Reading the Paper Trail: How Provenance Is Researched

In the Kupferstichkabinett in Berlin, the museum of prints and drawings, a researcher named Hanna Strzoda set out to check the drawings that had entered one collection between 1933 and 1945. The starting list held about 1,200 works. After removing wartime losses and works purged as “degenerate art,” about 950 remained to be investigated. A single sheet could send her through thirty files held in ten separate repositories. Roughly a third of the acquisitions had come through auctions, and the registration records for those sales yielded some 300 different sets of owners’ initials, each of which had to be traced.

The results, as reported by SPKmagazin in 2017, are sobering in their honesty. For around half of the works, the team could rule out any suspicion of confiscation through Nazi persecution. For the rest, the evidence did not yet allow a firm conclusion. That is what provenance research looks like in practice: patient, archival, and often inconclusive.

What a provenance line says

The Yale University Art Gallery describes provenance research as interdisciplinary work that traces the ownership history, or biography, of a work of art from its making to the present. The result is usually printed as a short chain of names, and the punctuation carries meaning. Yale lists owners in order, starting with the earliest. A semicolon between two owners means the work passed directly from one to the other. A full stop means the transfer was not direct, or that nobody knows whether it was. Words such as possibly, probably, reportedly and purportedly flag information that has not been confirmed. Each entry aims to give the owner’s name and life dates, the place, the method of transfer and the date.

Read this way, a provenance is less a pedigree than an argument, with its gaps marked in the grammar. The Johnson Museum of Art at Cornell is direct about the limits: a complete provenance is rare, particularly for works that are centuries old, because documents are lost, histories are forgotten and gifts go unrecorded.

Evidence on the object and on paper

Research begins with the object. The Berlin team looked at collectors’ marks, labels and auction lot numbers on the backs of the drawings. It then moved to the museum’s own records: inventory books, card indexes and acquisition files. From there it went out to databases and outside repositories, among them the Lugt database for identifying collectors, the Ferdinand Möller Archive, the Berlin State Archive and the German Federal Archives.

The paper trail has holes. Many files were lost in the war. Drawings were also documented less carefully than paintings because they were worth less on the market, so a single entry might lump several works together. One record cited in the Berlin account lists five drawings by Liebermann, sorted only into three of one subject and two still lifes, which leaves a researcher to work out which sheets those were.

For older ownership histories, the largest single resource is the Getty Provenance Index. Founded in 1974 by Burton Fredericksen in the Getty Museum’s paintings department, it now holds more than twelve million digitised, open-access resources covering the sixteenth to the twentieth centuries, with particular strength in European and American art. Its main datasets are transcribed archival inventories, which record objects in European private collections from 1520 to 1880; sales catalogues documenting works offered for sale in Europe from 1650 to 1945; and the stock books of two dealing firms, one based in Paris and one in New York. Alongside these sit files on more than 20,000 collectors and collections, records of payments to artists in Rome between 1576 and 1711, and provenance records for works in public collections in Britain and the United States. Sale catalogues matter here for a simple reason. A sale is one of the few moments in an object’s life when it is described, numbered and dated in print.

That printed moment is not always transparent. In Berlin, about a third of the drawings under review had been bought at auction, and the sale records often identified owners only by cryptic codes or initials. A lot number on the back of a sheet can tie it to a catalogue entry, but the entry may name the consignor only obliquely. Each set of initials becomes a small research project of its own, which helps explain why, in Strzoda’s account, checking a single work could take weeks or months.

The Nazi era

The years 1933 to 1945 receive particular attention because of the scale of Nazi looting. As Cornell’s museum summarises, Allied forces returned many works after the war, but thousands were lost, destroyed or sold, and wider access to historical records has since allowed survivors’ families to pursue restitution.

The international framework dates from December 1998, when the Washington Conference on Holocaust-Era Assets released eleven non-binding principles on Nazi-confiscated art. The text asks that such art be identified, that relevant records be opened to researchers, that findings be publicised so that pre-war owners or their heirs can be found, and that a just and fair solution be reached expeditiously when they are. The fourth principle speaks directly to the difficulty seen in Berlin: when deciding whether a work was confiscated, “consideration should be given to unavoidable gaps or ambiguities in the provenance in light of the passage of time and the circumstances of the Holocaust era.”

In the United States, the Association of Art Museum Directors and the American Alliance of Museums, working with the Presidential Advisory Commission on Holocaust Assets, issued guidelines in 1998 and 1999. Cornell applies the alliance’s definition of a covered object: a work created before 1946, acquired after 1932, which changed hands between 1932 and 1946 and may have been in continental Europe during that time. Cornell posts its covered objects online and cross-posts them to the alliance’s Nazi-Era Provenance Internet Portal. Yale keeps a comparable list of works made by 1945 whose histories suggest they were in Europe during the Nazi period and whose ownership in those years is not confirmed.

Antiquities and the 1970 line

For archaeological material the key date is 1970. On 14 November that year UNESCO adopted the Convention on the Means of Prohibiting and Preventing the Illicit Import, Export and Transfer of Ownership of Cultural Property. UNESCO currently counts 149 states parties. The convention asks them to build national inventories and heritage services, to introduce export certificates, to bar the import of objects stolen from museums and religious institutions, and to seize and return stolen cultural property on request. A second treaty, the 1995 UNIDROIT Convention on Stolen or Illegally Exported Cultural Objects, deals with private law questions, including privately owned objects and material from illicit excavations.

Museums have turned the convention’s year into a working rule. Cornell notes that the United States implemented the convention through the Convention on Cultural Property Implementation Act in 1983, and it cites the Association of Art Museum Directors’ 2013 guidelines: a museum normally should not acquire an ancient work unless research shows it had left the country where it was probably found before 1970, or that any later export was legal. Cornell treats 17 November 1970 as the critical date. Yale’s curators apply the same benchmark.

None of this makes provenance simple. The standards tell an institution which questions to ask and which dates matter. The answers still come one label, one sale entry and one box of files at a time, and, as the Berlin team found, about half the time they are clear enough to close a file.