The CMOA Digital Provenance Standard.

Draft Version 0.2
Published Oct 14, 2016
This supersedes version 0.1, published March 14, 2015..


At the Carnegie Museum of Art, as with many institutions, provenance is recorded in a text field within our collections management system. As such, there is no machine-readable structure, and it is cumbersome or difficult to automatically analyze the data contained within.

One of the goals of the Art Tracks project has been to generate a digital model for storing and capturing the data within provenance in a machine-readable format. This document contains our best understanding of that model, along with supporting information and examples.

There are three expressions of this standard. One is a textual model that functions as an extension of the AAM recommended provenance text. The second is as an abstract data model in JSON. The third is as a Linked Open Data, as an RDF model using the CIDOC-CRM. All three expressions are compatible and can be converted to each other.


Currently, there are two models for provenance within the cultural heritage space.

We have the standard, textual provenance document which has been used for the past two hundred-or-so years. This document describes a complex network of relationships between parties, events, locations, and objects in a concise, semi-structured format. It contains a remarkable wealth of detail about the history of the object, but it relies heavily on the expertise of the reader to be understandable. There are a large number of implicit statements, events, and entities that would need to be extracted from the data to represent the recorded knowledge of provenance in a machine-readable form.

We also have the CIDOC-CRM, a robust conceptual model for event-based modeling of the history of a object, built upon 20+ years of research into the practices of cultural heritage. CIDOC-CRM is an extremely comprehensive model capable of representing knowledge at a level of detail that far surpasses what we currently capture in our provenance documents. It is explicitly designed to support event-based knowledge like that which makes up provenance. However, the complexity of the CRM requires a deep understanding of both the field and the potential uses of the knowledge to appropriately choose a level of detail for modeling provenance.

The software in use at cultural institutions currently work with textual provenance. They contain a transcription of the text, but not a digital model of the meaning of the text. However, the tools museums use to record their collections information are robust, well-understood, and deeply embedded into daily museum practice.

To our knowledge, there are no tools currently in production that support the CIDOC-CRM model of provenance. There are several research projects underway to transition from the document-based structure of existing collections management towards an event-based model, but they are more than five years away from broad adoption in the museum community.

The Art Tracks project is designed to ease the transition between these two models. It is explicitly an interim solution, but one that is designed to work within our existing infrastructure, while preparing us to migrate to a semantic, event-based model in the future.

Textual Provenance Standard

New Sections

Traditionally, provenance consists of two semantically different sections. One is the paragraph of semi-structured text that contains a list of transfers ordered from first owner until the present day. The other is a collection of footnotes which provide unstructured content, commentary, and documentation of sources. In order for provenance written using this standard to remain legible for traditional scholars, we have not fundamentally changed the written representation of either of these sections. However, in order to add additional metadata that enhances the provenance record, we have added two additional sections to the text: an Authorities section and a Citations section.1

General format

Possibly purchased at auction by John Doe? [1910?-1995?], Boise, ID, for daughter of previous, Sally Moe, Baroness of Leeds [1940-], Pittsburgh, PA?, at “Sale of Pleasant Goods”, Christie’s, in London, England, sometime after November 5, 1975 (stock no. 10, for $1000) [1][a][b].


[1]: Purchased on the occasion of her birthday.


John Doe:
Boise, ID:
Sally Moe, Baroness of Leeds:
Pittsburgh, PA:
London, England:
Christie’s: Sale of Pleasant Goods: no record found.


[a]: See Arnau, F. (1961). The art of the faker: Three thousand years of deception. Boston: Little, Brown.
[b]: See curatorial file, Carnegie Museum of Art.

Artist as first period

The first period recorded in provenance should be the artist.

Created by Sue Smith, the artist;

This allows the provenance to cover the entire existence of the work. Within the aquisition method list, there are several phrases used to indicate this, and it is also appropriate (but not required) to add the artist as a clause after the name.

Capitalization of the first letter

If this is the first record or the previous record ends in a period, capitalize the first word. Otherwise, use standard capitalization. (if it’s a proper name, capitalize, if it’s not, don’t.)

Period Certainty

If the entire period is uncertain, begin the period phrase with Possibly , otherwise nothing.2

Acquisition Method

The method of acquisition or transfer.

Acquisition methods describe the means and terms around the transfer of custody and/or ownership of the work. This covers standard methods such as sales or gifts, but it also covers creation, destruction, loans, and other ways that artwork can be transferred. In particular, is is worth mentioning party transformations, which are transfers that happen when the legal owner of the work changes because the owner’s legal identity has changed, not because of any transfer of the work itself: for example, marriage, widowhood, or institutional consolidation.

An exhaustive list of these terms is available on our website, as well as a SKOS ontology that defines the vocabulary.

Each term has a name, but it also has a preferred phrase that signifies that term for use in written provenance. When writing provenance, this acquisition method phrase appears before the name.

Types of Parties

When reviewing these records, there are often multiple people involved: owners, custodians, agents, and auction houses. Often the owner and custodian are the same person, and agents may not always be present.

When writing this, the name that appears after the acquisition method is assumed to be the acquiring party UNLESS that name clause is followed by the word “for” followed by another name clause. In that case, the first name refers to the purchasing agent, and the second to the acquiring party.

