Chapter 23 answers

Fictional training example — not legal authority.

Letter from Mara Vale to Omar Reed 2 (Apr. 3, 1932), Folder 8, Box 4, Harbor Port Committee Papers, Series B, Harbor Civic Archive, item HCA-17-44, https://example.org/archive/HCA-17-44.

The citation identifies creator and recipient, document and date, supporting page, collection hierarchy, custodian, and item identifier. “Box 4” is only a container label and may recur across collections. The item number and catalog link help recover this letter while the collection information preserves its archival context. The page locator connects the item to the statement used.

A long citation is not automatically useful: each component here has a supplied retrieval role. Do not replace the creation date with a later catalog update or call the archive the letter’s author.

Could a researcher distinguish this letter from another item in Box 4 and reach page 2, while knowing both its creator and custodian?

Fictional training example — not legal authority.

“Vale reported not having received a reply as of April 3, 1932.” Cite the supplied letter at page 2. A claim that the committee rejected the request would require evidence of that decision, such as inspected minutes, a decision notice, or a responsive letter actually conveying rejection.

The available statement reports the writer’s receipt of correspondence. It does not establish the committee’s action, motive, or internal decision. The absence of a reply in this packet also does not prove that no reply exists elsewhere. Preserve both the writer’s perspective and the stated date.

“The committee never replied” is broader than “Vale had not received a reply by April 3.” “The request was rejected” adds a different event altogether. Neither can be repaired by adding a more detailed archival citation.

Does the revised sentence report only the supplied letter’s account, and is the additional evidence described as needed rather than as already found?

Fictional training example — not legal authority.

1932 is the letter’s creation date; 1980 is its accession into the collection; 2020 is the digitization date; 2025 is the later cataloging event. Cite it as the 1932 letter, retaining the other dates in separately labeled provenance fields when useful.

Each date answers a different question. A digital copy can be created decades after the historical item, and its description can be revised later still. The object’s identity should not drift each time a repository changes its records. When discussing digitization or catalog history as the proposition itself, those later events may deserve their own citations.

Selecting the newest date by default would turn a historical letter into an apparent contemporary communication. Conversely, omitting all provenance dates can make it difficult to tell which representation or description was inspected.

Can each date be paired with its actual event, and does the final citation avoid presenting accession, scanning, or cataloging as the letter’s creation?

Fictional training example — not legal authority.

Lina Vale, Port Records 18 (version dated Mar. 1, 2025) (unpublished manuscript, Harbor Civic Archive, item HCA-M9).

The model identifies the author, title, pinpoint, dated version, publication status, and repository item. The date belongs to the supplied manuscript version. “Unpublished manuscript” accurately describes what is given; the archive provides custody, not necessarily commercial publication. No publisher, publication agreement, or forthcoming status is supplied.

Do not label the work a book from Harbor Civic Archive merely because the archive holds it. Do not infer that public access means final publication, or that an unpublished manuscript is forthcoming. Those are distinct facts needing evidence.

Could a reader identify the specific March 1 version and page 18, and does the status description avoid adding a publisher or publication promise absent from the packet?

A descriptive note may place the repository information after the main citation rather than inside the status parenthetical, provided the version and custody remain clear.

Fictional training example — not legal authority.

Harbor Records Office, Finding the Register, at 03:12–03:27 (May 2, 2025), https://example.org/video/register.

The institutional creator, video title, date, direct URL, and timestamp range identify both the work and the relied-on segment. The title alone names the recording but leaves the reader to search its duration. The timestamp belongs to the supplied version; another edit may place the same words at a different time.

Do not use the upload date as a timestamp or provide an invented transcript page. If the argument concerns words spoken, inspect the segment and its context rather than relying only on a platform’s caption preview. The exercise supplies no speaker identity beyond the institutional publication metadata.

Can the reader open the named video and reach the full relevant statement between 03:12 and 03:27, with no unexplained version or locator change?

Fictional training example — not legal authority.

Use 00:10 in the inspected 30-second clip, and identify the source as that clip or edited excerpt. Record that the full recording was not inspected. Do not cite 41:35 as though you checked the corresponding passage in the full version.

Timestamps are coordinates within particular files. The fact that the packet supplies the full-recording coordinate does not turn an unperformed inspection into a completed one. If the claim requires broader context, obtain and watch the relevant portion of the full recording, then decide which source and locator to cite.

Combining the full recording’s title with the clip’s timestamp can misdirect the reader. Using the full timestamp without disclosure can also imply a scope of verification not performed. Keep version and locator together.

Does the citation identify the actual inspected representation and its 00:10 coordinate, while the research note preserves the full recording as uninspected?

Fictional training example — not legal authority.

Harbor Data Office, Filing Counts, version 2.1, tbl. B (June 1, 2025), https://example.org/data/filings-2-1.

Method note: The count uses Table B rows labeled “2024” and excludes withdrawn filings.

The citation identifies the dataset release and table; the method note identifies the subset and exclusion. Those details let a reader understand which records would enter the calculation. The packet supplies no DOI and no numerical total, so the answer should invent neither. A release-specific URL is the available access route.

Citing a general data homepage would conceal the version. Giving only a total would conceal the filtering decision. Excluding withdrawn filings is a substantive method choice that should remain visible, not be buried in an unexplained spreadsheet formula.

Can another researcher identify version 2.1, select the 2024 rows in Table B, and apply the stated exclusion without guessing which release or population was used?

Version 4.5 identifies the particular DataCite documentation release. The documentation DOI, https://doi.org/10.14454/g8e5-6293, identifies that documentation object. It should not be described as covering every later release or as the separate schema object, which has its own identifier on the official version page.

Related objects may share a version label while serving different functions: documentation explains a schema, and the schema object is separately identified. The model records the documentation’s supplied title, credited working group, publisher, and 2024 date. A substantive claim about a property still requires reading the relevant documentation passage.

A resolving DOI proves an access route or identity relationship, not that a particular technical claim was checked. Replacing the documentation identifier with the schema identifier would direct the reader to a different object even though both concern release 4.5.

Does the citation identify the documentation actually used, and does any broader claim avoid silently including later releases or uninspected schema properties?

Fictional training example — not legal authority.

Quote the clearly legible image as “not received,” correcting the OCR’s omission. Mark the genuinely unreadable word as illegible under the chosen quotation convention rather than guessing it. Preserve the image’s item and page locator, the OCR version, and a note explaining the comparison and remaining uncertainty.

The image resolves the first problem because the negative word is visible. The second problem remains evidentiary uncertainty. Treating both as routine typographic fixes would conceal the difference between a verified correction and an unsupported reconstruction. Read the surrounding passage to ensure any omission or bracketed notation preserves meaning.

Do not quote the flawed OCR as if it accurately represents the original, and do not silently supply a plausible word for the illegible portion. If analyzing the OCR error itself, identify that representation explicitly.

Can another reader reproduce the image/OCR comparison and distinguish the restored, visible “not” from the word that remains unreadable?