Writing Object Labels at Three Reading Lengths: A Practical Guide

Three lengths serve three different reading situations

An object label has a limited job: identify an object, provide useful context and give a visitor a reason to look more closely. A short label may contain only an identification and one significant fact. A medium label can explain the object's use or historical setting. An extended label can introduce a more complicated relationship between the object, its maker, its owner or the collection. These are differences in depth, not differences in what the institution claims to know.

For a practical object label template, begin with three working limits: 30 words for a short label, 60 words for a medium label and 100 words for an extended label. These are editorial targets, not universal museum standards. Local exhibition requirements, accessibility guidance and institutional style policies should determine the final limits. The important point is to establish the limits before drafting, so an AI assistant has a defined task.

• Short: identification and one essential fact.
• Medium: identification, context and significance.
• Extended: identification, context, evidence and a carefully qualified interpretation.

Prepare a source record before drafting

A writing assistant needs a controlled set of facts. The most useful starting material is an approved catalog record containing the object's preferred name, accession number, materials, dimensions, date or date range, maker or attribution, provenance information and documented use. Include the source of uncertain information where possible. A catalog field that reads 'attributed to' must remain distinguishable from a confirmed maker attribution throughout the writing process.

Consider a hypothetical museum record for a stoneware storage jar. Its documented information identifies the material as stoneware, records a height of 24 centimeters, gives a broad date range of approximately 1880 to 1900 and describes the object's use as domestic storage. No maker is recorded. The assistant should receive those facts and an explicit instruction not to supply a place of manufacture, owner or named maker. Missing information is a boundary, not an invitation to complete the story.

Give the writing assistant separate instructions for each length

A single request for three increasingly detailed versions can encourage repetition or introduce extra information in the longer version. Separate instructions make the task easier to control. Specify the audience, maximum word count, approved source material, preferred terminology and prohibited assumptions. Ask the assistant to retain approximate dates and uncertainty markers exactly where they affect meaning.

For example, a short-label instruction might read: 'Write a museum object label of no more than 30 words using only the supplied record. Identify the object, material and documented date range. Do not invent a maker, location or provenance. Use plain American English.' The medium and extended instructions can add context and interpretive requirements without expanding the factual permissions.

• Set a maximum word count rather than asking for a generally short text.
• Provide the approved object record as the sole factual source.
• Require uncertain dates and attributions to remain qualified.
• Ask for unsupported details to be identified as questions, not supplied as facts.

Compare the three versions against one factual baseline

The three lengths should be checked against the same record. A useful editorial comparison separates facts, interpretation and language. For the hypothetical jar, 'stoneware' is a documented material; 'approximately 1880 to 1900' is a qualified date; 'used for domestic storage' is a documented functional description. Calling it a farm kitchen jar would add a location and setting that the record does not establish.

A short version should not lose the uncertainty in a date merely to save words. A medium version should not acquire an invented household history. An extended version should not become a miniature essay containing unsupported claims about trade, regional production or social status. The additional space should explain the significance of known information, not disguise gaps in the record.

Reading level is not a substitute for historical accuracy

An exhibition label reading level concerns how easily a visitor can understand the text. It does not determine whether the historical content is sound. A sentence can be grammatically simple and still contain a false attribution. Conversely, an accurate sentence may be difficult to read because it contains unnecessary specialist vocabulary, passive constructions or several ideas at once.

Ask an assistant to replace unfamiliar terminology with ordinary language while retaining the underlying meaning. If a technical material name is essential, retain it and explain it briefly. For example, a label can identify stoneware and describe it as a type of fired ceramic. A local museum's preferred terminology and accessibility standards should guide the final wording. Collection management metadata tools may store technical information separately from public-facing text, allowing the label to remain readable without sacrificing catalog precision.

Keep the approved record and the public label distinct

An object label is an interpretive text, not a replacement for the catalog record. The record should retain fuller descriptions, source references, accession information and qualification notes. The label selects a small portion of that information for a particular audience and location. This distinction matters when a museum catalogue backlog contains incomplete descriptions: a polished label can make an uncertain record appear more settled than it is.

AI-assisted drafting should be recorded according to institutional policy. A practical note might state: 'Public-facing label text drafted with AI writing assistance from the approved catalog record; factual content and final wording reviewed and approved by collection staff.' The wording should be adjusted to reflect the actual contribution. If the assistant also summarized or restructured historical information, that work should be described accurately.

A controlled workflow for the next object

The most manageable approach is to select one object with a reasonably complete record, establish the three word limits and generate each version from the same approved information. Compare every proper noun, date, material, measurement and attribution with the source. Read the text aloud to catch awkward phrasing, then ask a second staff member to consider whether the interpretation goes beyond the evidence.

Keep the approved versions with their source record and identify which text is intended for each display context. Begin with a single 30-word label and its corresponding medium and extended versions. This gives the institution a reusable starting template before applying the approach to more complicated objects.

Questions readers ask