For example, “Purchased by John Doe” assumes that John is the acquiring party, but “Purchased by John Doe for Sally Moe” assumes that Sally is the acquiring party and John is her agent.

If the sale is a confirmed commission and the artist is not assumed to have ownership of the work would also be referred to this way, for example: “Commissioned from Fritz Franz, the artist, for Sally Moe” would assume that Sally Moe was the acquiring party, and that Fritz was the artist. In rare cases where the artist is not known, you can use the alternate form “Commissioned by Sally Moe”, though this is discouraged.

Seller’s agents, typically auction houses and galleries, can be referred to by a name clause followed by “from”. If there is a specific, named event involved, that event should be called out in double quotation marks, followed by a comma.

For example, ‘Purchased by John Doe from Bitforms Gallery’ or ‘Purchased by John Doe from “Digital Works”, Bitforms Gallery’.

Party Clauses

Each individual mentioned follows the same form.

The name of the acquiring party.

Claude Monet

Each party begins with the name of that party. Use the form of that name most appropriate for the provenance at that point. It it typical to use the name as it is recorded in the primary source where the transaction is documented, or the form of the name that would have be used at the time that the acquisition took place


Queen Elizabeth II

Often parties have titles that are associated with them. These should be included when relevant. However, they MUST be included as part of the name in the authority record for the party. See the section on Authority Records for more information about this.


Vincent van Gogh’s brother, Theo van Gogh.

Often it can be clarifying to state the relationship between the acquiring party and another party. Given the myriad ways that humans can be related, this is often better addressed through a comprehensive model in an authority file. Thus, we have chosen to not allow any possible relationship here—instead, we allow only a single, familial relationship. If there are multiple relationships, as occasionally happens, we prioritize marriage over any other relationship.

The allowed relationships are:

  • “brother”
  • “sister”
  • “sibling”
  • “mother”
  • “father”
  • “parent”
  • “son”
  • “daughter”
  • “child”
  • “grandchild”
  • “grandparent”
  • “nephew”
  • “niece”
  • “uncle”
  • “aunt”
  • “wife”
  • “husband”
  • “spouse”
  • “relative”

These clauses directly precede the acquiring party’s name, followed by a possessive 's, followed by the relationship and a comma.

Artist Clause

Jackson Pollock, the artist.

The only non-familial clause allowed is “the artist”, which should appear ONLY in the first period of a provenance. Occasionally, an artist may re-acquire a work, but in those cases it should not be explicitly mentioned that they are the artist. This clause directly follows the name, separated by a comma.

Name Certainty

If the name is uncertain, follow the name with a question mark.

George Strait?


Possibly George Strait

Using a question mark and not a certainty word prevents confusion around the question of which aspect is certain of uncertain, the party or the entire record.


If the party’s name has multiple initials, use a space between them.

J. J. Abrams


J.J. Abrams

Unknown Parties

If the party is not known, use Unknown party, Unknown collector, or something similar. Don’t leave it blank. Also, don’t associate unknown people with a generic authority file for all unknown people—leave the authority section as unknown unless there’s a need to create a specific unknown party.

If the acquiring party is a marriage, a partnership, a purchase in shares, or any other form of joint ownership, there must be an entity that holds legal ownership of the work. For instance:

Mr & Mrs. Marshall Fields III

This is the marriage of these two people, not either of these people themselves. Any dates or locations should refer to the marriage, and a provenance event (typically a Transformation of Party) should occur if a marriage ends in either death or divorce.

Life Dates

Francis Bacon [1561-1626]

For a party, the relevant dates are their birth and death dates. For organizations or legal entities such as a marriage or partnership, these are the dates between which the organization was in operation or the partnership was in effect.

These are written as years, enclosed within brackets and separated by a dash.


If the year is unknown, or the party is still alive, exclude that date, but leave the dash.

[1880-] or [-1955]

If either of the years are uncertain, follow that date with a question mark.


For years BCE,



A location associated with the owner of the artwork at the time of acquisition.

Locations in provenance have traditionally been used as a way to uniquely identify individuals. Because of this, the nature of the location traditionally recorded in provenance is ambiguous—it doesn’t have specific semantic meaning. This standard uses Linked Data unique identifiers as a technique for uniquely identifying individuals, but having these locations present still provides benefits to humans reading the written records.

Because of this, we have chosen to consider the location that directly follows any party’s name to be a location associated with that person. In general, it would be appropriate to choose a location most appropriate for the object under discussion, but that is not a requirement, just a suggestion.

For the specific location where an acquisition takes place, we’ve added the ability to have a location preceded by in. This location is explicitly the location where the acquisition took place, and is often associated with an auction house or sale. None of these locations are required, but it can be helpful if the location is known to explicitly call it out using the in form because it helps with location-based research.

For example:

Purchased by John Doe, London, England;

Would mean that London is a location primarily associated with John, even if the artwork was purchased while he was on holiday in Spain.

Purchased by John Doe in Barcelona, Spain;

Would explicitly place that purchase in Spain, but would not associate John with Barcelona (beyond his potential presence to purchase the work).

