ISAD-G, the international standard for archival description from the International Council on Archives, has spent more than thirty years saying what must be described in an archive and in what order, and its second edition, from 2000, is still the reference for archives in Latin America and Spain. The standard does not say how to bring that description into a digital system, and that is where decisions appear that are best made before configuring anything: how many levels to use, how to store dates that are not exact, how to allow searching by subject, and what to do with the conditions of access and use area, which almost always gets left for later.
Why an archive is described by provenance and in levels, from the fonds down to the item, is explained in How to describe an archive in ISAD-G with AI.
How many levels to configure
The standard allows fonds, section, series, file and item, with intermediate sublevels, and the system usually offers all of them. How many are needed depends on the archive: a municipal archive with fifty years of history may need four, a historical archive with colonial documentation may need six, and a business archive can get by with two.
The most common mistake is to turn on every level from the start and find out later that the description team does not have the information to fill them in. The result is section or subseries records with empty fields, and nobody knows whether an empty field means that it does not apply or that it is pending research. That is why it is worth settling, before configuring the system, the levels the institution can actually describe, because an archive with three well-described levels is more useful than one with six half-filled ones.
Dates that are not exact
ISAD-G asks for at least one type of date, depending on the material and the level: the dates of creation of the documents or the dates of their accumulation by the creator in the conduct of its business. In a historical archive many of those dates are approximate, uncertain or contradictory, and a field that only accepts the year-month-day format is of no use for a file known to date from approximately 1850 to 1870. ISO 8601-2, which incorporates EDTF, has a notation for this (1850~/1870~), but not every system accepts it.
Three solutions work in practice, and they can be combined. The first is to have a structured date, which serves for searching and sorting, with a free-text date note beside it that explains the uncertainty. The second is to use a start date and an end date that are independent of each other, both optional. The third is to add a date qualifier (“circa,” “before,” “after,” “between”). If the system offers none of these, the team ends up putting information that should be searchable into the notes, or entering approximate dates in structured fields as if they were exact.
Searching by subject without touching the hierarchy
People who use an archive usually search by subject, while the ISAD-G hierarchy is ordered by who produced the documents. The temptation is to create “subject” levels inside the fonds to make searching easier, but those levels break provenance and become impossible to maintain when new documentation arrives that fits none of them.
What works is to keep the hierarchy for the structure of the fonds and to handle subject searching with subject, place and name access points, which can be assigned at any level. The system has to allow both separately, because one serves to preserve the order of the archive and the other to find what is in it.
The conditions of access and use area
This is the area most often left empty at the start and the one that causes the most work afterward. ISAD-G gathers here the conditions of access, the conditions of reproduction, the language and script of the documents, their physical characteristics and technical requirements, and the finding aids that already exist, and in many systems these fields are filled with a generic value copied into every record.
The problem appears when the institution has to show that it complies with transparency and freedom of information laws, when it holds documents with personal data protected by the GDPR or by the law of its country, or when it needs to control who may consult each section of the fonds. If the area was not thought through when the system was configured, the records have to be revisited one by one to add what should have been captured from the start. It is better to decide beforehand what access conditions each section or series has, and to describe them at that level so that they hold for everything below it.
What to check in the system
Beyond whether the system has fields named after the standard, there are two things worth checking before choosing or configuring it. The first is that a value common to an entire fonds or series can be described once at its level, as the ISAD-G rule of non-repetition asks, instead of being repeated in every file. The second is that producers can be described separately, with ISAAR(CPF), and linked from the description, so that the name of an institution, a person or a family is written the same way throughout the archive.
To continue the conversation
These are decisions about the design of the description, and the archivist and whoever configures the system should make them together before loading the first record. If you would like to discuss how you have resolved them at your institution, or which of them is giving you trouble, write to us at info@janium.com.