A code is not a colour

Gold is a direction, not a value. The N codes exist because the industry needed shorthand for that direction, and they work well as a conversation and badly as a specification. A buyer who orders 2N and receives something that reads pale has not been supplied the wrong code; they have been supplied the code they asked for.

Where the codes come from

The N scale describes the colour of a gold alloy in the gold–silver–copper system. Copper warms the colour, silver cools it, and the numbered steps run from pale to red. The system was developed for karat gold and was adopted by PVD target makers because it gave buyers a familiar word for a coating that contains no gold at all.

Two things the table cannot show, and both matter. The steps are not equal: in most target sets the visual distance between 2N and 3N is larger than between 3N and 4N. And the names overlap. Champagne, honey and pale gold are marketing terms with no fixed position, so one supplier's champagne can sit near 1.5N and another's near 2.5N.

What the code does not specify

An N code names a family, and four consequences follow.

Two 2N golds are not the same colour. Two target suppliers, or two chambers running the same target, produce visibly different results from the same nominal code. The code sets the intent, not the point on the scale.

It says nothing about thickness. A 0.3 µm coating and a 0.8 µm coating of the same target read differently. Thickness — typically 0.2–0.5 µm for a gold tone, measured to ISO 3497 — is a separate specification, and it is frequently the first number reduced when a quote is squeezed.

It says nothing about gold content, and it must never be used as though it did. A gold PVD coating on stainless steel is a ceramic layer such as titanium nitride, not a gold alloy. The FTC's Jewelry Guides require coated goods to be described as coated, so a listing that says 2N gold without the word finish becomes a compliance problem the first time a marketplace or a customs officer reads it; the text of the Guides sits at 16 CFR Part 23. For pieces sold in the EU, the REACH nickel release limits — tested to EN 1811 on the finished item — apply underneath whatever colour is specified.

It does not travel. Approving a 2N sample from one supplier and ordering 2N from another is not a colour match. It is a new approval that happens to share a name.

The numeric layer

Colour is measurable, and measurement is what turns a code into a specification. A spectrophotometer returns the CIE L*a*b* coordinates: lightness, the red–green axis and the yellow–blue axis. For gold tones, b* does most of the work, a* separates a warm yellow from a rose, and L* separates a bright gold from a dull one.

The tolerance is expressed as ΔE, the distance between two colours in that space. As a working guide, ΔE under 1 is close to the limit of a trained eye, ΔE of 1 to 2 is acceptable for pieces sold as a pair, ΔE of 2 to 3 is visible side by side but reads as one family at a distance, and above 3 the pieces read as different metals when adjacent. Set the band against the use: two halves of a pair need the tightest tolerance, while two SKUs that never share a photograph can carry a wider one.

Locking a tone

Five controls, agreed in this order:

1. A physical master, two copies, signed and dated, kept by both sides, made in the specified surface finish.
2. A measurement of that master in L*a*b*, with the instrument and geometry noted.
3. A thickness in microns, written into the purchase order rather than understood.
4. A ΔE band for production, with a sampling frequency and the action taken when a lot falls outside it.
5. A re-approval trigger: any change of target, chamber, substrate polish or thickness restarts the approval.

The viewing condition belongs in the same document. ISO 3664 specifies the standard light sources for colour evaluation, and the practical version is a D65-class lamp at roughly 5000–6500 K against neutral grey. Approving a gold from a phone photograph under warm shop light is how a 3N sample gets signed off for a 2N order.

Why it matters commercially

A set that ships with a 2N pendant and a 3N chain will be photographed together, and a difference invisible across two approval meetings becomes obvious on one card. The customer sees two metals and concludes one is wrong. Nothing is defective; the specification permitted it.

The reverse failure costs more. A brand approves a 2N gold, builds a season on it, and the second order arrives at 2.5N because the target was replaced. Without a master and a ΔE band there is no basis for a claim and no way to sort the goods. Both failures are cheap to prevent at sampling and expensive to fix in a container. Buyers matching tones across necklaces and earrings should read the tolerance work in our guide to PVD colour matching and the pink end of the scale in rose gold PVD; to agree a master sample and a measured band before production, use the RFQ page.