Purchased by John Doe, London, England, in Barcelona, Spain;

Would explicitly state both facts. This is often not required, but can be done if it is clarifying in some way.

Writing Specific Locations

If the location is uncertain, use a question mark following the last word of the location.

Glasgow, Scotland?

If the location is in the US, record the city and the state (in zip code two letter abbreviated form, without periods, and with a comma between the city and the state.)

Boston, MA

If the location is outside of the US, record the city, the province or state (if required to prevent ambiguous locations), and the country. Each of these should be separated by a comma.

Paris, France

If the location is in a named, notable building, record that building.

Highclere Castle, Hampshire, England

Do not record street names or street numbers within provenance locations.

If the location has changed names over time, the name of the location is typically recorded the name of the location at the time of the acquisition. i.e. Leningrad, not St. Petersburg, if the acquisition took place between 1924 and 1991. All locations are typically recorded in English, and should be written using the most recognizable or appropriate form. Locations are included in the Authority section, which helps deal with the ambiguity of names for specific places. Think of the words as being a way to describe the location as it would be referred to in art history, and the Authority record referring to a specific physical location.

Named Events

from “Carnegie International”, Carnegie Museum of Art, Pittsburgh, PA from “Northbrook Sale” through John Doe from “Sale of Important Works” through Paul Durand

While many transactions are between private individuals and have no documentation, other come through participation in significant or named events such as gallery shows, auctions, or large sales. Also, many of the sales were facilitated by a seller’s agent. To record this information, we’ve added a section that we refer to as the “Named Event” section. This section begins with either the word “at” or the word “through”, depending on which makes grammatical sense.

This is optionally followed by the name of the event in quotation marks.

from “Northbrook Sale” from “Venice Biennial”

The name of the event should always appear in the Authority Section. The name of the event can be followed by the agent, auction house, or other person or organization who arranged, facilitated, or hosted the event. This person is referred to as the “seller’s agent”, and their format follows that of any other Party record.

from “Northbrook Sale”, Christie’s, London, England.

Note that this location is again the location associated with the actor, not the location of the sale.

Transfer Location

In cases where the location of the transaction is explicitly known, it can be recorded using the word “in”.

Purchased by Jane Doe in Boise, ID. John Doe, from “Northbrook Sale”, Christies, in London, England.

This is the location of the event, not that of any of the parties involved. When it is present, it can often be redundant to have both the location of the parties and the location of the event, and our recommendation in that cases to only use the location of the event, since it will be reasonably clear that the party is associated with the event. For historical reasons, however, it remains useful to be able to record one or both.

Period of Ownership

A description of the time period the work was owned by the party.

before 1996 Until some time after the 15th Century 1995 until 2006

For dates, the preferred forms, in increasing order of precision, are:

  • 19th century
  • 1990s
  • 1990
  • October 1990
  • October 11, 1990

Years before 1 CE are specified with BCE. Years between 1 CE and 999 CE are written with an explicit CE, years after 999 CE are written without an explicit era.

5th century BCE 566 CE August 1, 1990

Uncertainty is indicated with a question mark following the last part of the date phrase

until October 1885?

Years are recorded in Gregorian form, using October 4, 1582 as the first day of the Gregorian calendar. For years earlier than 1582, use the proleptic Gregorian Calendar.

The year immediately preceding 1 CE is 1 BCE. There is no year zero.

When recording periods, we always record the interval of possibility for the acquisition. This can be done by recording the known date with appropriate precision:

October 1990

implies that the work was acquired on a specific day between October 1, 1990 and October 31, 1990. Often, though, we need to record something where either the beginning or the end of that interval is unknown, or the period described by the interval is not able to be expressed as a single imprecise date, and so we use a list of connecting phrases to explicitly describe the beginning and end of the period.

after 1990 by 1990 or before 1990 between 1990 and 1995

These forms allow us to express both a precision and the bounds of the interval. So between 1990 and 1995 represents the interval between January 1, 1990 and December 31, 1995.

Typically, we do not need to record the deacquisition of the work within provenance events, because that transfer is recorded as an acquisition within the following period. However, sometimes while we know when a particular collector de-acquired the work, we do not know explicitly to whom a work was transferred, and rather than create an “unknown collector” with no data other than the date, we prefer to record the deacquisition within the period of the known collector. In these cases, we use the word until to indicate that the period being described is that of the deacquisition.

until 1990 until at least 1990 until before 1990 until between 1990 and 1995 1990 until 1995

Please refer to the Art Tracks date standard for a more complete description and explanation of of connecting phrases.

Purchase Information

It can often be useful to record information about the details of a sale. There are many possible details, but we feel that the associated stock or lot number and the monetary amount associated with the transaction are both sufficiently regular that they could be captured, and sufficiently useful as structured data that we should capture them.

(stock no. 55512, for $100,000) (for 5 shillings, three pence) (lot 25)

Both the stock or lot number and the amount should be enclosed within parentheses, and the amount should be preceded with the word “for”. If both are present, use a comma between them.

Footnote & Citation Markers

Footnotes are notes, additional context, or any other information needed to clarify the record that cannot be included in one of the previous sections. Citations are a reference to an external document that is the source of the information provided for this period.

