Building an Evidence Hierarchy for Pipe-Organ Attribution

Define the question. “Who built this organ?” is usually several questions in one: what was originally designed, what was actually built, what was altered later, and what is in the instrument now. The same document can be strong evidence for one of these and nearly worthless for another. A factory specification can provide strong evidence of what was contracted, but may not record later changes.

Gather the records that bear on that question. Contracts, specifications, ledgers, registers, database entries, published work lists, journals, newspaper accounts, correspondence, finding aids, dealer listings and builders’ project pages all belong in the file. Each establishes only what it states, as of the date it was written. A record describes what it documented; it does not automatically reflect later changes.

Compare the records with the instrument itself. Current inspection evidence, including the resulting photographs and reports, helps establish present configuration. Where a record and the instrument disagree, the question needs more evidence before anything is concluded.

Check independence. Trace repeated information to its origin. Three copies of one claim are not three independent confirmations; records from the same archive may still document different events or aspects of the work.

Qualify the conclusion. State what the evidence supports, which question it answers, and where corroboration was sought and not found. A negative finding, recorded with the search that produced it, is part of the result.

No single ranking of evidence classes settles every historical question. The ranking depends on what is being asked, and it should be rebuilt each time the question changes.

A current inspection helps document what an instrument contains. Control systems are the clearest case: no trade standard records control systems by builder, they are commonly made by third parties, and they are commonly replaced. What identifies the system fitted today is the console and its wiring, with the manufacturer and model read from labels or boards, not any brand-era literature. That is a method conclusion drawn from those three conditions, not a rule stated by any single authority.

A photograph can document an inspection; its date, subject and coverage determine what it shows. A photograph of an identifying mark establishes what was visible on that instrument at the moment of the image; it does not by itself show where the same builder placed marks on other instruments.

A builder’s records rank by how close each one stands to the named instrument. The record classes that survive for M. P. Möller show the pattern for the question of what was contracted: a factory specification or contract file for the instrument settles what was ordered; an opus-indexed ledger band places that contract in a period; a sales brochure establishes only what the company was offering at a date; and a patent establishes only a claimed design. That ordering comes from the limits each record class carries, it belongs to one builder’s records, and it is a pattern to reason from rather than a universal rule. Ask instead what the instrument contains today, and every one of those records drops beneath current inspection.

A factory publication can settle something as narrow as a name. A 1939 Estey trade catalog digitized on the Internet Archive from the Brattleboro Historical Society’s holdings prints the corporate style “Estey Organ Corporation,” not “Estey Organ Company.” That is what the item settles, the style Estey used on its own literature that year, and the item page alone shows nothing about the product lines inside.

A register, database record or published work list establishes that the entry exists and what it says, not what the instrument is. At least three separately maintained record sets can exist for a historic organ: the builder’s own business archive, where it survives; the Organ Historical Society’s Special Collections repository of builder business records, held at Stoneleigh and an off-site facility; and the Pipe Organ Database, a public searchable index of builder, location, images and stoplist data. As of September 2, 2026, none of the three is guaranteed to be complete for a given instrument, and whether a builder’s archive survives, or whether OHS holds records for a builder, varies from builder to builder.

Even an institutional form draws the line at the plate. As of September 1, 2026, the Organ Historical Society’s Historic Pipe Organ Awards application asks for the builder’s name to be listed as it appears on the nameboard or builder nameplate. That is an instruction to transcribe what the plate carries; the name it captures is the name on the plate.

An attribution rests on the same pairing that conservation practice relies on. The Organ Historical Society’s 2008 conservation guidelines treat reconstruction of vanished portions of an organ as appropriate where documentary and physical evidence survive and conjecture is minimal, and an attribution likewise rests on the instrument’s own records and its own fabric read together. That threshold is the OHS position; other conservation bodies arrange the same treatment vocabulary differently, and their definitions are not the same.

The same limit applies at the smallest scale. An artefact record lists a Casavant demonstrator organ of about 1925, 33 cm wide, 115 cm tall and 42 cm deep, with one octave of pipes labeled by note name and marked Melodie 8, as held by the University of Alberta physics department and acquired around the time a Casavant Memorial Pipe Organ was installed in the university’s Convocation Hall. Read as of September 2, 2026, that record does not describe the object’s internal action in detail, and its custody statement is the record’s own claim, not a present confirmation that the object is where it says.

A patent establishes a claimed design at its date, and nothing about which instruments received it. US Patent 965,896 describes a divided-chamber organ pipe that sounds the pitch of an orthodox sixteen-foot pipe from a pipe about nine and a half feet high. That is the construction as filed. The patent does not establish how many Estey instruments received such pipes, or that every short 16-foot pipe in an Estey is built that way.

A phrase in a patent or trade literature is not exclusive to one firm. “Direct electric action” names an Estey patent as well as the Wicks usage: United States Patent 1,887,076, titled “Direct electric action for pipe organs,” granted November 8, 1932 to Warren Samuel Russell and assigned to the Estey Organ Company, covers a different bipolar-magnet and pivoted-armature pallet arrangement. Shared terminology opens a line of inquiry; it does not settle whose mechanism is in the instrument.

A compiled patent index sits a further step down. A privately maintained list of William E. Haskell’s patents gives twenty-five pipe-organ patents dated between 1892 and 1928 alongside four unrelated ones, and because it carries no assignee column, whether any one of them was assigned to Estey has to be read from the patent text itself. Finding a patent and reading what it establishes is the subject of using patents to research organ mechanisms.

