OHS Pipe Organ Database Limits: Sampled Records, Attribution, and Update Boundaries

None. As read on September 1, 2026, the Pipe Organ Database makes no statement anywhere on its public site about how complete its coverage of any builder’s production is; its /about page instead describes a project whose potential depends on the continued participation and contribution of organ scholars, and its FAQ page held no questions at all. The public pages reviewed on the same date carried no published policy on editing, no completeness warning, no submissions-review procedure, and no error-correction procedure.

The silence holds at the level of a single firm as well. As read on September 2, 2026, neither /about nor any builder entity carries a per-builder completeness statement, so the relation between the database’s Casavant records and Casavant’s own production is left unstated by the publisher, and the same is true of every other builder in it.

That silence is the governing rule for everything below. With no published coverage or accuracy statement, every count, attribution, opus number, and year in the database is a fact about what contributors have entered, and never by itself a fact about what was built, where it went, or what stands today.

A builder entity’s holdings count the records filed under that entity on the day you read it, and nothing more. As displayed on September 1, 2026, the entity “Skinner Organ Co.” at /builders/7374, typed “Builder” and noted “Operated by E. M. Skinner, 1919-1932”, reported 626 Instruments, 51 Consoles and 0 Blowers. Those three figures are a holdings snapshot of the database on that date: they say how many records contributors had attached to that entity, not how many instruments, consoles, or blowers that firm made.

Because the publisher states nothing about how much of any firm’s production is recorded, a holdings figure carries no coverage claim with it. Six hundred and twenty-six could be most of the firm’s work or a fraction of it, and the database does not say which; the figure changes as records are added and is only ever true on the date it was read.

The same limit applies to the database as a whole. On September 1, 2026 its sitemap index advertised four child sitemaps on the API host’s /sitemaps/ path that together list 175,737 front-end URLs, and even that enumeration should not be treated as complete, because its class counts differed from the landing-page counters.

The database’s own counts disagree with each other, and the discrepancy runs in both directions. Against landing-page counters of 75,220 Instruments, 8,477 Builders, 61,623 Locations and 30,713 Stoplists read on September 1, 2026, the sitemaps listed 74,926 instrument URLs (294 fewer), 8,462 builder URLs (15 fewer), 61,734 location URLs (111 more) and 30,600 stoplist URLs (113 fewer). Neither figure corrects the other; both are dated readings of a live system, and neither the sitemap nor the counter is a complete enumeration.

The image counter was read twice on that date with different values. One reading of the landing page on September 1, 2026 reported 93,932 Images alongside 75,220 Instruments, 8,477 Builders, 61,623 Locations, 30,713 Stoplists and 11,154 Documents; another reading on the same date reported 93,975 images, while the /images index surface reported 93,973 and the /stoplists index surface reported 30,757. Both readings stand as recorded. The counters carry no derivation note and no date of computation, and the 11,154 Documents figure is counted although there is no documents index route and no documents entry in the sitemap.

Quote a counter with the date it was read. With no derivation note, a difference between two counts cannot be assigned to growth, deletion, or error, and none of the figures is a count of pipe organs in existence.

The builder named first on an instrument record is whoever the record is headed under, and as of September 1, 2026 that first-line attribution may identify later work, an owner, or an unknown-builder placeholder rather than the original builder. The original attribution, where the database holds one, sits on the subordinate “Originally” line; the revision chain, the work label, the notes, and the “Related Instrument Entries:” list have to be checked before the headline name is read as the maker.

A rebuild shows the pattern plainly. The single result for an opus search on Peragallo Opus 772 displays “Peragallo Pipe Organ Co. ( Opus 772, 2022)” above “Originally Reuter Organ Co. (Opus 1368, 1962)” at St. Malachy Catholic Church, Brownsburg, Indiana, 98 Ranks; the two opus numbers belong to different firms and different numbering systems on one record. A second record, /instruments/17890, reads “Schantz Organ Co. ( Opus 2197, 2001)” over “Originally Reuter Organ Co. (Opus 1368, 1963)”, so the rebuilder’s own opus is the record’s opus while the original firm’s number survives only on the subordinate line.

The headline can also hold no firm at all. Records read on September 1, 2026 showed “Unknown Builder (2026 +/-)” and “Owner (2026)” as the current builder, and one read “A. E. Schlueter Pipe Organ Company (2026)” over “Originally (Opus 9468, 1961)” with the original builder name blank while its opus number was present.

Nor does the headline reliably place work in the right corporate era. As read on September 1, 2026, a 2024 instrument, Opus 2807 at First Presbyterian Church, Colorado Springs, rebuilt from a 1958 Reuter, is displayed under the name “Austin Organ Co.”, a name the database’s own builder entity for that firm limits to 1898-1937. An older case shows the same field disagreeing with an outside account: the database attributes Opus 2000 of 1937 to “Austin Organ Co.”, while a secondary account states that Opus 2000 was the first instrument produced by the reorganized Austin Organs, Inc. Neither the database field nor the encyclopedia account is a factory record, and this page does not choose between them.

