In sieve notation, −2 means the stone passes through a number 2 plate and is not caught by anything below it. In practice that is a diameter of roughly 0.80 to 1.00 mm and a weight around 0.003 to 0.005 carats — somewhere between two hundred and three hundred and thirty stones to the carat.
It is the floor of the commercial diamond trade. Below it, goods exist but are generally sold as dust or industrial rather than as cut stones.
The same fifty-seven facets
The thing that surprises people is that a −2 round brilliant is not a simplified stone. It carries the full architecture: one table, eight stars, eight bezels and sixteen upper halves on the crown; sixteen lower halves and eight pavilion mains below, with or without a culet. Fifty-seven surfaces, cut in a fixed order, on something under a millimetre wide.
Single-cut melee — seventeen facets — does exist, and it has legitimate uses in antique restoration and certain designs. But it is a different product, and it should be asked for rather than substituted. A single cut delivered against a full-cut order will look visibly duller in a pavé field next to full-cut neighbours.
Why it is hard
Tolerances do not scale. A crown angle is 34.5 degrees whether the stone is five millimetres across or one. But at one millimetre, the physical distance corresponding to half a degree of error is vanishingly small, and the cutter is working against that with a wheel that does not know how big the stone is.
You cannot see the result unaided. Every judgement is made under magnification. Fatigue matters more, light matters more, and a cutter who is having a bad afternoon produces goods that will not sort into the same parcel as the morning's.
The girdle is proportionally enormous. Getting a girdle thin enough to look right and thick enough to survive setting is a genuine constraint at this size. Too thin and the setter chips it; too thick and it reads as a grey band that kills the stone's brightness in a pavé run.
Yield decisions are brutal. A single rough crystal might give thirty −2 stones or thirty-six depending on how it is planned. That is a twenty per cent swing in output from one decision on the planning table, which is why we argue with the software rather than accepting its first proposal.
Why most houses buy it in
Because it is slow, it needs specialists, and the per-stone value is tiny. A cutter who spends the day on −2 produces a fraction of the carat weight of one working on half-carat goods. Economically it only makes sense if you are set up for it: benches configured for small goods, cutters who do one step of the fifty-seven all day, and a sorting operation that can turn thousands of stones into a coherent parcel.
Most manufacturers are not set up for it, so they subcontract. That is a perfectly rational decision — but it is also the reason so many melee parcels arrive inconsistent. The house selling you the goods did not make them and cannot tell you who did.
What good −2 looks like
Put a hundred stones face-up under 10× and look at them as a field rather than individually. In a good parcel the tables all sit at the same height, the stars are the same length, and the whole field flashes together when you rock it. In a poor one you will see individual stones going dark at angles where their neighbours are bright — that is inconsistent pavilion depth, and in a finished piece it reads as patchiness.
At −2 there is nowhere to hide a bad angle. The stone is too small to argue about and too numerous to inspect one at a time — so the only control that works is consistency at the bench.
Buying it
Ask who cut the goods. If the answer is vague, the sortment probably will be too. Ask for a sample parcel and look at it as a field. And agree the colour and clarity band in writing before the order rather than negotiating it after — at this size, the band is the product.