Why Stones Fall Out of Finished Jewelry — and What Production Does About It

A lost stone is the failure a brand hears about first — from the customer, not the factory. It is rarely a bad stone. In almost every case it is a sequence problem: the order in which the operations happened, and the moment the setting was last checked.

Where a stone actually escapes

A set stone is held by geometry, not glue. The seat is cut to the stone's real girdle, the metal is pushed over it, and the piece is checked. Every escape route is a way that geometry gets compromised — usually by a later operation undoing an earlier one.

The bearing was cut to the nominal size, not the actual stone

Calibrated stones are sold in size ranges, not exact diameters. A seat cut to the nominal size can leave a stone rocking microscopically in its bearing. It looks perfectly set on the bench. Under the directional load of daily wear — a ring against a doorframe, a bangle against a desk edge — a rocking stone works its prongs open over months. The seat has to be cut to the stone in hand, and the stone in hand has to come from the lot that will be used, not a sample of the lot.

The seat was dirty

Wax residue, polishing compound and oxide film all make a stone sit high in its bearing. The prongs close over the wrong plane, the stone is mechanically trapped rather than seated, and it leaves at the first impact. This is why cleaning the seat is a step with its own place in the sequence, not something that happens if there is time.

The prong was polished away

Metal removed by a file or a wheel does not come back. If the prong is trimmed to thickness before final finishing, the finishing takes it below the workable minimum and nobody notices, because nobody measured it afterwards. Prong thickness is checked on the finished profile — after finishing, not before.

Heat moved the seat

Soldering near a set stone without protecting it shifts the seat: alloy flows, the bearing tilts, the stone sits proud on one side. Every heat operation belongs before the last stone goes in. That single sequencing rule prevents a large share of stone loss, and it costs nothing but discipline.

Wear opened the setting

A ring shank flexes thousands of times a day. Prongs take directional load; a pavé row distributes it; a single large stone on a thin shank concentrates it. This is a design question as much as a production one — the wall thickness and stone seat decision made at the CAD stage decides how much abuse the finished setting can absorb.

Why the order of operations decides the outcome

The bench sequence that prevents stone loss is unglamorous: all heat before final setting, final polish before the last stones go in, retention checked under magnification after finishing rather than before. None of this is visible in the finished photo, and all of it is visible in the warranty record two years later.

It also means the honest answer to "will the stones stay in" is never a guarantee — it is a description of the sequence and the checks. A supplier who answers with a guarantee is selling confidence; a supplier who answers with the sequence is showing you their work.

What to specify before production starts

Stone sizes with their tolerances, the setting method per stone group, and where the piece takes wear in real use. A bangle that will be worn every day and a display piece for a trade fair do not need the same retention regime. Saying which one yours is lets the workshop make the right trade-off instead of guessing at it — and it changes what the workshop checks at the end.

What to ask your supplier

Two questions cover most of it: at what point in the sequence are stones set, relative to soldering and final polish? And when is retention checked — before or after finishing? A workshop with a real sequence answers both in seconds, because the sequence is written down. The setting methods overview and the inspection page describe how these gates sit inside our own routing.