Nabest — Factory-Direct 925 Sterling Silver & VVS Moissanite  |  Stock from 1 pc · Custom from 10 pcs  |  OEM / ODM & Dropshipping Ready
Home / Buying Guides / Decide how it is made and checked / Silver Jewelry Defects: How to Specify Against Them
Buying guide

The defects buyers actually find in silver jewelry, and how to specify against them

Last updated: October 2026

By Jack Lv, General Manager — 925 sterling silver and GRA moissanite manufacturing in Foshan since 2011

Silver jewelry fails inspection in a small number of repeating ways: porosity and gas holes from casting, stones that are loose or missing, plating that is dull or worn through, blackened solder joints, dimensional and weight deviation, and surface scratches from uneven polishing. Each one can be written into a purchase specification as a measurable requirement, which is what makes it findable at inspection rather than arguable after delivery.

Which defects cause most claims on 925 silver jewelry orders?

Porosity and gas holes, loose stones, plating that wears through, weak solder joints, and dimensional drift between the sample and the bulk. They dominate because four of the five are invisible on a photograph and only appear after wear or after a customer opens the box.

Defects in silver jewelry cluster. They come from a handful of process steps, and each one appears at a predictable point in the workflow. If you know where a defect is created, you know which inspection gate can catch it and which specification line prevents it.

The table below maps the main families to their origin and to how they present when the parcel is opened.

Defect familyWhere it is createdHow it presents to the buyer
Porosity and gas holesCasting, from gas trapped in the metal or from a poorly filled mouldPits on a supposedly polished surface, often visible only at an angle or under magnification
Loose or lost stonesStone setting, from an under-sized seat, weak prong contact or a burr left in the seatA stone that rotates or rattles on arrival, or a missing stone in a returned piece
Dull or worn platingPlating, from thin deposit, poor surface preparation or contamination in the bathA finish that looks flat at unboxing, with base metal showing at edges and high wear points
Blackened solder jointsAssembly and soldering, from oxide left in the joint or from solder that does not match the alloyDark lines at joints, splayed discolouration around a repair, or tarnish that starts at the joint
Dimensional and weight deviationCasting, forming or finishing, from tool wear and from grinding metal awayPieces that do not fit a size chart, components that do not assemble, or a batch lighter than the approved sample
Scratches and uneven polishPolishing and handling, from coarse compound, mixed work in one barrel or unprotected storageSleeve marks, flat patches where detail should be crisp, and fine scratching across a plain surface

How do you specify porosity and gas holes so they become rejectable?

Write a maximum pit diameter and a maximum count per surface area, and name the inspection light. Without a number a defect is an opinion, and an inspector on the factory floor will resolve the opinion in the factory's favour.

Porosity is the defect buyers most often accept without knowing it, because it often needs magnification or a raking light to see. A specification that says the surface must be free of pits is almost unenforceable. One that names the inspection method, the light condition and the magnification is enforceable.

Porosity cannot be polished out. It sits below the surface, so polishing a porous casting removes metal, exposes further pits, and leaves a piece that is thinner than the approved sample. That is why porous castings should be rejected rather than reworked.

Require the finished surface to be free of pits, pinholes and open blisters when examined under a loupe or microscope and under raking light.
State that a porous casting may not be corrected by polishing, and that pieces with porosity are a reject, not a rework.
Require any porosity found at an edge, at a solder joint or on a sealing surface to be treated as a critical defect, because those are the points where a crack will start.
Ask the factory to describe how it avoids porosity, for example through metal temperature and mould filling control or the use of vacuum or centrifugal casting, and record the answer in the specification.
Require that the alloy used for casting is recorded per lot, which is where incoming material inspection and the XRF spectrometer do their work.

How do you specify stone setting so a loose stone is detectable?

Specify the setting type, the prong count and a pull or probe test per batch rather than per piece. A loose stone is the defect most likely to reach your customer, because it usually passes a visual check and fails in wear.

A stone that falls out is the claim that costs the most, because it damages the customer relationship as well as the piece. The specification has to define what tight means, and it has to define when the check happens.

Timing matters more than most buyers realise. Setting faults are easiest to detect before plating, and harder afterwards, because a piece that has already been plated cannot easily be re-seated without marking the finish. Nabest runs setting checks at the in-process stage for exactly this reason.

Require each stone to sit level in its seat, with no visible gap between the girdle and the metal, checked by eye and by loupe or microscope.
Require a light pressure or probe test at the bench, and state that a stone which moves, rotates or rattles fails regardless of appearance.
Specify the minimum contact requirement in words, such as full contact along the seat wall and prongs that are pressed onto the stone rather than folded over it.
Require loose or damaged stones to be re-seated during in-process inspection, and require any re-seating after plating to be declared, since it affects the finish at that point.
Fix the stone grade and size in the specification, and state that a replacement stone must match the grade supplied, with its own report, because GRA moissanite reports are issued per stone rather than per order.
Note that moissanite sits at 9.25 on the Mohs scale, so a stone that looks dull after wear usually indicates a soiled or damaged surface rather than abrasion, whereas cubic zirconia at 8 to 8.5 scratches and dulls more readily in the same setting.