A project or service record settles what the firm that wrote it observed and did, at the date it wrote it. Evaluate each claim separately; a gap in one part of the history does not invalidate well-supported findings elsewhere. One Estey case, Opus 1233 of 1914, traced end to end, shows what a chain can and cannot deliver: the opus number and build year come from a secondary list derived from cost cards; the builder and fabrication dates are corroborated by physical evidence on the instrument; the specification is recorded after restoration rather than from a factory document; the alterations are known only from the restorer’s observation; and the present-condition statement is a database status line last dated 2009. The contemporaneous shop order, the one record that would independently fix the original specification, was not consulted.

That chain is complete in coverage and uneven in authority at every link. It is useful because it shows where each link came from. It is not evidence about Estey instruments generally, and it is not a model of what a finished chain should look like.

A builder nameplate establishes the name the plate carries; it does not by itself establish who manufactured the complete instrument or its components. The console is where a nameplate is most often over-read. Austin states that from the 1970s to the 1990s it installed nearly one new console a month on non-Austin organs as pneumatic consoles by other builders failed. The practical consequence is that an Austin console does not by itself establish that the organ behind it is an Austin. The figure is Austin’s own account of its work.

A nameplate can also be older than the work behind it. Letourneau’s published list of its own rebuilds and restorations names Casavant Frères as the original builder in eight of its eighteen entries and a Wadsworth-Warren-Casavant Frères composite in a ninth, so an instrument bearing a Casavant nameplate may contain Letourneau work. That list records only Letourneau’s projects, does not say which are in the United States or what was done, and is not a measure of how often Casavant instruments are rebuilt.

A maker’s mark on a component identifies the maker named on that component, not the builder of the whole instrument. Record it as evidence about that layer of the organ and keep the builder of the whole as a separate question. A photograph of the mark carries the same limit as the mark itself.

A number in a builder’s series does not by itself establish new work by that builder. Documented series can include rebuilds, and rebuilding firms can assign their own numbers to other builders’ organs. A bare opus number can also be ambiguous between firms. Both Saint-Hyacinthe builders keep independent series that begin at 1: Casavant’s runs from opus 1 in 1880 to at least opus 3920, and Letourneau’s from opus 1 in 1979 to Opus 143 as published on September 2, 2026. A reference such as “opus 63” or “opus 118” identifies nothing without the builder’s name. Within these two published series as reviewed on September 2, 2026, the overlap extended through 143.

A series can also carry more than one number for one project, or none at all. Casavant’s own timeline lists opus 1587-1588 for a 1938 installation in Lewiston, Maine and opus 3812/3813 for a 2001 installation at Central Synagogue in New York, and a rebuilder’s project record describes a 1922 Casavant acquired as opus 940, 941; the sources do not explain what the second number in each pair denotes. Casavant’s announcement of its first Inspiration Organ in the United States, installed in December 2020 in the restored Crawford Chapel at Rye Presbyterian Church, Rye, New York, identifies the instrument as II-24/9 rather than by an opus number, without saying whether an opus number was also assigned.

A construction feature seen from outside supports an attribution without settling it. The founding Austin patent and an OHS account describe universal-type instruments as having a large enclosed chamber beneath the pipework with an airtight entrance, sometimes including a two-door vestibule. Observing such an enclosure can support identification of an Austin universal-chest system; it is not conclusive, and it does not authorize an owner to open or enter the chamber. Console appearance, an enclosure and a shared phrase in the literature each narrow the field. Each is answered finally by the instrument’s own records and its own fabric, not by resemblance.

Assess a source’s provenance, specificity and reliability for the claim being made. Seek corroboration where a record is incomplete, disputed or ambiguous. The Schantz chest-and-action dates published in the company’s 150th-anniversary feature in The Diapason in 2023 come from one builder-authored piece written by three Schantz employees. Independent corroboration for them has not been found, and they should not be used alone to date an unidentified instrument. A builder’s account may document a specific historical fact, but a general account of its mechanisms does not establish what an unidentified organ contains.

Trace agreement to its source so that repeated copies of one statement are not counted as independent confirmations. Organ Clearing House listings carry the house’s own numeric catalogue identifiers, some of them in the same range as builder opus numbers, so a listing number corroborates nothing about an opus number until it has been traced to the system it belongs to. Trace each number to its origin. Copies of one claim are not independent confirmations, while different records from the same archive may document different aspects of the work.

A documented negative finding constrains what may be asserted, and it belongs in the file. The Schantz dates above show the pattern: corroboration was sought, it was not found, and the finding is recorded so that the dates stand as the builder’s statement and nothing more. A negative finding is limited by the search that produced it. It shows that corroboration was not found where it was sought, not that it does not exist, and it should be reported with that limit attached.

One Pipe Organ Database record for a Schantz instrument at First Presbyterian Church, Morristown, Tennessee, read through an unaffiliated mirror, carries opus 1116 and 1973 in its header while the stoplist attached to the same record is labeled Op. 1306 and 1974. That is a conflict between the indexed field and the document the record is built from. One record shows such a conflict can occur; it does not say how often. Conflicting records may describe different events, or one of them may simply be wrong, and the right response is to hold both positions until a further record decides between them. How to do that for opus dates, builder names and institutional records is set out in reconciling conflicting opus dates, builder names and institutional records.