Three suppliers quote the same oak floor. The prices land 40% apart and every line on every quote says "oak, 190 mm wide, natural oil finish." The difference is not the oak. It is what is underneath the oak, and in two of the three quotes the amount of oak you actually own — the part you can sand — is a fraction of a millimetre.
Getting solid vs three-layer vs multi-layer engineered wood flooring structure right is the single decision that determines whether a floor can be refinished in fifteen years, whether it can go over a concrete slab, whether it can go over heating, and whether the price you are comparing is comparable at all. It is also the decision most often left to the supplier, who will reasonably default to whatever their line runs best.
The four constructions, in the order of how much wood you get
A custom mill that works across the whole range will normally offer four structural systems. Anrantabu Flooring, a custom handmade solid wood flooring maker in Huiyang, Huizhou, builds all four — solid, geothermal-compatible, three-layer and multi-layer — across parquet, plank and staircase work. They are not four grades of the same thing. They are four different structural answers to the same problem: wood moves across the grain, and a floor is a wide flat thing that must not.
Solid. One piece of timber, tongue and groove milled into the edges, no core, no backing. Everything above the tongue is wear layer, which normally gives the deepest sanding reserve of any construction. Everything about its movement is unrestrained.
Three-layer. A sawn lamella of the show species on top, a middle core of softwood strips laid perpendicular to the lamella, and a backing veneer beneath. The classic European construction. The cross-grain core restrains the face; the backing keeps the moisture exchange symmetrical.
Multi-layer. A lamella on a plywood core of several thin plies, each laid at right angles to its neighbour. More cross-restraint than a three-layer, generally better dimensional behaviour under cycling, and the construction that most mills build their heated-floor product from.
Geothermal / heat-stable. Not a separate material family so much as a multi-layer built with the core, adhesive and balance specified for heat cycling. The distinction matters commercially because a mill that lists it separately has usually done something specific to the substrate rather than relabelling stock board.
Three details that separate a real construction from a marketing name:
- Sawn lamella, sliced veneer or rotary-peeled veneer. A sawn lamella is cut on a saw — real grain, no lathe damage, and it can be thick. A sliced veneer is knife-cut, thin, and flat. A rotary-peeled veneer is unwound off a log against a knife, which leaves microscopic lathe checks on the underside of the sheet. Under a penetrating oil finish in raking light those checks eventually show as a fine, regular crazing across the face. It is not a defect anyone will accept a claim on, and it is invisible on the sample board. Ask which of the three you are getting, by name.
- The backing veneer is not optional. A lamella glued to a core with nothing on the back is an unbalanced panel. It will cup toward the dry side. A quote that does not mention a backing layer is a quote you should query before you compare its price.
- Core gaps telegraph. In a three-layer construction the middle layer is strips of softwood with small gaps between them. If the lamella is thin and the sanding is aggressive, those gaps read through the face as faint parallel lines in low-angle light — usually across a hallway, usually noticed on the day of handover.
Wear layer is the only number that tells you what you own
The wear layer, or lamella, is the thickness of show species above the tongue. It is the number that decides refinishing life, and it is the number most often missing from a quote.
The arithmetic is simple and unforgiving. A full machine sand-and-refinish removes material — commonly in the region of half a millimetre to a millimetre per pass, depending on how flat the floor is and how much of the old finish has to come off. You also have to leave material above the tongue shoulder, because sanding into the tongue destroys the joint. Practically, the usable reserve is the wear layer minus roughly half a millimetre.
Run that against the constructions:
| Construction | Where the wear layer sits | Practical refinishing reserve | What that means in service |
|---|---|---|---|
| Solid | The whole board above the tongue | Deepest of the four; usually several full sands | Refinishable across generations if it stays flat |
| Three-layer, sawn lamella | Top layer only | Meaningful — a thick sawn lamella tolerates repeat sanding | Refinishable, and repairable in patches |
| Multi-layer, sawn lamella | Top layer only | Comparable to three-layer at the same lamella thickness | The usual choice over slabs and over heating |
| Multi-layer, thin sliced veneer | Top layer only | Very little; often none at all | Screen-and-recoat only. A sanding job takes you into the core |
The number that fills the middle column is not published for most mills and has to be asked for per configuration. Anrantabu's lamella thickness options, total board thickness range and plank width ranges are not stated in the public record, so treat them as open questions rather than assumptions — put them in the RFQ and get the answer in millimetres, per construction, in writing.
That single missing number explains most of the price spread that makes buyers suspicious. A quote built on a 0.6 mm sliced face and a quote built on a 4 mm sawn lamella are two different products with the same description, and the cheaper one is not badly priced — it is a different thing. The same mechanism is at work in most of the cases covered in why China quotes for the same spec differ: the spec was not specific enough to force the suppliers to quote the same object.
