What an archive search can and cannot close

Eighteen units, with different roles

The audited Inquisition research currently represents 18 distinct archival units: seven Ribadavia targets, four cross-border candidates or comparators, and seven contextual or unlinked units. This is a count of represented records, including books and dossiers. It is not a count of fully inspected trials, proven family cases, or all surviving material relevant to Ribeiro.

The known units have very different evidentiary states. The Méndez civil dossier is held and has selected working editions. Justa Rodrigues's Coimbra process is currently known through its catalogue description. A two-page archive publication about Fernando Álvarez's confiscated-land rents describes a 93-folio original; possessing the publication is not possessing those 93 manuscript folios. [1] [2]

Searching a series

The official Santiago fiscal subseries description reports 89 contained units. That number gives a useful task: enumerate the children, preserve their identifiers, and inspect the relevant descriptions. It does not mean that 89 Ribadavia cases have been found, or that every original can be opened in the present research session. The full child-level sweep remains unfinished. [3]

Civil cases have a separate route. So do criminal matters, relations of cases and autos, correspondence, personnel records, and other administrative material. A search restricted to one title field or one tribunal label can miss documents describing the same people in another role. A useful plan therefore follows archival structure as well as names. [4]

Access failures belong in the record of the search. A timed-out page or an unavailable image means that route was not successfully inspected. It cannot be counted as a negative finding about the archive's holdings.

Names, roles, and false matches

Historical spelling creates obvious search work: Ribadavia and Rivadavia; Rodríguez and Rodrigues; Méndez and Mendes. The harder problem is deciding whether a result concerns the same person. Chronology, spouse, occupation, residence, a signature, or an explicit cross-reference can help. A spelling resemblance cannot do that work alone.

The search should also follow the roles already supplied by a source. A dowry can lead to a payer, a notary, an heir, or an assignee. A property can lead to a neighbour or a convent. The 1585 receipt supplies two Porto payers, but it does not license calling them Isabel's relatives. A place such as Dueñas may be a person's residence rather than part of a compound surname. Preserving those distinctions prevents a search result from becoming a false family link. [5]

What forty-eight queries mean

The current bounded pass comprises 48 public-web-index queries. This records what was tried in that interface. It is not an exhaustive native PARES search, and it does not show that every relevant catalogue record was indexed, returned, or inspected. A statement such as no relevant hit in this query is therefore much more defensible than no record exists.

The search counts also remain separate from the translation counts. Finding a catalogue item does not mean acquiring its images. Acquiring a file does not mean reading every face. Surveying the structure does not mean transcribing its text. Finishing the boundary of one instrument, while retaining gaps, does not finish the case.

A useful next question

The next step is often narrower than search everything. It might be to identify a cited deed, finish enumerating a particular subseries, read a witness from opening to signature, or resolve one ambiguous relationship clause. A search can be closed when its stated scope has been inspected or its precise blocker recorded. The historical question can remain open.

That is how this journal measures progress: by source identity, bounded coverage, and conclusions that another reader can check. It leaves room for a new record to change the account without requiring the earlier uncertainty to be hidden.