Replicating Historic Parquet: What Restoration Buyers Should Ask

Finding a historic parquet replication supplier for a custom reproduction floor starts from a bad position: you have a room where sixty per cent of the original survives, a crate of salvaged blocks, some photographs taken before the ceiling came down, and a budget that was set before anyone lifted the carpet. The question is whether a workshop on the other side of the world can cut new pieces that sit convincingly next to hundred-year-old ones.

The technical answer is more encouraging than most restoration buyers expect, and the commercial answer has one honest catch. A hand-cutting mill can reproduce almost any pattern geometry you can measure. What it cannot do — what nobody can do — is deliver a new floor that matches an old one in colour on the day it is laid.

Matching is five separate problems, not one

"Match the existing" is a single line in a schedule of works and five distinct pieces of work in the shop. Separating them is the most useful thing you can do before you write to any supplier.

Four of the five are procurement problems with clean answers. The fifth is a design decision, and it needs to be made by whoever is responsible for the finished room, not left to the mill.

Measure the block, not the pattern

The single most common error in a replication brief is sending a photograph of the pattern and a room size. Pattern names are unreliable across periods and countries, and two floors that both get called Versailles panels can differ by a hundred millimetres in module.

What the shop needs is the block. Concretely:

• Measure at least ten blocks and record the range, not one block and a claim. Historic parquet was cut by hand or on machines with wider tolerances than a modern CNC, and a set of blocks that read as identical will typically vary by half a millimetre or more. If you send one measurement, the shop will hold it to a tenth of a millimetre and the new work will look mechanically regular next to the old — a giveaway more visible than colour.
• Record the moisture conditions on the day you measured. Timber moves across the grain with humidity. A 70 mm oak block can measure noticeably wider in a damp February building site than in a heated finished room. Note the date, the room condition and, if you can, a moisture meter reading; then state the moisture content you want the new material shipped at.
• Find an unworn edge. The blocks in the middle of a room have been sanded and walked on for a century, so their thickness is not the original thickness. Lift a block from under a skirting, behind a radiator or inside a cupboard and measure that. Then decide the new thickness against the surviving floor, not the original — because the new material has to finish flush with what is actually there.
• Record the joint, not just the outline. Tongue-and-groove, square-edge butt joint, loose spline, or blocks bedded in bitumen on joists — these change the whole substrate design and the laying method. Photograph a block from underneath and from the end grain.
• Establish whether it is blocks or panels. A panel floor is assembled in the shop into modules that arrive complete; a block floor is laid piece by piece on site. Reproducing a panel floor as loose blocks doubles the site labour, and reproducing a block floor as panels changes how it reads.

A workshop running hand cutting alongside CNC, parquet assembly and relief carving in one building — the production chain described for a mill like Anrantabu Flooring, from timber drying and substrate lamination through hand cutting and assembly — can work to any of these. It can only work to what the drawing says, and a drawing built from one measured block is a drawing with an error already in it.

The sample you send, and the paperwork it triggers

Sending a physical salvaged panel is the best brief you can give, and it is the point where restoration projects hit an administrative surprise. A piece of worked timber crossing a border is an import, even at 300 grams. Depending on species and destination it can attract a phytosanitary requirement, a timber legality declaration, and — if the species is listed — a CITES permit. Rosewoods are the ones that catch people out; the permit position for rosewood and Sonokeling flooring applies to a sample as much as to a container.

Before you ship anything irreplaceable, settle three things in writing: whether the mill will work from a physical sample at all, whether it is returned and at whose cost, and what formalities apply at the receiving end. None of that is published for most workshops, so ask. Where the sample is genuinely irreplaceable — a fragment from a listed interior — send a high-resolution scan, a silicone or plaster mould of the profile, and a modern block of the same dimensions instead.

Species and cut: name the sawing pattern, not just the wood

Most European historic parquet is oak, and the reason it looks the way it does is usually the cut rather than the species. Quarter-sawn and rift-sawn oak shows the medullary ray fleck — the pale ribbon figure that reads instantly as an old floor. The same oak, flat-sawn, produces the cathedral grain of a modern plank and will never look right next to it, no matter how well the colour is matched.

So the specification line is not "oak". It is "European oak, quarter-sawn, ray fleck present on the face, minimum X per cent of blocks". Any competent mill can hold that, but only if it is written down, because flat-sawn yields far more usable board per log and is the default everywhere.

This is also where you need to be realistic about a supplier's home ground. A mill whose published species range runs to Burma teak, black walnut, rosewood, Indonesian rosewood and cedar is a mill that buys imported precious hardwoods to order rather than one sitting on a European oak yard. That is not a disqualification — timber is a purchased input and species are sourced against orders — but it does mean the species and cut question should be settled explicitly at enquiry stage, with a sample board from the actual intended parcel before the main order is cut. The general approach to writing a spec sheet a Chinese factory can build from applies directly.

Structure: a historic pattern on a modern base

One thing a modern mill offers that a nineteenth-century one could not is a choice of what sits under the pattern. The same geometry can be produced as solid blocks, or as a three-layer or multi-layer engineered panel with a solid wear layer over a stable core.

For restoration that choice is often the difference between a floor that works and one that does not. A ground-floor room with a modern damp-proof membrane and underfloor heating will destroy a solid block floor bedded the way the original was. An engineered build-up with the same face material, the same block dimensions and the same joint appearance gives you the look on a base that survives the building's current services. Specify the wear layer thickness explicitly — it determines how many times a future generation can sand the floor, which is the whole argument for doing the job properly.

