drafts

Give every label or panel a stable identifier, working location, text type and current version. The inventory should distinguish an object caption from a section introduction or an operating notice. These pieces require different information and different approval. A folder of documents named final, final revised and final new makes the current state harder to identify precisely when production pressure increases.

Keep one recognised place for the current wording. Design artwork can refer to that inventory, but it should not quietly become the only location of factual changes. Agree how an edit made during layout returns to the master text record. Otherwise the team can end with an approved document and a different approved-looking panel that nobody has compared directly.

Record who is responsible for each check. A curator may confirm object facts, an editor may resolve clarity and a designer may check hierarchy and fit. Some tasks overlap, but general approval from several people does not show that any one of them checked a specific date, attribution or object association. A simple responsibility record makes the process more reliable without requiring a large editorial department.

Check factual claims against their actual sources

Review names, dates, titles, quotations, technical terms and factual interpretations against the appropriate project sources. Identify what establishes each significant claim. A familiar anecdote should not pass into print merely because it has appeared in several internal drafts. Where evidence remains uncertain, revise the wording or attribution rather than allowing polished language to imply that the uncertainty has disappeared.

Verify the relationship between text and the particular displayed object. A description prepared for one loan may remain beside a replacement after selection changes. The text can still be accurate about the original example while misleading the visitor about what is actually shown. Use object identifiers and reference images to connect the proof with the installed selection, including reproductions and components displayed separately.

Check quotations in context and through the project's agreed contributor process where relevant. Editing punctuation or shortening a passage can alter meaning, especially when a qualification disappears. Keep the approved source and any required review record traceable. Do not invent a clearer quotation or treat an interpretive summary as the contributor's own words.

Test what unfamiliar readers think the text says

Select a representative sample of labels before laying out the entire exhibition. Include a short caption, a contextual panel and any text carrying an important or difficult distinction. Ask people unfamiliar with the drafting discussion to explain the main point in their own words. Their explanation is more useful than a polite answer to the general question of whether the writing is clear.

Ask a few targeted tasks: identify which object is discussed, locate a particular fact, explain a technical term or describe what remains uncertain. These questions distinguish different problems. A reader may understand the argument while associating the caption with the wrong example. Another may find the identifying information easily but misunderstand the central comparison. Each requires a different revision.

Avoid coaching the intended answer before the exercise. A long introduction by the writer can supply context absent from the actual label. Record the task, material and conditions so the team knows what the feedback refers to. A small diagnostic exercise can reveal an ambiguity, but it should not be presented as representative audience research or a numerical guarantee of understanding.

Check the order in which information becomes available

Read the headline, opening and caption separately as well as together. A visitor might encounter only one of these layers. The headline should make an honest promise, the opening should provide a useful entry and the caption should identify the actual object. Deeper paragraphs should add evidence or complexity rather than repeatedly announcing that the subject is important.

Identify references that depend on an earlier panel. Words such as this process or the next stage may be unclear to somebody arriving midway. Add the smallest necessary local context or change the wording. Testing should not assume that every visitor reads all text in a compulsory order unless the exhibition's actual operating format genuinely supports that requirement.

Look at the text as a group within a section. Individually concise labels can still repeat the same introduction several times. Conversely, a necessary definition can disappear because each writer assumed another panel contained it. A section-level review identifies these relationships before the physical arrangement makes them less obvious to the editing team.

Print representative proofs at their intended size

A screen review cannot establish how a label works on the wall. Print a representative proof using the intended scale and inspect it beside the relevant object or a suitable full-size stand-in. Check the actual relationship between text, object and viewing position. A label may be legible in isolation but difficult to connect with the thing it is explaining.

Use the expected lighting and material conditions where possible. Reflections, shadows and the contrast between text and background can affect reading. Do not infer physical usability from a high-resolution design file or rely on a generic type-size number as proof of access. Appropriate design guidance, specialist input and user feedback should inform the actual combination of typography, surface, placement and context.

Inspect longer passages from several realistic positions and with neighbouring information present. A panel may seem inviting when held at a desk but become visually dense among several competing displays. The response may involve editing, hierarchy, location or an alternative format. Treat the physical proof as a chance to improve the explanation rather than only to confirm that all the words fit.

