Una Vilana Windisch — A New Chapter in Dürer Authentication

In 2023, Art Recognition was approached by a European private collector with a delicate portrait on vellum titled Una Vilana Windisch, attributed to Albrecht Dürer. The work depicts a humble peasant woman and is dated 1505. The work drew immediate interest, given that a nearly identical sketch, rendered on paper and officially attributed to Dürer, is preserved in the British Museum.

The existence of two versions of a composition is not unusual in Dürer’s body of work. Known for his technical precision, Dürer often created preparatory studies before completing a final version. But what made this case exceptional was the near-identical alignment of the two works: once superimposed, the vellum and the paper versions matched almost perfectly.

Over the years, Una Vilana Windisch had been subjected to extensive investigation. From art historical studies dating back to the 1970s to cutting-edge conservation techniques and material analysis, the artwork had travelled a long path of inquiry. 

A forensic handwriting expert compared the monogram, handwriting, and date to authenticated Dürer letters, concluding that the signature was a genuine, spontaneous autograph. Technical analysis further supported the work’s age: infrared imaging revealed an underdrawing, UV light showed no modern retouching, and material testing confirmed the use of iron-gall ink on lime-treated vellum—typical of the early 1500s.

Despite these findings, the art world remained divided. Some leading connoisseurs, including Theodor Musper and Herbert Peter, supported the attribution and even included the piece in the 1989 Berlin exhibition Dasein und Vision. Yet others, like Dr. John Rowlands, curator at the British Museum, dismissed it, asserting that only the paper version in the museum’s collection was authentic.

In the midst of this scholarly standoff, the collector turned to Art Recognition for a new layer of insight—AI authentication. Using a dataset of 144 genuine Dürer artworks (ink, chalk, and charcoal) and an equal number of forgeries, imitations, and even synthetic images, the AI was trained to recognize the artist’s unique style.

The result? Una Vilana Windisch was classified as authentic with 82.2 % probability. Interestingly, the British Museum’s paper version—long considered the “true” Dürer—was also analyzed and received a slightly lower authenticity score. This unexpected twist added a new dimension to the story and captured the attention of the art world, eventually being covered in The Art Newspaper.

This case exemplifies how AI is not replacing traditional connoisseurship, but complementing it—offering a data-driven layer of verification in complex attributions. With no universal standard in the art market, the addition of AI may be exactly what’s needed to break long-standing debates around authenticity.

< Back