Colour will not match, and that is the decision to make

New wood next to old wood is lighter. Oak ambers over decades; walnut lightens and loses its purple cast; teak darkens and then greys under UV. A floor laid in 1890 has a century of oxidation, wax, dirt in the grain and sunlight-driven colour shift that no finishing process reproduces on fresh timber, and any supplier who says otherwise is selling you a stain that will look wrong from a different angle.

There are four honest strategies. Pick one before you order.

1. Match now, diverge later. Tint or reactively treat the new material to sit close to the aged floor today. It looks best at handover and drifts apart over five to fifteen years as the new wood ages under the applied colour.
2. Match the original as new, and let time close the gap. Lay the new material untinted, accept an obvious difference for a few years, and let it converge. Right for a floor that will be lived on for a long time and wrong for a property being sold next spring.
3. Break the junction deliberately. Use a border, a threshold or a change of pattern where old meets new, so the eye reads two related floors rather than one failed match. Frequently the most convincing answer, and the one an inlay and marquetry capability makes easy — the commissioning workflow for borders and medallions is the same process.
4. Redistribute the old material. Lift the surviving original into the field of view — the centre of the principal room — and put new material under furniture, in the closet and along the perimeter. Costs site labour, saves the room.

Two practical notes. Judge every colour sample on the floor, in the actual room, in daylight and under the room's evening lighting; a sample approved at desk height under a window is approved against a different light path than the one it will live in. And photograph the existing floor next to a neutral grey card before you send references, because a photograph without a colour reference is a photograph of your camera's white balance.

Why the finish choice matters more in restoration than elsewhere

A hand-rubbed wax oil penetrates the grain rather than forming a film on the surface. In a restoration that is the property that counts, because a penetrating finish can be spot-repaired and feathered into the surrounding floor. A cured lacquer film cannot be blended — a repair is visible as an edge, and the standard fix is to sand and recoat the whole room, which on a floor with limited thickness left is a cost you cannot pay twice. The trade-offs on both sides are set out in wax oil versus UV lacquer; for heritage work the repairability argument usually settles it, and it is also closer to what the floor originally had.

The replication brief: what to send

A complete brief, sent once, gets a usable quote. A brief in six emails gets six revised quotes and a two-month delay.

• Dimensioned drawing of the block and the assembled pattern, with the measured range across at least ten blocks
• Photographs: the whole floor, a metre square in raking light, one block from above, below and end-on, all with a scale rule and a grey card
• The unworn thickness, and the finished thickness you need to meet the surviving floor
• Species, sawing pattern and grade, with an agreed reference for figure and colour variation
• Structure and wear-layer thickness, plus the subfloor and any underfloor heating
• Area required, split into replacement field, borders and attic stock
• Your colour strategy from the four above, stated explicitly
• Finish system, coat count and whether a maintenance kit is supplied
• Target moisture content at shipment, set for the finished building
• Any consent conditions from a conservation officer or heritage authority

That last one is not a formality. On a listed or protected building, the consent may specify like-for-like materials and methods, and a replica in a different structure or species can breach it. Get the consent position confirmed before the timber is bought, not after the container ships.

Common questions

Is a custom mill actually cheaper than a specialist heritage joiner?

Sometimes, and the saving is in labour rather than material. A workshop with several hundred hand-craft workers set up for small-to-medium batch custom work can absorb an unusual pattern without the setup cost that stops a volume mill from quoting at all. Against that, put the sample cycle, freight, duty, the cost of the measurement study and the risk of getting it wrong at distance. For a single small patch the arithmetic rarely works. For a room, a staircase hall or a whole floor of a building it often does.

What is the minimum area worth ordering?

Not published by most mills, and worth asking as a direct question rather than assuming. Project-based MOQ generally means the mill prices the job rather than applying a fixed threshold, which is favourable for restoration but makes the answer supplier-specific — the same negotiation logic as any MOQ discussion with a China factory applies. Expect setup-heavy pricing on very small areas: the drawing, the cutting jigs and the first-article approval cost the same whether you order 8 m² or 80 m².

How many sample rounds should I budget for?

Two as a minimum, three realistically: one for pattern geometry, one for species and cut, one for colour and finish. Approve each in writing with photographs before the next begins, and keep a retained half of every approved sample. The mechanics of running that cycle are covered in the factory sample order process and in what to ask for on a custom flooring sample board.

Can the new floor be laid in the same pattern but at a different scale?

Physically yes, visually rarely. Historic patterns were sized to the rooms and the material of their period, and rescaling a Versailles panel to suit a modern room reads as a reproduction. If the room has changed shape, borders are a better answer than rescaling — the construction logic of the panel itself is worked through in how a palace parquet panel is built.

What to ask the supplier next

1. Will you work from a physical salvaged sample, do you return it, and what do you need at your end to receive it?
2. Can you source European oak in a named sawing pattern, and can I see a sample board from the actual parcel before the main cut?
3. What is your process for colour-matching against an aged floor, and what do you work from — a sample, a photograph, or a physical reference you keep?
4. What is the smallest area you will produce, and how is a patch order priced against a full-room order?
5. What tolerance do you hold on block dimensions, and can you deliberately widen it to avoid mechanical regularity next to hand-cut original work?
6. What moisture content will you ship at, and how is the material packed for a long sea transit?

Anrantabu Flooring in Huizhou is one workshop of the type this article describes: custom parquet, marquetry and inlay work hand-cut in-house, with solid and engineered bases and custom designs developed to a buyer's specification. Its capability detail and contact route are on the Anrantabu Flooring factory page — send the six questions above to it and to two others, and read the replies for how precisely each one answers the colour question. That answer tells you most of what you need to know.