The colour was never
in the stone.

Notes from a session at the World Diamond Congress, Singapore, July 2026, "The Unsolved Mystery of Color in Diamonds," with Roy Safit of the Fancy Color Research Foundation, cutter and former Harry Winston CEO Ray Simpson, and Janak Mistry of DiaDNA, moderated by Richa Goyal Sikri.

Everything the trade knows about grading colourless diamonds, the four Cs, breaks down almost immediately once colour enters the picture. Fancy colour diamonds, Safit argued at the opening of the session, behave more like coloured gemstones than like their own colourless cousins.

The most disorienting idea on the panel came from Mistry. Colour in a transparent object depends on how long light stays trapped inside it before escaping. Cut a coloured rough down small enough and light passes through so quickly the stone reads as essentially colourless, regardless of how saturated the raw material actually was. Cut the same material larger, with proportions engineered so light bounces repeatedly before exiting, and the colour deepens dramatically. He calls this optical size, distinct from carat weight.

That was the clearest illustration Mistry gave: a rough that GIA examination found actually contained both type IIa and type IIb material in one crystal. Cut as two separate stones, they came out two completely different colours, discovered only because someone chose to analyse the stone scientifically before cutting rather than after.

Both cutters on the panel were clear that planning software has limits. Simpson described judging final colour as coming down to human stereo vision looking directly at a stone, something no photograph or spectrometer fully captures. Different cutting centres, New York, Antwerp, Surat, have each developed their own inherited style for fancy colour material, almost none of it written down. He mentioned sourcing old estate pieces during his time at Harry Winston specifically because some historic cutting techniques had simply been lost with the people who knew them.

The session description had promised a closing announcement with a profound impact on value development in this segment. That turned out to be a new partnership between the Fancy Color Research Foundation and a rough diamond planning platform, Cutwise. FCRF has run a quarterly pricing index for more than two decades, built from dealer-submitted pricing data across colour categories. Through the new integration, a manufacturer can scan a rough stone and see, for each possible cutting outcome, an estimated real dollar market value before any cutting decision is made. One example given: a rough planned as a ten carat cushion came out looking heavy and overbuilt; recutting as an eight and a half carat heart shape would have been worth roughly 40% more, a comparison only visible because the value estimate sat alongside the plan itself.

A dealer in the audience asked the question that mattered most: does formal, granular pricing data risk doing to fancy colour diamonds what standardised grading arguably did to colourless ones, replacing negotiated market judgment with a number that starts to dictate price rather than reflect it. Safit's answer was that FCRF deliberately does not connect buyers to specific stones or broker deals, even when long-standing members ask directly, on the basis that doing so would cross from providing information into interfering with trade.

A separate question asked whether the same rarity-based framing FCRF uses for fancy colour, describing a stone as one that surfaces only once every five or six years, could be extended to colourless diamonds as an argument against lab-grown competition. Safit said the analytics could technically support it, but the historical database does not exist for colourless stones the way it does for fancy colour, where nearly every stone of consequence already passes through GIA. It is the same gap we wrote about in our note on the ethics of the lab diamond conversation: natural diamonds have a rarity story lab-grown can never claim, but rarity only works as an argument when someone has actually built the data to prove it.