Footnotes are indicated by a number contained inside brackets at the end of a record, and Citations are indicated by a letter contained within brackets at the end of the record. They should be the last thing within the clause, preceding the completing punctuation. There should only be a single footnote per-period; if multiple notes are needed, they can be multiple sentences within the footnote. There can be as many citations per-period as needed, one for each clarifying document.

John Doe, 1990 [2]. John Doe, 1990 [2][a]. John Doe, 1990 [a][b]. John Doe, 1990 [a]. John Doe, 1990 [1][a][b].

For more information about the content of footnotes or citations, see the respective sections of this document.

Closing Punctuation

If the artwork is known to be a direct transfer, use a semicolon at the end of the period of ownership.

Michel Buble, 1990;

If the artwork is not known to be a direct transfer, use a period.

Michael Jackson, 1986.

If uncertain, assume that the transfer is not direct.

Notes Section

The footnote section follows directly after the full Provenance paragraph.


[1]. This is a note. [2]. This is also a note.

Each footnote section begins with a blank line, followed by the string ‘Notes:’ on its own line, followed by another blank line, to indicate that the footnote section is beginning. Each individual footnote is indicated by a number within brackets, followed by a period, followed by the text of the note.

Footnotes can contain as much text as needed—they are for clarity and people, not computers. They are designed to capture additional content, stories, clarifying notes, and information directly related to the period that they are referenced from.

Even if a footenote could apply to multiple periods within the provenance, you should create a duplicate footnote that contains the same text for each period. That way, unintentional changes of one owner do not accidentally propagate to other sections.

If there are no notes for a provenance, omit this section entirely.

Authority Section

The Authority section references Linked Open Data URIs for named individuals, locations, and events.

This section begins with a new line, followed by Authorities:, followed by a blank line. It should directly follow the Notes: section if there are notes, otherwise it should follow the provenance paragraph.


John Doe: see
Jimmy James: No record found.

This section should contain a line for each Actor, Location, and Named Event (collectively, the Entities) mentioned in the provenance text. Each name (or appellation) used to refer to an entity within a provenance text should only be included once.

The appellation should exactly match how it is represented in the provenance text, including capitalization and punctuation3. If titles, honorifics, prefixes, suffixes, or other text is included as part of the name, they should be included in this section. Each appellation should be followed by a colon, followed by one or more spaces.4

If a Linked Data URI from an Authority File is known for the entity, it should be included, preceded by the word see. For Authority Files that differentiate between the entity and the record, this should be the URI of the record.5

John Doe: see

These SHOULD be included in the order they appear in the provenance text.

Entities that do not currently have any known URI should be represented using “No record found.” instead of a URI.

Jimmy James: No record found.

Citation Section

The Citation section should always be the last section of the document. The section begins with a blank line, followed by the string ‘Citations:’ on its own line, followed by another blank line, to indicate that the citation section is beginning. Each individual citation each note is indicated by a letter within brackets, followed by a period, followed by the text of the citation.


[b]. C.L. Sulzberger, A Long Row of Candles, Macmillan, 1969, p. 8. [f]. See, cat. 376.

Currently, there is no defined format for citations.

Linked Data

Design Principles

As we have been developing the linked data model for this standard, we have had three guiding principles:

The first is to maintain the integrity of the CIDOC-CRM, and to use the predicates and classes that it (and its extensions) provide whenever possible.
The second is to consider that the major purpose of modeling provenance as structured data is to enhance the usability of the provenance—particularly the searchability of the provenance. At every step, we would like to enhance the ability of a user to locate specific events in the history of an object, and we aim to provide as many hooks into the provenance as possible.

Third, we aim to capture completely the information that makes up the written provenance text. Provenance data is extremely concise, and to omit any word from the provenance is to omit a detail that the provenance researcher deemed important.

The structure and model presented below is our best attempt to create a standard that accomplishes our mission while abiding to the principles as closely as possible.

Levels of Provenance

Linked data is still a relatively new concept in museums, and very few museums are prepared to embrace it to the level that fully modeling provenance in all its specificity entails. To ‘shorten the on-ramp’, we have defined four levels for the digital encoding of provenance. These levels have been designed to function as strict supersets—each level can also function as each level below. Each level provides a more complete graph of information, additional search abilities, and a more nuanced understanding of the provided provenance text.

Level One

When we model a provenance text we are modeling two separate but related structures. One is the written text that makes up the traditional provenance record, which we refer to as the “provenance text”. The other is the actual events and parties that the provenance text describes, which we refer to as “provenance events”. We maintain and model both of these structures in Linked Data because they both contain nuances that are difficult to capture in the other’s form.

Level One describes a provenance text in which the events described within the document are not modeled in any fashion, nor are the people, places, nor any other entity. This level is intended as model for provenance within an institution where the provenance is represented a single text field and where no additional work is done to understand or interpret the provenance. The RDF model at this level merely models the existence of a provenance text, the existence of an event, and relates both to the work described.

At this level, full text search is possible, but it is not possible to facet the information in any way.

