Each entry names the original document and the publisher that holds it.
How the register is arranged
The register is divided into three sections. Earth
& Environment holds six entries on geology, energy resources, the physical earth and
the environment. Life Sciences & Health holds
eight, on biology, ecology, medicine and public health.
Engineering & Standards holds six, on
engineering, computing, measurement and technical standards.
The division is a finding aid and not a claim about the subjects. Several entries could sit
in a different section without doing violence to anything: a toxic cyanobacterial bloom is as
much environmental as biological, a suppressed pistol is as much history as engineering, and a
procurement dispute over a transport aircraft is policy conducted in engineering terms. In
each case the entry is filed where a reader would look for it first, and each section page
says openly where its own boundary was drawn by judgement. When the judgement does not match
yours, the index lists every page on the site on one page,
regardless of section.
The sections are uneven in size and that is not an oversight either. The register follows the
subjects that other work actually cited, so its shape is determined by what researchers
reached for rather than by a plan to cover a field evenly. A section with eight entries does
not indicate a priority; it indicates eight documents.
What an entry contains
Every entry opens with a block of five fields, and those five are the ones a citation needs.
Subject gives the section. Source names the publisher that
holds the original document. Reference gives the original's own identifier
where it carried one — for the Amu Darya Basin entry, for instance, that is USGS Bulletin
2201-H. Pages gives the extent of the original, and Entry
gives this register's own reference for the record.
The extent is more useful than it looks. A citation to page 214 of a document that runs to
twenty-eight pages is pointing at a different document, or at a different edition of the same
one, and knowing that before you start looking saves the search that was never going to
work.
The body that follows is a written summary of the subject, organised under its own headings,
and it is a summary of the subject rather than a summary of the document. The distinction
matters: it means the entry is useful for deciding whether the original is the thing you need,
and it is not a substitute for reading the original if it is.
Most entries carry one drawn plate. A plate appears where the content is numeric, structural
or sequential and a diagram states it more precisely than a sentence can — a stratigraphic
section, a set of transform offsets drawn to relative scale, a temperature band, an effect
size against the sample it needs. Where the content has no such shape, there is no plate,
because a figure that restates a paragraph is padding with a border around it. Each section
page says which of its entries carry one and which deliberately do not.
Each entry ends with two things: a short section on where the primary material sits, and
three questions with direct answers. The questions are the ones that recur in the literature
on that subject and tend to be answered inconsistently — the difference between a transform
fault and a fracture zone, or why a range extension for a toxin-producing organism carries
more weight than a single record usually would.
Addresses that carry a record but no summary
Not every address in this register has a written entry behind it. A large number carry a
record instead: the reference, the name the document was published under, how many referring
domains cite the address, and a list of some of the pages that do.
That is a smaller thing than an entry and it is deliberately not dressed up as a larger one.
It exists because a reference pointing at an address should resolve to something that says
what was there, rather than to a dead end — and because the citing pages are often the fastest
route to the document itself. Someone who cited a document usually said something about it,
and a bibliography that includes it will frequently give the publisher, the year and the full
title that the address alone does not.
These records are marked noindex and are not listed in the sitemap. They are
addressable, not promoted: they answer a reference that is already pointing at them and are
not offered to a search engine as content.
Why some subjects have an entry and others a record
The difference is not importance. It is whether the subject could be established well enough
to write about honestly.
A written entry exists where the subject of the original document is recoverable with
confidence — where what the document was about is clear enough from how it was referenced that
a summary can be written and checked, rather than guessed at. The twenty entries here are the
cases where that held. Each one covers a subject that is documented independently in the
published literature, so the summary can be written from that literature and is verifiable
against it, and the entry names the body that published the original so the claim can be
taken further.
Where the subject is only a name — a title with no context, an author with no field, a
reference with nothing attached — there is a record and no summary. Writing a summary in that
position would mean inventing a subject and attributing it to somebody else's document, which
is worse than useless to a reader chasing a citation: it would look like information and
behave like noise. A record that says plainly what is known and stops there is the more
honest artefact, and for the reader's actual purpose it is also the more useful one, because
the citing pages it lists are a real lead.