How do you specify plating thickness and appearance?

State the thickness in microns and the coverage areas, then verify on a sample from the bulk rather than on the pre-production sample. Plating thickness is the single line where a cheaper quote most often hides, because it is invisible to the eye at delivery.

Plating is the specification most often written as an adjective. Plated in white rhodium tells an inspector nothing they can measure. Plating thickness stated in microns, controlled at the plating line and recorded per lot, tells them exactly what to check.

Two honest trade-offs belong in the conversation with your supplier. A thicker deposit costs more and takes longer, and on fine detail a heavy deposit can soften the crispness of a pattern. Thin plating also fails fastest at the points that rub, so the specification should name where the plating will be examined, not only how thick it should be.

State the target plating thickness in microns and the tolerance you will accept, and require the deposit to be controlled and recorded at the plating station.
Name the inspection points, including the inside of a band, the edges of a pendant, the tips of prongs and the clasp on a chain, which are where wear-through appears first.
Specify the appearance under a defined viewing condition, such as uniform diffused light against a white background, so that dull and sheened pieces are judged the same way twice.
Require no exposed base metal at any point, and state that a piece showing plate wear-through on a high point fails as a major defect.
Require an adhesion check on the finished deposit, such as a tape test and a rub test, and require any piece that fails to be rejected rather than re-polished.
Require the surface preparation steps to be described, because plating over contamination produces a deposit that looks correct on the day of packing and fails later in wear.

How do you specify solder joints and polishing quality?

Specify joint strength by a pull test and polishing by a named standard finish with a reference photo. A reference photo settles most polish disputes because it converts a matter of taste into a comparison.

Solder faults are visible and predictable. They appear as a dark line at a joint, as a patch of discolouration around a join, or as a flow of solder onto a surface where it should not be. A specification should require the joint to match the surrounding metal in colour after finishing, and to be free of oxide and of visible solder spread.

Polishing faults are the mirror image: they are about what should have been removed and what should not have been removed. Written carelessly, the requirement becomes an instruction to polish everything flat, which destroys the design and the weight. Written properly, it protects detail as well as removing marks.

Require joints to be clean and colour matched, with no blackened line, no oxide bloom and no solder bleed onto adjacent surfaces.
Require joints to be dressed so that they do not create a step or a ridge that catches a cloth or a finger.
State that pieces from different stages, such as plating and pre-plating work, must not be processed together in the same polishing or finishing batch.
Require the approved sample to define detail: crisp edges, sharp engraving and visible texture must survive, and a piece polished to a smooth blank is a reject.
Require handling discipline from polishing to packing, including finger protection and individual bagging, because most fine scratching is created between the last polish and the pouch.

How do you specify dimension and weight tolerance?

Give a nominal plus or minus tolerance for the critical dimensions and for finished weight, and state which dimensions are critical. Weight tolerance matters commercially as well as technically, because on a silver piece it is also a proxy for how much metal you paid for.

Dimension and weight are the two properties that catch everything else. A batch that is light usually means metal was removed during finishing, which often accompanies porosity repairs and over-polishing. A batch that is dimensionally out of line usually means a worn mould or die, which drifts gradually rather than failing suddenly.

Both should be measured with instruments rather than judged by eye, and both should be tied to the approved sample rather than to a drawing that nobody has compared against a physical piece.

Fix dimensions against the approved sample and state the tolerance you will accept on each critical dimension, measured with calipers or gauges.
State a weight tolerance in grams per piece, and require an electronic balance check on finished goods, since a light piece is the fastest indicator of hidden metal loss.
Require the alloy to be verified with an XRF spectrometer, so that a light piece cannot be explained by a lower silver content, and record the reading.
Require critical assembly dimensions, such as clasp fit, ear post thread and any component that must mate with another part, to be checked as a functional test rather than as a measurement alone.
State that components from a changed mould or die require a new approval sample, because a tool change is the point at which dimensions drift.

What is AQL sampling for, and what can it not do?

AQL sampling tells you whether a batch is acceptable at an agreed defect rate; it does not prove the batch is free of the defect you care most about. It is a decision rule, not an inspection of every piece, so pair it with a full check on the one characteristic that would end the order.

AQL stands for Acceptable Quality Limit. It is a sampling method in which a defined number of pieces is drawn from a lot and inspected, and the lot is accepted or rejected against a threshold number of defects in defined classes, usually critical, major and minor. It is a lot-level decision rule, not a measurement of individual pieces.

Its limits are worth stating plainly, because buyers often treat an AQL result as a guarantee. A passed inspection does not mean the lot contains no defective pieces. It means the number of defects found in the sample stayed inside the agreed threshold. A small sample can miss a fault that appears later in the same lot, and a defect class that was not defined at inspection cannot be counted against the supplier afterwards.

AQL also cannot fix anything. Applied only at the end of production, it can only sort good pieces from bad ones. Applied at the in-process stages, the same method prevents value being added to a piece that will be rejected.