This model is made up of two core CRM entities: a E31 Document and an E7 Activity. The document is the provenance text itself, and the activity records the ongoing activities of the owners of an object that involve transfers of ownership and custody6. Both of these entities are assigned a E55 Type of aat:300055863, which is the AAT term for “provenance”.7

Level Two

Level Two models builds upon Level One by associating the provenance text and event with the people, places, and URLs present within the text. This level is designed to accommodate software that is capable of processing the provenance text but doesn’t have special knowledge of provenance, such Named Entity Recognition or string matching against authority files.

At this level, it becomes possible to connect artwork to the people and places that are mentioned in the provenance. It also enables search for provenances using alternate names (when linked to authority files), and to use URLs to link to digitized primary sources. Individual transactions and dates are not modeled at this level.

Level Three

Level Three models the individual transactions as specific provenance events, but considers there to only be a single event representing the transaction between periods of ownership. These transactions follow each other in time, but are not casually related. To implement this level, we assume the existence software capable of parsing individual acquisition or transfer events within provenance. All software developed as part of Art Tracks is designed to work at this level.

At Level Three, it becomes possible to analyze and search for specific provenance transactions, and relate those transaction to people and locations. It also becomes possible to search for transactions that occurred within specific space-time volumes, such as “Continental Europe between 1932 and 1946”.

Level Four (unimplemented)

At Level Four, we explicitly model the sub-events that make up each individual transaction. These events are casually related: each one is motivated by and dependent on the event that directly precedes it. For example, rather than modeling a loan as a single event, you would model the transfer of custody to the trucking company, followed by a transfer to the art handlers at the new location, followed by a transfer to the exhibiting institution. This level of provenance is not currently recorded in traditional provenance texts, but is the next logical step for integrating provenance into a larger model of the history of movement, ownership, and custody of art.8


I. Why Text?

Given the underlying structure of the provenance information, we considered several possibilities for the canonical reference for this information. To do so, we developed the following set of guidelines for describing our goals:

  1. Provenance must be represented as unambiguous structured data.
  2. There should only be a single source of truth for provenance data.
  3. The provenance information should follow industry standards when available.
  4. Provenance should balance simplicity, capturing nuance, and unambiguity.
  5. Information should not be discarded.
  6. The system should be designed for usability.
  7. The system should be as simple as possible.
  8. The system should be technologically feasible.
  9. Not all information can be unambiguous.
  10. Not all information can be structured.
  11. The current industry standard for provenance records is a paragraph of standardized text.

Designing for usability, as well as designing for a single source of truth, means that there exists a conflict between the traditional provenance record as a block of text and the requirements of structured data.

One option was to ignore the requirement for a single source of truth, and have both structured and unstructured representations of provenance data. While technically the easiest option, it is very easy for the two sources of data to diverge, and there would be no method to detect such divergences. Enforcing data constraints with policy tends to be less than optimal, and without a detection mechanism, experience has shown that one of the two systems will become untrusted.

Given that two versions of the data is non-optimal, the second option of converting provenance to structured data and treating the textual record as a produced artifact of the data is tempting. This would allow the data to be the single source of truth. This is also a technically appealing solution, since it allows the use of existing technological tools (such as relational databases) and simple data modeling constraints.

The major issues with this option revolve around usability. Given that most museums do not have structured provenance data, in order for the structured data to be the single source of truth, all existing records would have to be converted immediately, or a transition period would have to be enforced where either no new information is added, or all information is added to both places. Additionally, extensive training would have to be provided to all staff on how to read and write provenance information.

Another constraint is the flexibility of existing Collection Management Systems. Either the CMS has to be extended to handle the new data, or there has to be a new tool added and maintained to store and handle the data.

Given these issues, the decision was made that the primary description of the structured data should remain the paragraph of provenance text. This has the benefit of working with existing tooling, and also allowing a gradual conversion of records without problems; however, this presents its own issues. The largest of these issues is that text is not traditionally an excellent source of unambiguity, and the structure of either English or the AAM guidelines is not sufficient to guarantee parsability.

II. Example Provenance as Text

This is the provenance for The New Testament, owned by the Freer|Sackler. The provenance of this object has been reformatted to use the CMOA Provenance Standard. This is being used as a demonstration of syntax and should not be considered an authoritative record.

Mrs. Serunian [1][a]; by inheritance to Dr. H. H. Serunian, son of previous, Worcester, Massachusetts [b]; purchased by Freer Gallery of Art, 1937.


[1]. Excerpt from the Curatorial Remark 1 in the object record: “Dr. Serunian inherited the book from his mother who was an Assyrian and brought it with her to this country. Dr. Serunian says that the covers were removed from the book to make it more easily portable when his mother’s relatives escaped from Kurdistan to Persia in 1916.”


Mrs. Serunian: see
Dr. H. H. Serunian: see
Worcester, Massachusetts: see
Freer Gallery of Art: see


[a]. Curatorial Remark 1 in the object record.
[b]. Curatorial Remark 1 in the object record.

III. Example Provenance as RDF

This is an example of the same provenance record as above, but represented as RDF in turtle. Each of the levels represented here build on each other. For conciseness, we do not duplicate information at each level; assume all of the statements at previous levels exist at the following ones, as well.