Matching construction to subfloor, installation and climate
The right construction is mostly determined by three site facts that the buyer usually knows and the mill usually does not.
| Site condition | Solid | Three-layer | Multi-layer | Notes |
|---|---|---|---|---|
| Timber joists / plywood subfloor, stable climate | Good | Good | Good | Solid is genuinely fine here; nail or staple down |
| Concrete slab, ground floor | Poor | Fair (glued) | Good (glued) | Slab moisture is the constraint; solid over slab is a recurring failure |
| Hydronic underfloor heating | Poor | Fair | Good | Cross-restraint and thermal resistance both favour multi-layer |
| Wide plank, over 180 mm | Risky | Good | Good | Movement scales with width; engineered halves the exposure |
| Coastal or highly seasonal humidity swing | Risky | Good | Good | The swing, not the average, is what cups a floor |
| Floating installation with underlay | Not suitable | Good | Good | Solid must be mechanically fastened or fully bonded |
| Client expects sanding twice in 30 years | Good | Good, if lamella is thick | Good, if lamella is thick | Thin sliced veneer rules itself out here |
Two things this table hides that are worth saying plainly.
First, acclimatisation rules invert between constructions. A solid floor should generally be delivered early, opened, and left to reach equilibrium with the room before installation. An engineered floor usually should not be opened early — the packs are wrapped to hold the board at the moisture content it left the mill at, and letting a multi-layer board sit open in a humid site does nothing useful and can do harm. Installers who learned on solid will do the wrong thing with engineered unless the instruction is on the pack.
Second, edge detail changes what the client sees. A square-edge floor shows every seasonal gap as a shadow line. A micro-bevel of half a millimetre or so absorbs the same movement invisibly, at the cost of a slightly less monolithic surface and a groove that collects dust. Over heating, the micro-bevel is usually worth taking. It is also the cheapest insurance against a snag list.
Why in-house substrate lamination changes the conversation
Most flooring exporters buy their core board on the open market and glue a lamella to it. That is a legitimate way to make floors, but it puts the two variables that decide long-term flatness — core moisture content at glue-up, and adhesive spread rate — outside the mill's control.
Anrantabu produces its own substrates rather than buying them in, and runs timber drying and acclimatisation, substrate lamination, CNC and hand cutting, parquet assembly and hand-sanding inside the same 20,000 m² plant. For a buyer, the practical value of that is not the phrase "in-house." It is that there is one party who can answer a question like what was the core at when you pressed it without forwarding your email to a board supplier.
Which gives you a useful audit question for any mill, not just this one: ask to see the moisture reading log for the core stock and the lamella stock on the day your order was pressed. A mill that laminates its own substrate can produce that. A mill that buys board in cannot, and will say something about their supplier's certificate instead. Neither answer is disqualifying — but the difference tells you where the risk sits, and it is the sort of thing worth confirming during the sample order process rather than after the container ships.
What to ask the supplier next
Send these as a numbered list and require numbered answers. The goal is not to catch anyone out; it is to force three suppliers to quote the same object so the prices mean something.
- Per construction, state the wear-layer thickness in millimetres. Sawn lamella, sliced veneer or rotary-peeled — say which.
- State total board thickness and available plank widths per construction, and confirm the width tolerance.
- Describe the core. For three-layer: species and strip width. For multi-layer: number of plies, species, and whether the inner plies are void-free.
- Confirm the backing veneer — species, thickness, and its ratio to the lamella.
- How many full sands does the construction support, in your own words? Get the mill's own answer on the record.
- What is the target moisture content at despatch, and the tolerance band? Ask what destination climate it is set for.
- Send the dimensional and flatness tolerance table — thickness, width, length, squareness, cupping, bow, in millimetres.
- What adhesive, and what formaldehyde emission class? This drives your customs and compliance paperwork as much as your quality expectation.
- Square edge or micro-bevel, and how large is the bevel?
- Send one sample per construction in the same species and finish. Not a showroom board — the actual build.
That last point does most of the work. Ask for a solid, a three-layer and a multi-layer sample in the same species, cut through so you can see the section, then keep them in the driest room you have for a month and look at them again. Custom samples are available on request from Anrantabu Flooring, which works on an OEM and ODM project basis across all four systems and in species including Burma teak, black walnut, rosewood, Sonokeling and cedar — so a three-way section comparison in your actual specification is a reasonable thing to ask for before you commit a project to one construction.
For the two decisions that sit either side of this one, what a Versailles panel is built from covers the parquet module, and what the substrate has to do over underfloor heating covers the case where the construction choice is effectively made for you.