Authentication isn’t a button you press. It’s a process — one that requires science, context, and – dare we say it – authentic experience. That’s why we built ArtDiscovery, with its multimodal approach that demands advanced materials analysis, imaging, AI, and research. The result? Certainty instead of doubt.
Our team works out of labs in London and New York, using cutting-edge instrumentation and the largest private pigment database in the world — including over 4,000 samples spanning ancient to contemporary painting.
We run molecular and elemental analysis, benchmark results against our internal datasets, and use them to filter out impossibilities. We provide best-in-class radiocarbon dating of canvas, paper, and wood — and we pioneered the dating of oil binders. We’re the only firm in the U.S. offering dendrochronology for panel paintings. And yes, we built AI systems that can tell Canaletto from Bellotto — tools originally designed for the Webb Space Telescope, now repurposed for less celestial concerns.
Each technique contributes to the authentication. That’s why we combine science with deep art historical knowledge and provenance research. When all three align, we can say with confidence: this work is by the artist, or it isn’t.
Below, you’ll find a breakdown of the techniques we use — what they do, and why they matter.
With scanning electron microscopy and energy dispersive X-ray spectroscopy (SEM-EDX), we examine paint samples at the microscopic level — identifying individual pigment grains and pinpointing the elements they contain. This allows us to match materials with known historical pigments and build a precise picture of the work’s composition.
Raman is highly effective for identifying pigments and materials, especially organic pigments invented in the nineteenth and twentieth centuries. Combined with our proprietary pigment database — built over 16 years — it gives us high-confidence results rooted in comparative data.
By examining the optical properties of pigments — including particle shape, morphology, and birefringence — we can determine how they were manufactured. These characteristics offer valuable clues about when and where the materials were produced, helping us date the artwork with greater accuracy.
X-ray fluorescence (XRF) lets us identify the elemental composition of pigments and materials. Traditional point analysis gives us data from a single spot — useful, but limited. That’s why we also use MA-XRF scanning, which produces full pigment distribution maps across the surface. It turns isolated results into a complete visual and chemical landscape of the work.
Cross-section analysis allows us to study the full stratigraphy of a painting: the layering of ground, paint, and varnish. This reveals the artist’s process and, sometimes, features specific to that artist.
ATR-FTIR lets us go beyond pigments and into the binders, varnishes, and organics that hold a painting together. It gives us insight into what was used — egg, oil, resin, wax — and sometimes how it was altered. Understanding the chemistry of a binder can make or break a timeline.
We carbon date canvas, paper, wood — and even oil binder, using protocols developed in-house. Our pre-treatment methods are designed specifically for artworks, not archaeological material — removing contamination and increasing precision. For many works, it’s the most objective way to pin down when it could have been made.
Panel paintings are dated through tree-ring analysis, matching growth patterns to established chronologies. This allows us to determine the earliest possible felling date of the wood — and by extension, the earliest possible moment the artwork could have been created. We’re the only US-based lab offering this service in-house.
X-rays penetrate paint and ground layers, revealing what lies beneath the surface. Denser materials like lead white or metallic pigments appear bright, helping us trace an artist’s working method or spot later alterations. It’s not just what you see — it’s what you don’t — that can change the story of a painting.
Infrared reflectography (IRR) allows us to see through the upper paint layers to detect underdrawings, carbon-based sketching materials, and pentimenti — visual evidence of the artist planning and executing their composition. These glimpses into the artist’s process can support attribution and uncover workshop practices.
UV light causes certain varnishes, overpaints, and restoration materials to fluoresce — revealing what's been added, retouched, or altered. It’s particularly useful for identifying conservation history and distinguishing original surfaces from interventions.
Some techniques we keep to ourselves. 20% of our lab time goes into original research — developing new methods, mapping artist-specific techniques, and building the datasets that keep us ahead. Forgers adapt fast — so we don’t give them a blueprint.
While scientific analysis can confirm whether materials align with a given time period and whether techniques are consistent with an artist, it is not always conclusive in matters of attribution.
To address this, we turned to astronomy – where machine learning, combined with traditional observation, detects complex patterns in light diffusion and chemical composition to pinpoint the origins of stars. Similarly, our AI-driven approach isolates artistic characteristics undetectable by the human eye.
A forger must not only use historically accurate materials but also replicate the unique pressure and technique of an artist’s brushwork – an almost impossible task. Pictology, ArtDiscovery’s proprietary AI system, analyzes these microscopic details, providing a scientific foundation for digital connoisseurship.
By leveraging high-quality training data, our algorithms detect subtle nuances in brushstrokes, composition, and other artistic elements that traditional methods might overlook. However, AI is not a standalone solution – it is one tool in our multi-modal approach, complementing scientific analysis, provenance research and expert connoisseurship.
Each attribute – including brushstroke length, width, and curvature, as well as the spatial relationships between compositional elements – is mapped for every painting in the dataset. These data points form a multidimensional distribution unique to the artist, creating a visual fingerprint of their technique. By comparing a painting against these expected values, we can assess whether it aligns with the artist’s known body of work, providing a data-driven approach to attribution.