Reading a “cited by” count
Records carry a count of referring domains, and it is worth being precise about what that
number is and is not.
It counts distinct domains that reference the address, not individual references and
not citations in the bibliographic sense. Ten domains may represent ten separate pieces of
work or one piece of work syndicated ten times; one domain may carry twenty references from a
single research group. The count measures how widely an address was referenced across the
web, which is a different quantity from how often a document was cited, and a very different
one from how much the document mattered.
What it is genuinely good for is triage. A higher count means more independent places to look
for a fuller description of the same document, which is exactly what you want when the address
itself has stopped being enough — and that is the use this register puts it to.
The publishers, and what each kind does to a citation
The originals behind this register sit with seven quite different kinds of body, and the kind
is the single best predictor of whether a citation will still resolve in ten years.
National surveys and agencies — the United States Geological Survey, ADEME,
the Institutul Geologic al României — publish in numbered series with stable designations.
These are the easiest originals to find years later, with one caveat: a national agency
publishes in the language of its administration, and that is where a citation trail most often
goes cold for a reader working in another one. A search that returns nothing in English is
frequently a search conducted in the wrong language rather than evidence that the document
does not exist.
Intergovernmental bodies — UNESCO through its Intergovernmental
Oceanographic Commission, the Commonwealth Secretariat — publish technical series and meeting
documents with their own numbering. Stable, but distributed through the organisation rather
than through academic indexes, so the reliable search key is the series and number, or for a
declaration, the meeting and its date.
Standards organisations — ISO, and ANSI through AIIM — issue documents whose
designations survive revision. That makes a designation without a year or edition ambiguous in
a way it would not be anywhere else: the same number can identify several materially different
documents published decades apart.
Commercial academic publishers — Springer, MDPI — attach a DOI that resolves
permanently. The record is stable even where the text is not free, and a DOI is the one
identifier worth trying before anything else, because it is designed for exactly this
failure.
Learned societies and not-for-profit publishers — the European Society of
Arachnology, CSIRO Publishing acting for the Australian Healthcare and Hospitals Association —
publish journals that are fully citable but indexed less uniformly than the commercial houses.
Present in the major indexes, usually; guaranteed, no.
University publishers and institutes — Universitas Brawijaya, Universitas
Lampung, the Universidad Autónoma del Estado de Hidalgo — publish genuinely findable work
through institutional and national channels rather than international ones. These are the
citations most often written off as untraceable, and they are usually the ones where the
document is sitting in plain sight on a university repository in Indonesian or Spanish.
Trade periodicals and independent technical authors — Small Arms
Review, and self-published technical writing — sit outside the academic indexing systems
entirely. The work is real; locating it means going to the publication rather than to a
database, and a failed database search says nothing at all about it.
Finding the original: what actually works
The order to try things in is not the order most people use. Search the identifier
first. A DOI, a bulletin number, a series designation or a standard number is
unambiguous and machine-readable, and it is the only part of a citation that does not decay.
Search the title second, and expect it to be imperfect — titles are
transcribed loosely into bibliographies, shortened, and translated inconsistently between
languages, so the same document appears under several strings that no exact-match search will
connect. Search the author last, because a common surname in a large field
returns everything and settles nothing.
When a document exists in more than one form — a preprint and a journal version, a draft
standard and its published edition, an agency report and a translation, a conference paper and
the article that grew out of it — treat them as different documents, because that is what they
are. Their content may be close, their pagination will not match, and a citation to a page
number is a citation to one specific version of one specific issue. This is the most common
reason a page reference appears to point at the wrong passage.
If the identifier and the title both fail, go to the citing pages rather than to another
search engine. Where a record here lists the domains that reference an address, those
bibliographies are the practical route: they were written by people who had the document in
front of them and frequently record the publisher, the year and the full title. A document
that cannot be found from its own address can often be found from somebody else's footnote.
Finally, this register records what each document was and who published it. It does not
reproduce the documents themselves, and it does not hold copies of them. What it offers is the
part a broken citation actually needs: enough to identify the original with confidence, and
the name of the body that has it.