Define the defect classes in writing before the order, with two or three examples of each, and attach photographs so that everyone inspects the same thing.
Agree the sampling plan and the acceptance and rejection thresholds in the purchase order, so a failed inspection has a stated consequence.
Use a random draw from the packed lot, not a tray selected by the factory, and record the cartons the sample came from.
Apply the method at in-process stages as well as at the end, since that is where a rejection still costs a fraction of the finished value.
Keep the inspection record with the shipment, including photographs, so a claim can be argued from evidence rather than from memory.

What can each of the four inspection gates catch?

Pre-production sample catches specification errors, in-line inspection catches process drift, pre-shipment inspection catches the bulk, and incoming inspection catches freight damage. Skipping the in-line gate is the commonest cause of a pre-shipment failure, because by then the whole batch carries the drift.

Nabest runs four inspection gates, IQC on incoming material, IPQC during production, FQC on finished goods and OQC before shipment. The gates are not interchangeable, and the clearest way to write a specification is to say which gate each requirement belongs to.

GateApplied toCatchesCannot catch
IQC, incoming materialSilver alloy, loose stones, findings, clasps, chains, packagingWrong alloy grade, wrong stone grade or size, undersized findings, damaged packagingAnything created later on the line
IPQC, in processWax or dies, castings, settings, polishing, platingPorosity at casting, loose seats before plating, plating appearance and thickness at the tank, joint faults before finishingDefects introduced by packing or handling after the stage
FQC, finished goodsAssembled, finished and plated piecesStone security, surface finish, solder colour, dimensions, weight against the sample, function testsHow the batch is packed, labelled or counted
OQC, before shipmentPacked cartons, labels, paperworkShort or over shipment, mixed models, wrong or missing labels, packing damage, documentation errorsFine surface defects, because it inspects the packed lot rather than re-finishing every piece
Write each requirement as a line with an owner and a gate, so that a failure can be traced to a step rather than to a department.
Treat packing and labelling as product requirements, because a mislabelled carton in a marketplace warehouse is as expensive as a finished-goods defect.
Ask for photographs of the finished goods and the packed cartons before dispatch, which is the evidence you will use if the parcel arrives wrong.

How do you turn this list into a specification sheet?

One page, one line per characteristic, each with a nominal value, a tolerance and a test method. Tie it to your commercial terms: Nabest quotes a minimum order quantity of 10 pieces per style, with sampling at 5 to 7 days and bulk production at 15 to 20 days, which is the window in which a specification dispute has to be settled.

A specification sheet that survives a production run has four sections: material, dimensions and weight, finish and function, and inspection. Every line in it should be checkable by someone who was not in the negotiation, with an instrument or a defined viewing condition, and against a physical approved sample.

Keep it short enough to be read at a bench. A specification that runs to many pages and repeats itself is usually ignored, and the paragraphs that get ignored are the ones about plating and finish, which is where the smallest differences show up most visibly to a customer.

Finally, tie the sheet to your commercial terms. Nabest quotes a minimum order quantity of 10 pieces per style, with sampling at 5 to 7 days and bulk production at 15 to 20 days, which means the specification is finalised at sampling. Changing a critical line after sampling has begun restarts the clock rather than adjusting it.

One line per requirement, with the gate that checks it, the method used, and the action if it fails.
One approved physical sample, signed and dated, held by both parties, with photographs attached to the specification.
Explicit defect classes, so that accept, rework and reject are decisions rather than debates.
A agreed tolerance for every measurable line, including plating thickness in microns and weight in grams.
A place to record what was actually inspected, so the next order starts from a record instead of a conversation.
Questions we get

Answered

Which silver jewelry defect causes the most claims?

Lost or loose stones, followed by plating complaints. Stones cause the highest cost per claim because the piece usually cannot be repaired by the customer. Plating causes the highest volume of complaints because it is the first thing a customer sees and the property most affected by wear, perfume and household chemicals.

Can a passed AQL inspection guarantee my order has no defects?

No. AQL sampling decides whether a lot is accepted or rejected against a threshold of defects found in a sample. It is a lot-level rule, not a promise about individual pieces, and a defect class that was not defined before inspection cannot be counted against the supplier afterwards.

How should plating be written into a purchase specification?

As a thickness in microns with a tolerance, controlled and recorded at the plating station, plus named appearance requirements under a defined viewing condition and named inspection points where wear-through shows first, such as the inside of a band or the tips of prongs. Adjectives are not checkable and should not appear as the only requirement.

Where in production should porosity be caught?

At the in-process gate, immediately after casting and before polishing or plating. Porosity sits below the surface, so polishing a porous casting removes metal and exposes further pits, and plating over it hides the fault until the customer wears the piece.

Do I need my own inspection if the factory runs four inspection gates?

A supplier's own gates are a process control, and they are worth asking about in detail. Your own inspection, whether by AQL sampling at the factory before dispatch or by a third-party inspector on site, is the evidence that the goods you paid for are the goods that were specified, which is a different question.

Want to see the plant and the process?
Ask us for factory photos and audit documents.

  WhatsApp Us