Peter Paul Biro
Forensic Studies in Art



Works of art often preserve the fingerprints of the artist who created them. Such crucial evidence can go unnoticed even by connoisseurs, art experts and conservators. If present, such evidence may prove valuable in addressing questions about authorship. We have used this approach since 1984.

For example, fingerprints left in clay and wax models by sculptors have helped pinpoint connections to the master, assistant, or the workshop. Similarly, fingerprints impressed in still fresh paint, or left with paint from the artist's fingers can also lead to answers.

But, detecting fingerprints is not a straightforward matter. It requires a full understanding of the conservation history of the object first, then applying the right techniques and tools, and the appropriate methodology. Read more

Conventional setup vs. scanner

On the left: Detail of a painting, taken with what is widely used equipment by police forensic examiners.
On the right: Image taken of the identical area. Our hyper-spectral scanner reveals a clear fingerprint hiding in just 10 nanometers of the spectrum. The conventional equipment reveals nothing in the same area.

DNA and art

Joann Friedrich Miescher

Johann Friedrich Miescher

While most think DNA analysis is modern 21st century CSI science, it is important to recall that Johann Friedrich Miescher already synthesized DNA in the late 1800s. He called it “nuclein” and considered it a new and unique part of cell structure. Watson and Crick, some 75 five years later in the 20th century, proposed the double helix structure that eventually led to today’s understanding of the genetic role of DNA. Many others contributed to the research in this field but one could hazard a guess that none of them foresaw that one day DNA may provide clues to the authorship of a work of art.

Crick and Watson

James Watson and Francis Crick

Our genetic fingerprints are left behind in so many ways. A single cell, a single hair can provide information that may lead to identification. Today, the oldest surviving DNA sample is 419 million years old, from one of the earliest forms of life on Earth – a saltwater bacterium.

In our work with art, we started considering DNA evidence about a decade ago hoping to shed light on attribution issues. The idea first came when we discovered human hair embedded in original paint on canvas. Then, some years later we conducted tests on various other hair samples to see if DNA can indeed be extracted from them. Some worked, some did not. The principal issues were contamination and degradation of the samples over time.

Model of DNA strand

Model of DNA strand

The small amount of sample available for analysis was also an issue. Human hair contains little genetic material unless the hair follicle still remains. But, it does contain mtDNA (mitochondrial DNA) that holds part of the genetic code inherited from the mother. With the relentless progress in technology, these issues tend to shrink away, opening up new possibilities. We continue to work on finding new ways of discovering genetic material in works of art. While it may not always be the painter’s, it may arguably be someone very close (an assistant or family member) and, like fingerprints, such evidence may establish connections and provide leads.

We are no longer alone in attempting to use this approach. Just recently a debated Van Gogh attribution is awaiting results from DNA analyses. A strand of human hair found embedded in paint is being examined for presence of DNA and, if successful, could be compared to DNA samples from descendants.

Human hair embedded in paint

Human hair embedded in paint. The paint was still fresh when it was deposited and in likelihood belonged to the artist

Multispectral imaging

Multispectral imaging has been around for decades in various forms like special cameras on spacecraft, but much more recently in the examination of works of art. The usefulness of the technique cannot be underestimated as results can shed light on what happened to a work of art even over the span of centuries.

Our multispectral imaging equipment  was designed  in-house but with a significant difference and purpose. It was originally concieved to fill a need to explore large surfaces for the presence of fingerprints. However, fingerprints do not always appear in the visible part of the spectrum. Taking images with sufficient resolution in ultraviolet and infrared light on large surfaces is far too cumbersome and time-consuming – hence prohibitively expensive. In addition, not knowing where a fingerprint may turn up, the entire surface would have to be imaged in close-up. This kind of macro mapping can produce thousands of images even if using only a few broadband filters. 

The key to that bottleneck was automation.

Essentially, the scanner is a highly precise motion control device that can carry payloads up to 60 pounds or more. The payload may be a camera, a spectrophotometer, or an x-ray unit – whatever cluster of sensors is needed for the task. Typically, in our work it carries two cameras, one chilled CCD imager and one true-color camera. In addition, it also carries light sources for imaging from 178 to 1100 nanometers, including fluorescence imaging. Full integration of motion control, camera, filter changer, and lighting control in a computer-automated loop now permits performing tasks that would have been impossible before. The new ability to use large numbers of narrow band-pass filters also permit spectral mapping through creating what is called a spectral image cube of a painting.

Instead of relying on a single sensor to produce a multispectral series of one view (which limits resolution significantly) we opted for a three-axis motion control device that carries a platform on which a number of instruments are deployed. The motion control moves and positions a detector with a precision better than the size of a pixel on a CCD sensor. This permits digital tiling of a great number of macro images, providing unprecedented resolution, plus reflectance spectral information.

The key component is the fully integrated software controlling all functions of the device – motion, sensors, and lighting. In addition, our software programming permits full automation; remote control, web access and sharing of data between geographically separate users. Much like astronomical observatories on high mountains or the Hubble Space Telescope, the device can be used remotely and data retrieved. It is indeed a tool for multiple concurrent users such as experts examining a painting from remote locations. Presently, we are developing Raman and imaging Raman applications that will be utilized from the platform and be capable of examining very large areas, typically up to 5x8 feet, and derive non-destructive chemical information form a painting. In essence, approaching the ability to know what each brushstroke is made of without ever touching the painting.

Infrared 1064 nm
Near infrared 980 nm
Near infrared 851 nm
Visible light 654 nm
Visible light 607 nm
Near infrared at 1064 nm Near infrared at 980 nm Near infrared at 851 nm Visible light at 651 nm Visible light at 607 nm

Visible light 561 nm
Visible light 471 nm
UV 405nm
UV 371nm
UV 350 nm
Visible light at 561 nm Visible light at 471 nm Ultraviolet at 405 nm Ultraviolet at 371 nm Ultraviolet at 350 nm
350 nm detail
Ultraviolet at 350 nm, detail revealing a fingerprint, perfectly useful for comparison, left in original surface. Such evidence may lead to identifying the artist even when the painting's provenance is unknown.
An excellent example demonstrating that finding fingerprints on a work of art is not a straightforward process as one cannot tell in advance in what range of the spectrum they may be detected.