Level One

# The Art object (only the essential bits for this)
freer_object:10173 a crm:E22_Man-Made_Object ;
  rdfs:label "The New Testament";
  crm:P52_has_current_owner freer_constituent:3326.

# The Owner is:
freer_constituent:3326 a crm:E40_Legal_Body;
  rdfs:label "Freer Gallery of Art".

The provenance text itself is documented as a E31 Document. This is the traditional, written text of a provenance statement, typically written following a museum’s standard form.

freer_object:10173_provenance a crm:E31_Document ;

  # We use `P2 has type` to distinguish this text from any other
  # document about the object.
  crm:P2_has_type aat:300055863;  # "provenance"

  # We use `P70 documents` to refer to period covered by the documented history
  # of ownership for the object.  This is what the content of provenance is.
  crm:P70_documents freer_object:10173_provenance_period ;

  # We use `P129 is about` to refer to the object that is the subject of the
  # the provenance.  This differs from `P70 documents` in that it is what the
  # provenance refers to, not what it is documenting.  
  crm:P129_is_about freer_object:10173 ;

  # We use `P3 has note` to document the actual text of the provenance.
  crm:P3_has_note "Mrs. Serunian [1][a]; by inheritance to Dr. H. H. Serunian, son of previous, Worcester, Massachusetts [b]; purchased by Freer Gallery of Art, 1937.\n\nNotes:\n\n[1]. Excerpt from the Curatorial Remark 1 in the object record: “Dr. Serunian inherited the book from his mother who was an Assyrian and brought it with her to this country. Dr. Serunian says that the covers were removed from the book to make it more easily portable when his mother's relatives escaped from Kurdistan to Persia in 1916.” \n\nAuthorities:\n\nMrs. Serunian:            see ...\nDr. H. H. Serunian:       see ...\nWorcester, Massachusetts: see ...\nFreer Gallery of Art:     see ...\n\nCitations:\n\n[a]. Curatorial Remark 1 in the object record.\n[b]. Curatorial Remark 1 in the object record." ;
  # We have a `rdfs:label` to provide information to a human user who wants to 
  # understand what this entity represents. 
  rdfs:label "The provenance of Freer|Sackler object 10173, 'The New Testament'." .

The complete sequence of events that make up the provenance are documented as a E7 Activity. This event contains all of the events in the object’s history that relate to the ownership and custody of the object. It stretches from the moment the work is considered to exist until either the current moment or the documented destruction of the work. At this level there are no individually modeled acquisition events.

freer_object:10173_provenance_period a crm:E7_Activity ;
  crm:P2_has_type aat:300055863 ; # Provenance
  crm:P16_used_specific_object freer:10173 ;
  crm:P4_has_time-span freer_object:10173_provenance_period_timespan;
  rdfs:label "The transfers of ownership and custody of Freer|Sackler object 10173, 'The New Testament'.".

freer_object:10173_provenance_period_timespan a crm:E52_Time-Span;
  crm:P82_at_some_time_within "1213/open";
  dct:conformsTo <>;
  rdfs:label "1213 until the current day".

Level Two

To the document, we add references to all locations mentioned. At this level, we can’t distinguish between the various types of locations mentioned, so we associate them all with the document.

freer_object:10173_provenance a crm:E31_Document ;
  crm:P67_refers_to geonames:4956199.               #Worcester, Massachusetts

We also associate the parties and URLs found within the document with the overall provenance period.

freer_object:10173_provenance_period a crm:E7_Activity ;
  crm:P11_had_participant freer_constituent:14160;  #Mrs. Serunian
  crm:P11_had_participant freer_constituent:14160;  #Dr. H. H. Serunian
  crm:P11_had_participant freer_constituent:3326;   #Freer Gallery of Art
  crm:P70i_is_documented_in <>

Level Three

freer_object:10173 a crm:E22_Man-Made_Object ;
  crm:P50_has_current_keeper freer_constituent:3326;

The List of people involved are:

# The Current Owner
freer_constituent:3326 a crm:E40_Legal_Body;
  crm:P53_has_former_or_current_location geonames:4140980. #Washington, DC.

# The first owner:
freer_constituent:14160 a crm:E21_Person;
  rdfs:label "Mrs. Serunian";
  crm:P53_has_former_or_current_location geonames:4956199. #Worcester, Massachusetts

# The second owner:
freer_constituent:4274 a crm:E21_Person;
  rdfs:label "Dr. H. H. Serunian";
  ulan:1511_child_of freer_constituent:14160;
  crm:P53_has_former_or_current_location geonames:4956199. #Worcester, Massachusetts

This is the timeline of significant events in the history of this manuscript

freer_object:10173_provenance_period a crm:E4_Period ;
  crm:P9_consists_of freer_object:10173_initial_ownership ;
  crm:P9_consists_of _:10173_prov-1 ;
  crm:P9_consists_of _:10173_prov-2 ;
  crm:P9_consists_of _:10173_prov-3 .

The initial ownership by an unknown person.

freer_object:10173_initial_ownership a crm:E8_Acquisition, crm:E10_Transfer_of_Custody;
  crm:P2_has_type mprov_acq:creation;
  crm:P30_custody_transferred_of freer_object:10173.
  crm:P24_acquired_title_of freer_object:10173.