Each of these is a single record. Together they establish that a headline attribution can fail to identify the original builder in several distinct ways; they do not establish how often that happens, and none of them settles who built any particular instrument.

The record may not tell you. As of September 1, 2026, instrument records carry no structured field recording how an opus attribution was established; the only provenance signals are a courtesy credit on an attached stoplist, free-text notes, attached documents and webpages, and the contributor name and e-mail stored with the record. A database opus attribution may therefore rest on a nameplate, a contract, a published list, or on none of these, and nothing in the record’s structure says which.

When a note does name its basis, read it and weigh it. /instruments/20650 carries a note dated 23 November 2005 stating that the instrument was “Identified from the Reuter Opus List” supplied by Reuter Organ Co. in November 2005, which is a builder-supplied opus list named as the source. A note can equally show that a record rests on a third-party online report rather than on inspection, together with the editor’s unresolved uncertainty: /instruments/76016 records a July 2026 entry identified “through online information” about a 2022 rebuild with a new solid-state relay, followed by “Editor-JS [July 23, 2026]: Unknown what, if any, changes in the unification may have taken place.”

An attribution whose basis is not visible in the notes is a lead to be corroborated from an independent record class, not an attribution of record.

Both fields are contributor-entered data, and four behaviors have to be expected before either is used in a dating argument. First, the year field holds free-form and imprecise values; as read on September 1, 2026, the forms “1903ca.”, “1920’s”, “1920s”, “ca. 1993” and “2026 +/-” all appear. Second, opus numbers are not strictly chronological: an “Austin Organ Co.” result set sorted by year showed Opus 192 dated 1904 preceding Opus 132 dated 1905, and Opus 100 dated “1903ca.” preceding Opus 192. Third, an opus number is not unique within a builder: two distinct original Reuter Organ Co. instruments both carry Opus 1368, /instruments/21164 dated 1962 at “AGO Convention”, Houston, Texas, badged Not Extant, and /instruments/21165 dated 1963 at Northminster Presbyterian Church, Indianapolis, Indiana, each 25 Ranks and 3 Manuals, and each with its own downstream rebuild record by a different firm. Fourth, in the Reuter sample, the field was empty in roughly eleven percent of records: 294 of 2,751 Reuter Organ Co. records carried no opus number at all, which means no contributor supplied one, not that the instrument never had one.

An opus number on its own therefore does not point to one record or to one builder entity. A search on a builder name plus an opus number matches the number wherever it sits: the query builder=Reuter&opus=1368 returned four records on September 1, 2026, two of them rebuilds by other firms whose only Reuter Opus 1368 value is on the “Originally” line. The opus has to be read together with the builder, the year, the location, and that line before it identifies anything.

Outliers are present as well. Seven opus values recorded against Reuter Organ Co. records exceed 3,000, the highest being 50,298, 50,152 and 50,037; those are inconsistent with any plausible Reuter opus series and are best read as data-entry artifacts, an inference from their distribution rather than a statement by the database. A builder’s opus range therefore cannot be inferred from the database’s maximum value. Both Reuter proportions describe records in the database, not what Reuter produced.

The opus field also accepts alphabetic prefixes, and a prefix is data to verify, not a code to decode. As read on September 1, 2026, /instruments/70906 reads “Schantz Organ Co. (Opus A346, 2023)” over “Originally Schantz Organ Co. (Opus 346, 1958)”, so the record numbers a rebuild of Opus 346 as Opus A346. Search-indexed results also show several Möller designations with an “R-” prefix, with examples dated from the 1950s through 1980; those records were not read directly and the meaning of the prefix is unresolved, so they remain identification leads rather than a decoded numbering rule.

A record cannot be dated from the record because the interface shows no creation date, no modification date, and no history of changes on any record page or in any table view, even though a stored creation timestamp exists and is offered as a sort key. As of September 1, 2026 the instrument, image and stoplist sort controls offer “Entry Created (Oldest to Newest)” and “Entry Created (Newest to Oldest)”, which shows the value is stored, while the instrument TABLE view’s columns are ID, Instrument, Location, Address, Ranks, Manuals, Pipes and Resources, with no created or modified column. The /changes and /pending routes that would presumably expose moderation history are disallowed to automated agents and were not visited.

The address of a record looks stable, but the content at it is not fixed. On September 1, 2026 public edit controls were available on records, no visible modification date appeared, and identifiers had been reassigned when the site was rebuilt, so citations made against the previous system do not resolve: note text on /builders/3912 still contains links in the retired form “BuilderDetails.php?BuilderID=<n>”, and the legacy link for Richard O. Whitelegg carries BuilderID=6591 while his current builder entity is /builders/6686.

