Optical Metrology in Luxury Horology: Chasing the Rolex 18K 2N Perfect Match

Have you ever gone to the optometrist for an eye exam? When they put the lenses in front of your eyes and go, “This one, or this one? This one, or this one?” for 15 minutes?

That is exactly what I have been doing with 18K gold pieces for weeks. I have been deep in the plating process for a while now, testing sample after sample, tweaking the chemistry, and trying to hit a flawless visual match for Rolex’s proprietary 18K yellow gold (Swiss 2N). It has been said I can be a wee bit obsessive. I started getting results that were incredibly close, close enough that I could hold a plated piece next to a genuine component under normal light and could not tell which was which 19 out of 20 times.

But my visual assessment was no longer cutting it in the microscopic margins I was working within. That was when I realized I needed a different method to measure the colours I was looking at.

The Capture Rig: Eliminating the Guesswork

If I was going to measure colour accurately, how could I capture it accurately? Phone cameras are horrific at this. They auto-adjust white balance, apply heavy processing, and lie to your eyes.

I built a controlled capture environment instead. Here is what I used and why it matters:

• Camera: Canon DSLR shooting in RAW. RAW files capture the full sensor data with no in-camera processing, no white balance baked in, no sharpening, no colour enhancement.
• Lighting: 4000 K arch light, >95 CRI (Colour Rendering Index), arching 180 degrees over the workbench. 4000 K gives a neutral white reference, and >95 CRI renders colour with near-perfect fidelity without the spiky spectrums that distort warm gold tones.
• Background: 18% neutral gray card. This gives a known mid-tone luminance reference point in every frame.
• Environment: A completely blacked-out room. No ambient light contamination.

The result is a capture pipeline that produces repeatable, accurate colour data. A difference in the pictures taken is real, not an artifact of the camera.

Enter CIE LAB: Measuring What Eyes Cannot

Staring at photos on a monitor still comes down to subjective judgment. I needed hard numbers.

Enter CIE LAB. This is a colour space designed specifically to model human perception, also used in ISO 8654, the international standard for gold alloy colour classification in jewellery and watchmaking. It has three axes:

• L* (Lightness): 0 (pure black) to 100 (pure white). A bright 24K yellow vs. a rich, warm vintage gold, a lot of that character lives in L*.
• a* (Green ↔ Red): Negative shifts toward green, positive shifts toward red. This is where you see the difference between a cold, lemony gold and a warmer, rosier tone.
• b* (Blue ↔ Yellow): Gold lives heavily in the positive b* territory. The magnitude tells you how saturated that yellow character is.

Genuine Rolex gold has a specific L*, a*, b* signature. My tests were designed to move the colours I was plating through that coordinate space to land on that exact target.

By calculating the Euclidean distance between my plating tests and the genuine target in this space, I get ΔE (Delta E):

ΔE = √[(ΔL*)² + (Δa*)² + (Δb*)²]

In precision manufacturing, a ΔE of 1.0 to 1.5 is the threshold below which trained observers cannot reliably distinguish parts under controlled conditions. Below 1.0, trained observers cannot reliably distinguish two samples in direct comparison under controlled conditions.

The Grind: Sample by Sample

Going through my 55+ samples, here is exactly what the numbers taught me.

The Curing Trap

There is a cruel trap in electroplating: hitting your colour straight out of the bath means absolutely nothing.

I pulled a sample, measured a sub-1.0 ΔE, and celebrated. Two days later, I put it back under the rig to shoot photos, and the numbers were wrong. The a* had crept positive (warmer), and the L* had dropped (duller). The colour shifted while sitting on the bench doing absolutely nothing.

It came down to three unavoidable physical reactions that occur in the first 48 to 72 hours:

1. Oxide film formation. Trace cobalt reacting with ambient oxygen creates a microscopic, warm-tinted oxide film.
2. Lattice relaxation. The violently compressed crystal lattice slowly relaxes, shifting the microgeometry and scattering light differently.
3. Hydrogen effusion. Trapped hydrogen gas bleeds out of the metal, leaving microscopic voids that alter surface reflectance.

I had to bake a 72-hour cure into my protocol to pre-compensate for this shift. Once I did, I locked in a reproducible, stable ΔE of 0.30.

Why Your Eyes Lie

A camera sensor resolves colour differences that the human visual system smooths over. When light hits a camera lens, ISP algorithms, auto-white balance, and web compression alter the raw data. By the time that image reaches your uncalibrated phone screen, fighting ambient room light, your brain is decoding heavily distorted information.

You cannot eyeball 2N Champagne. Scratches, lighting angles, and room reflections will lie to you every single time. By combining mechanical agitation, precise codeposition of base metals, and rigorous CIELAB colour analysis, I engineered the luck out of the process.

What’s Next for K-Plating

Now that the Swiss 2N 18K-colour hard-gold plating is solved and reproducible, the focus shifts to expanding the colour library. White gold and palladium finishes are already in production at 394 HK. Since this article was written, deposit hardness has been formally measured: 234 HK for the cobalt-hardened gold and 394 HK for palladium, both by Knoop microindentation to ASTM E384. The ongoing targets are broadening the range of verified CIELAB targets across the full spectrum of factory gold tones, all over bare steel or intact PVD with no intentionally applied nickel underlayer.

Same methodology. More colours. Higher standards.

Measurement note: The ΔE values in this article were derived from controlled DSLR RAW captures analysed with spectro software on PC, under the standardized conditions described above. Production colour acceptance is now measured by a dedicated hardware spectrophotometer under the conditions published on the colour method page, with an acceptance threshold of ΔE < 1.5.