Key idea & worked example
Source dependence means that apparently separate records may derive from the same underlying evidence. Corroboration needs attention to that shared origin.
Count independent evidence, not the number of places where a claim is repeated.See the worked example →On this page
The MODAVIS Pipe Organ Dataset (POD) connects instrument identities with source accounts, people, places, specifications and documented events. Its release structure provides a fixed basis for reuse and comparison.
| POD 1.6.0 | Published corpus |
|---|---|
| Canonical organs | 215,769 |
| Actors | 11,778 |
| Places | 203,831 |
| Source collections | 25 |
| Virtual instruments | 897 |
These are release-specific corpus counts, not a census of all surviving instruments. Coverage depends on the contributing collections, their documentation and the reconciliation work embodied in this release.
A record is a meeting point for accounts
The corpus connects instrument identities with attributed descriptions. Several sources can discuss the same organ, and one source can document more than one state. That makes the distinction between an instrument and an account central to reuse. A convenient instrument summary should remain traceable to the statements from which it was assembled.
Do three copies of a value count as three measurements?
Source dependence means that apparently separate records may derive from the same underlying evidence. Corroboration needs attention to that shared origin.
- Original sourceOne recorded observationThe evidence begins at an identified source.is copied or transformed
- Derivative copiesSeveral displaysSpreadsheets, exports and graphs may repeat it.retains the same lineage
- Evidence countOne lineageRepetition does not create independent measurement.
Work through the example 3 STEPS
Imagine the Cuntz value 1281 mm appearing in a workbook, an exported JSON observation and a catalogue summary based on that export. Only the first two are the concrete record types used here; the catalogue repetition is illustrative.
Trace each occurrence
The JSON observation points to the workbook cell. The hypothetical summary points to that observation. These records form a derivation chain.
Count the evidence correctly
Three appearances of the value do not establish three independent measurement sessions. Repeating an extraction in another file adds accessibility, not an independent observation.
Seek a genuinely different check
An independently documented measurement or suitable calibrated survey could support comparison. Its own conditions and uncertainty would still need inspection.
- Cuntz Positiv · VAO 0.5.0-rc.2 ↗
Published manifest: identities, measurement observations, realization digests, profiles and rights. Exact examples checked against this release.
- MODAVIS Ontology Network 0.1.0 ↗
Identity, state, configuration, evidence, assertions, context and provenance. Teaching examples explain its distinctions without inventing new normative properties.
This example supplies no additional independent measurement of the Cuntz pipe.
The useful distinction. Count independent evidence, not the number of places where a claim is repeated.
What the numbers count
| Unit | Meaning for analysis |
|---|---|
| Instrument identity | A canonical point of reference for an organ within the released corpus. |
| Source membership | A connection between a record and one contributing source collection. |
| Specification account | A particular attributed description of an instrument’s configuration. |
| Stop occurrence | A stop recorded within a source account; not necessarily a unique physical stop. |
| Documented event | An attributed historical occurrence, distinct from a processing operation. |
These levels should not be added together or substituted for one another. Two accounts can describe the same physical stop, and an account may preserve only part of a specification. A larger number of stop occurrences can therefore reflect richer documentation rather than a larger population of instruments or components.
What a release makes stable
A release fixes an interpretable state of the corpus, including its methods, distributions and known limits. It provides a common basis for comparison even while a public interface or development workspace changes. Keep that release identity with any extracted table, graph or derived dataset; the values alone do not record the conditions under which they were produced.