When Ms. Serunian. acquired the manuscript

_:10173_prov_1 a crm:E8_Acquisition, crm:E10_Transfer_of_Custody;
  crm:P2_has_type mprov_acq:acquisition;

  crm:P30_custody_transferred_of freer_object:10173;
  crm:P24_acquired_title_of    freer_object:10173;
  crm:P22_transferred_title_to freer_constituent:14160;
  crm:P29_custody_received_by  freer_constituent:14160;
  crm:P3_has_note "Mrs. Serunian";
  crm:P129i_is_subject_of _:10173_prov_1_footnote.

_:10173_prov_1_footnote a crm:E33_Linguistic_Object;
  crm:P2_has_type aat:300265639;
  crm:P3_has_note "Excerpt from the Curatorial Remark 1 in the object record: “Dr. Serunian inherited the book from his mother who was an Assyrian and brought it with her to this country. Dr. Serunian says that the covers were removed from the book to make it more easily portable when his mother's relatives escaped from Kurdistan to Persia in 1916.” See Curatorial Remark 1 in the object record.".

When Ms. Serunian. acquired the manuscript

_:10173_prov_2 a crm:E8_Acquisition, crm:E10_Transfer_of_Custody;
  crm:P2_has_type mprov_acq:bequest;

  crm:P120i_occurs_after _:10173_prov_1;

  crm:P30_custody_transferred_of freer_object:10173;
  crm:P24_acquired_title_of    freer_object:10173;

  crm:P23_transferred_title_from freer_constituent:14160;
  crm:P28_custody_surrendered_by freer_constituent:14160;

  crm:P22_transferred_title_to freer_constituent:4274;
  crm:P29_custody_received_by  freer_constituent:4274;

  crm:P3_has_note " To 1937\nDr. H. H. Serunian, Worcester, Massachusetts; by inheritance from his mother, Mrs. Serunian.".
  crm:P129i_is_subject_of _:10173_prov_2_footnote.

_:10173_prov_2_footnote a crm:E33_Linguistic_Object;
  crm:P2_has_type aat:300265639;
  crm:P3_has_note "Curatorial Remark 1 in the object record.".

The third transfer:

_:10173_prov_3 a crm:E8_Acquisition, crm:E10_Transfer_of_Custody;
  crm:P2_has_type mprov_acq:purchase;

  crm:P120i_occurs_after _:10173_prov_2;

  crm:P30_custody_transferred_of freer_object:10173;
  crm:P24_acquired_title_of    freer_object:10173;

  crm:P23_transferred_title_from freer_constituent:4274;
  crm:P28_custody_surrendered_by freer_constituent:4274;

  crm:P22_transferred_title_to freer_constituent:3326;
  crm:P29_custody_received_by  freer_constituent:3326;

  crm:P3_has_note "From 1937\nFreer Gallery of Art, purchased from Dr. H. H. Serunian, Worcester, Massachusetts.";
  crm:P129i_is_subject_of _:10173_prov_3_footnote;

  crm:P4_has_time-span _:10173_prov_3_timespan.
_:10173_prov_3_timespan a crm:E52_Time-Span;
  crm:P82a_begin_of_the_begin "1937";
  rdfs:label "From 1937".

_:10173_prov_3_footnote a crm:E33_Linguistic_Object;
  crm:P2_has_type aat:300265639;
  crm:P3_has_note "See note 2.".

IV. Example Provenance as JSON

Level One

provenance: {
  object_id: "",
  provenance_text: "Mrs. Serunian [1][a]; by inheritance to Dr. H. H. Serunian, her son, Worcester, Massachusetts [b]; purchased by Freer Gallery of Art, 1937.\n\nNotes:\n\n[1]. Excerpt from the Curatorial Remark 1 in the object record: “Dr. Serunian inherited the book from his mother who was an Assyrian and brought it with her to this country. Dr. Serunian says that the covers were removed from the book to make it more easily portable when his mother's relatives escaped from Kurdistan to Persia in 1916.” \n\nAuthorities:\n\nMrs. Serunian:            see ...\nDr. H. H. Serunian:       see ...\nWorcester, Massachusetts: see ...\nFreer Gallery of Art:     see ...\n\nCitations:\n\n[a]. Curatorial Remark 1 in the object record.\n[b]. Curatorial Remark 1 in the object record.",
  existence_interval: "1213/open"

Level Two

provenance: {
  object_id: "",
  provenance_text: "Mrs. Serunian [1][a]; by inheritance to Dr. H. H. Serunian, her son, Worcester, Massachusetts [b]; purchased by Freer Gallery of Art, 1937.\n\nNotes:\n\n[1]. Excerpt from the Curatorial Remark 1 in the object record: “Dr. Serunian inherited the book from his mother who was an Assyrian and brought it with her to this country. Dr. Serunian says that the covers were removed from the book to make it more easily portable when his mother's relatives escaped from Kurdistan to Persia in 1916.” \n\nAuthorities:\n\nMrs. Serunian:            see ...\nDr. H. H. Serunian:       see ...\nWorcester, Massachusetts: see ...\nFreer Gallery of Art:     see ...\n\nCitations:\n\n[a]. Curatorial Remark 1 in the object record.\n[b]. Curatorial Remark 1 in the object record.",
  existence_interval: "1213/open"