Run a separate consistency and mechanical check

After meaning and factual issues are resolved, check spelling, punctuation, names, dates, capitalisation, units and terminology consistently. Use the publication's agreed conventions. Do not let a software spelling suggestion silently alter a proper name, quotation or specialist term. Automated tools can help locate candidates for inspection, but responsibility for the final wording remains with the editorial process.

Review practical information with the colleagues operating the exhibition. Room names, entry arrangements, help contacts and programme references can change independently of curatorial text. An accurately written notice can still be out of date. Keep time-sensitive information in a form that can be maintained appropriately rather than embedding an uncertain operating detail inside an expensive permanent panel.

Check accessibility alternatives and related outputs alongside the main text. A transcript, large-print version or digital guide should correspond to the approved public account while respecting the needs of its format. Copying an older draft into an alternative version can create conflicting information. Identify which linked materials need updating whenever a significant correction is made.

Approve the final artwork, not just the text document

Compare the laid-out proof against the current approved wording. Look for omitted lines, old captions, broken characters, unintended emphasis and changes made to make text fit. A shortened paragraph can lose an important qualification even when the designer's intention is simply to improve the layout. Any substantive edit should return to the relevant checker before release.

Confirm panel identifiers and file names through the production schedule. The printer needs to know which final file corresponds to which output, in what quantity and under which agreed specification. A message saying all panels approved leaves room for ambiguity if several versions remain in circulation. Identify the actual artwork and approval record explicitly.

Freeze the release set through a clear process while allowing a route for essential corrections. The point is to stop uncontrolled parallel changes, not to prevent a known factual error from being fixed. If a late correction occurs, identify which files and related outputs it affects, obtain the relevant approval and make the superseded version unmistakably obsolete.

A hypothetical proof catches an old object association

Imagine a hypothetical architecture exhibition replacing one model with another during late loan coordination. The master caption is updated correctly, but the designer's layout still uses the previous object's date and image. A general copyedit finds no spelling mistakes because both sets of information are internally plausible. This is an illustrative production scenario rather than a report of a real exhibition error.

The object-association check compares the artwork with the final selection register and reveals the mismatch. The team updates the correct panel, checks its new layout and identifies the affected alternative text version. The print-release record names the revised file so that the earlier artwork is not sent simply because it was attached to an older approval email.

A physical proof then reveals that the revised caption's identifying line is hard to distinguish from the interpretive paragraph. The hierarchy is adjusted without changing the factual meaning. Two different checks have solved two different problems: the first protected identity and accuracy, while the second made that accurate information easier to use in the actual display.

Complete the check after delivery and installation

Inspect the produced material against the approved release set. Check that the correct panels, finishes and quantities have arrived and report discrepancies through the actual supplier arrangement. A correct artwork file does not establish that every physical output matches it. Keep appropriate records of any replacement or agreed deviation so the installation team knows what is current.

After installation, check placement and association again. Panels can be installed in the wrong order, captions can migrate between similar objects and a correct label can be obscured by a final mount or open door. These are public-experience issues that no desk proof can fully resolve. Include them in the opening checks with the people responsible for interpretation and operation.

Preserve the final installed text, meaningful corrections and approval information in the project archive. If an error is identified during the run, use a proportionate correction route rather than treating print as an irreversible decision. The strongest result is not a claim that testing eliminated every possibility of error. It is a traceable explanation of what was checked, what was installed and how the team can responsibly improve it.

CheckPrimary questionUseful evidence
Factual reviewIs the claim and object association supported?Sources and final object identifiers
Reader taskWhat meaning is actually understood?Task-specific observations
Physical proofCan the information be used here?Full-size sample in context
Artwork releaseWhich exact version was approved?Named files and approval record
Installed checkDoes the public arrangement match the decision?Placement and association review

Questions specific to this guide

Can a single proofread cover all the checks?

A skilled reviewer can identify several kinds of issue, but the process should still distinguish facts, meaning, physical presentation and final-version integrity. Each needs an appropriate information basis.

Should the writer explain the exhibition before a reader test?

Provide only the context the actual visitor would have. Otherwise the test may assess the writer's verbal introduction rather than the label's ability to communicate independently.

Is approval complete when the Word document is signed off?

No. Confirm that the final artwork contains the approved meaning and that the installed panel is associated with the correct material. The different stages can introduce different errors.

Further reading

Authoritative background on related planning and access questions. These resources provide context; they do not endorse this guide or replace the current requirements for your specific setting.