The only visible provenance trail is the set of dated, attributed note entries on a record, and those cover what contributors added, not what was changed. The status sentence carries no date of its own, so a record alone does not establish an instrument’s present physical condition, continued existence, playability, or current specification. As of September 2, 2026 the record for Estey Opus 1233 at Piper City Presbyterian Church, Piper City, Illinois carries the status “Extant and Playable in this location” while its most recent dated note is from 8 March 2009; that present-tense statement rests on a status line whose latest dated entry is seventeen years old and which carries no separate verification date. It does not say the organ is not playable. It says the database statement is not a dated inspection.

The consequence for citation follows directly: record the record’s numeric URL, the date read, the values relied on, and whether the record was marked “Pending Review”, because nothing on the record will reconstruct those for you later.

A correction enters through one of two published routes: the per-record contribution and edit controls, or the site-wide contact form. As read on September 1, 2026, those were the only public correction or extension pathways identified, and the pages of the database itself did not publish the review process applied after submission.

The Organ Historical Society (OHS) describes what happens next in two places, and the two accounts differ in stated timing. Its own about-the-OHS page states that any information submitted through the online contribution forms is subject to review and verification by the OHS Database Committee before it is accepted into the published record; the Committee’s membership, review criteria, and turnaround time are not stated, nor whether a submission is visible while pending. The January 2024 announcement of the current interface states only that “ALL comments and updates are sent to the editing team through a contact form and will be handled,” without saying whether that handling precedes or follows the change becoming publicly visible, who the editing team are, what standard of evidence they apply, or how long handling takes.

The two statements may describe one process at different levels of detail, or genuinely different procedures; OHS publishes no single consolidated statement resolving the ambiguity, and this page does not choose between them. What a reader can rely on is the route, not an outcome, a standard, or a timeline.

The database keeps conflicting values side by side, and a reader has to carry both rather than pick one. The clearest case is designed in: as of September 1, 2026, one physical organ that has been rebuilt is modeled as several instrument records, one per revision, each with its own numeric ID, linked by a header reading “This instrument has multiple revisions:” with “VIEW OLDER (<year>) REVISION” and “VIEW NEWER (<year>) REVISION” controls and a “Related Instrument Entries:” list in the notes area. Each revision record is a full instrument record with its own values, so figures differ along the chain on purpose: in one worked chain the 1950 Reuter record shows 6 Ranks while the ca. 1993 and 2022 records both show 12 Ranks. The size on any one record describes that revision, not the instrument’s original state or its current state generally.

Other conflicts are simply left standing. Two records disagree on the number of the same 1968 Möller rebuild contract at First Presbyterian Church, Washington, Pennsylvania: one entry is headed Opus R-786, 1968, while a 2024 note on the related entry describes the same work as R-768 in 1968. Both values remain in the database, and a prefixed number read from a database note is verified against the factory specification before it is relied on. A conflict can also sit inside one record: in a mirrored copy of one Schantz record for First Presbyterian Church, Morristown, Tennessee, the header reads “Schantz Organ Co.”, opus 1116, year 1973, while the stoplist attached to the same record is labeled “Op. 1306” and 1974, so the indexed opus field and the document the record is built from disagree. That shows record-level conflict as one possible cause of an anomalous opus and year pair; it does not say how often it occurs.

Two cautions on builder entities are declared here and explained elsewhere. As read on September 1, 2026, the structured “Active: <start> – <end>” range can be “Unknown” at one or both ends even where the same entity’s description states explicit dates: “Schantz Organ Co.” reads “Orrville, Ohio from 1873” with “Active: Unknown – Present”. An “Active: … – Present” value is not evidence that the entity currently operates: “Austin Organ Co.” shows “Active: Unknown – Present” while its note reads “Firm name from inception until 1937”, and “Richmond H. Skinner” shows “Active: 1931 – Present” against a note reading “Methuen, Massachusetts, 1931-1942.”

The builder search also renders results in two views that expose different fields, and the entity “Wicks” (/builders/7432) prints “Salt Lake City, UT 84124 UT, US” in the CARDS view against its own note “Highland, Illinois from 1906.” while its detail page shows no address line at all; both displayed values stand and neither is adopted as the firm’s location. How to search across those views and across a firm’s several name forms is the subject of searching builder name forms, successor names, and merged companies.

Two further conflicts belong to the database overview and are only named here. As read on September 1, 2026, the database’s /about page calls it “the definitive compilation of information about pipe organs in North America” while its /related page says it “accepts information about instruments across the globe”. The OHS site says the relational database was established in 2004 while the database’s /about page says the project first started in the 1950s, statements that may refer to different stages rather than a single disputed founding event.

These limits settle how much a record that exists can carry; they do not settle what a search that returns nothing means. That question has its own rule, set out in what an absent database record does and does not show. For the lookup path, the anatomy of a record, and the citation practice in full, return to how to use the OHS Pipe Organ Database.

When a database record is feeding a decision about an actual pipe organ, treat it as the opening lead and plan for inspection findings and written project documentation to carry the weight the record cannot.