Level Three

provenance: {
  object_id: "",
  provenance_text: "Mrs. Serunian [1][a]; by inheritance to Dr. H. H. Serunian, her son, Worcester, Massachusetts [b]; purchased by Freer Gallery of Art, 1937.\n\nNotes:\n\n[1]. Excerpt from the Curatorial Remark 1 in the object record: “Dr. Serunian inherited the book from his mother who was an Assyrian and brought it with her to this country. Dr. Serunian says that the covers were removed from the book to make it more easily portable when his mother's relatives escaped from Kurdistan to Persia in 1916.” \n\nAuthorities:\n\nMrs. Serunian:            see ...\nDr. H. H. Serunian:       see ...\nWorcester, Massachusetts: see ...\nFreer Gallery of Art:     see ...\n\nCitations:\n\n[a]. Curatorial Remark 1 in the object record.\n[b]. Curatorial Remark 1 in the object record.",
  existence_interval: "1213/open",

  periods: [
      period_certainty: true,
      original_text: "asdadadasd",
      footnote: "blah blah blah",
      citations: ["http://...", "...", "asdasd"],
      owner: {
        owner_id: "ulan:12345",
        owner_certainty: "true",
        owner_relationship_to_previous: "her daugher",
        owner_residence_id: "geonames:12345"
        owner_residence_certainty: "true"
      custodian: {
        custodian_id: "ulan:12345",
        custodian_certainty: "true",
        custodian_relationship_to_previous: "her daugher",
        custodian_residence_id: "geonames:12345"
        custodian_residence_certainty: "true"
      transfer: {
        direct_transfer: true,
        location_certainty: true,
        location_id: "geonames:1234",
        method_id: "prov_acq:gift",
        botb: "1995",
        bote: "1995",
        eotb: "1995",
        bote: "1995",
        botb_certainty: true,
        bote_certainty: true,
        eotb_certainty: true,
        bote_certainty: true       
      agent: {
        agent_id: "ulan:12346",
        agent_certainty: "true",
        agent_residence_certainty: "true",
        agent_residence_id: "geonames:12346"
      purchase: {
        price: 1234,
        currency: "lira",
        lot:  "l. 12345",
        sale_id: "getty:12345",

  agents: [{
    id: "ulan:12345",
    name: "Jane Doe",
    birth: "1995",
    birth_certainty: "true",
    death: ""
    death_certainty: "true",
    id: "ulan:12346",
    name: "John Doe",
    birth: "1800",
    birth_certainty: "true",
    death: "1900"
    death_certainty: "true",

  places: [{
    id: "geonames:12345",
    name: "London, England"
    id: "geonames:12346",
    name: "Pittsburgh, PA"

  events: [{
    id: "getty:12345",
    venue: "ulan:12345",
    location: "geonames:12345",
    title: "Northbrook Sale"
  1. Both of these were designed to be human-readable without specialized software and should to be able to be written and maintained by humans. They also are designed to provide additional context not only to the computer, but to a human reader.

  2. We have decided not to include a vocabulary or “levels” of uncertainty, because we don’t have a good system for determining how these levels would be specified. We also differentiate between “Probably”, which refers to the entire period, and “?”, which we use to indicate that individual parts of the record might be questionable.

  3. This resolves the problem of allowing the user to maintain nuance in their choice of appellation for individuals while also providing an unambiguous ID for that individual. “Jane Doe” can be explicitly linked to the same URI as “Jane Pendleton”, a maiden name, maintaining context while disambiguating identity.

  4. We allow one or more spaces so that the LOD identifiers can be aligned when presented in a text interface using a monospace font. This is not a requirement, but is permitted.

  5. Best practices for Linked Data also assert that these URIs should be dereferenceable using HTML, which means that a human researcher can gain immediate context for the people and places represented by the IDs by copying and pasting them into a browser.

  6. We choose to use E7 Activity instead of E4 Period to model this series of activities for two reasons. First, because it allows us to associate actors and objects directly with the Activity, which becomes useful at later levels, and second because we feel that it the class that most accurately represents the agency of human activity that defines this period.

  7. We feel that using AAT to define our types is a suitable best practice, and one that we will continue throughout this model whenever possible. Our feeling is that the implicit assignment of AAT vocabulary terms the class E55 Type through the use of crm:P2_has_type is an acceptable practice despite lack of explicit statement. An alternate model of crm:P2_has_type _:type. _:type skos:exactMatch _:aat_term . is possible, but does not add any additional abilities to our model while adding complexity, and as such is discouraged.

  8. We do not fully define this level, and none of the Art Tracks tools are capable of modeling or mapping these events. It may seem unusual to define a level that do not implement or fully define, however we understand the utility of this level of specificity. While we believe there would be benefits to modeling at this level, none of the institutions that we are working with have data at this level of granularity. However, as would also like to make sure that the model we are developing does not preclude the future existence of this data, we have designed the model to accommodate it when it exists.

Table of Contents