Industries / Consumer goods / Jewelry
Casting and forming
Casting and forming makes the metal parts of a piece: pouring molten metal into molds, striking or bending metal into shape, joining parts, printing them from powder and finishing their surfaces.
7
Milestones achieved, from c. 4500 BCE to 2016.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
c. 4500 BCE
The oldest lost-wax casting
In lost-wax casting a model is made in wax, covered in clay or plaster, and heated so the wax melts away, and molten metal is poured into the hollow it leaves. A small copper amulet shaped like a spoked wheel, found at Mehrgarh in Baluchistan, Pakistan, in a layer dated 4500 to 3600 BCE, is the oldest known object made this way, by the account of a team led by Mathieu Thoury and Loïc Bertrand, who imaged its metal with a new light-based method in Nature Communications in 2016.
Credit
the metalworkers of Mehrgarh; identified by Thoury, Bertrand and colleagues
c. 1858
Drop hammers strike rings
In die striking, a heavy weight drops onto metal and drives it into a steel die cut with the shape. Around 1858 drop hammers, die makers and die sinkers spread through Providence, Rhode Island, starting with the city's first drop hammer at the Gorham Company, and ring shanks were die-struck before being chased and engraved. Before this, around 1840, rings were made by passing a die through a rolling mill. One die-struck signet design could be made in eight finger sizes.
Credit
the Gorham Company and Providence die makers, by the account of Peter DiCristofaro of the Providence Jewelry Museum
1940s
Rubber molds multiply wax models
In 1907 William H. Taggart, a Chicago dentist, published a precise lost-wax method for gold inlays, with his own pattern wax, investment (the plaster-like mold material) and an air-pressure casting machine. In the 1940s New York ring makers, "taking their lead from the dental industry", made rubber molds of master models, and Aaron Finelt developed the first wax injector to fill them, by Peter DiCristofaro's account. One master could now yield many identical wax copies, cast together on a wax "tree" in one mold. Sources differ on timing: Jerry Taylor of Hoover & Strong wrote in 2017 that production lost-wax casting came to jewelry about 80 years earlier, before the Second World War, while a World Gold Council text says it was not widely used in commercial jewelry until after the war.
santafesymposium.orghitchiner.compmc.ncbi.nlm.nih.govcigem.ca
Credit
William H. Taggart; Aaron Finelt; New York ring makers
1994
Europe limits nickel release
Nickel hardens and whitens many jewelry alloys, and sweat dissolves it into the skin, where it can cause a lifelong allergy. In 1991 the Swedish dermatologist Carola Lidén and colleagues put evidence to the European standards committee for jewelry that up to 10% of women were affected. EU Directive 94/27/EC of 30 June 1994 banned post assemblies for new piercings with 0.05% or more nickel, and earrings, necklaces, rings and other items in long skin contact that release more than 0.5 µg of nickel per cm² per week. Agreeing on the test methods took more than four years, and the rule came into force in 2001. Patch tests of 180,390 patients in Denmark, Germany, Italy and the UK then showed nickel allergy falling among women under 30.
LessonA limit protects only as far as it is enforced; a 2025 review found that 24.7% of European earrings for piercing failed the EU's standard release test (EN 1811), and that simpler spot tests flagged 34.5% of earrings from Asia and 31.1% from North America.
Credit
Carola Lidén and colleagues; the European Parliament and Council; the fall in allergy measured by Garg and colleagues, 2013
1998
Lasers weld at the bench
Jewelers traditionally join parts by brazing, which the trade calls soldering, with a torch that heats the whole area. Costantino Volpe of Tiffany & Co. and Richard Lanam of Engelhard-CLAL wrote in 1998 that laser welders had shrunk to less than the size of a jeweler's bench and fallen in price from several hundred thousand dollars to under $50,000. In their tests on platinum and 18-karat gold, laser welds were stronger and less porous than brazed joints, though slower to make. John C. Wright wrote in 2001 that, under ideal conditions, a laser can weld as close as 0.2 mm to stones, hinges and even pearls, so stones often need not be removed for repairs.
Credit
Costantino Volpe and Richard Lanam (tests); John C. Wright
2010
Cadmium replaces banned lead
The US Consumer Product Safety Improvement Act of 2008 set a strict limit on lead in children's products. In January 2010 the Associated Press reported that some Chinese makers of children's jewelry were using cadmium, a more toxic metal, instead. Lab tests for the AP by Jeff Weidenhamer of Ashland University found that 12 of 103 items bought in four states contained at least 10% cadmium, and one charm 91% by weight. A lawyer for the Fashion Jewelry Trade Association said its members did not see widespread substitution. The EU banned 0.01% or more cadmium by weight in the metal parts of jewelry from 10 January 2012.
LessonBanning one toxic metal without limiting its substitutes can move the hazard instead of removing it.
Credit
the Associated Press and Jeff Weidenhamer; the European Commission
2016
Lasers melt gold powder
Laser powder bed fusion, also called selective laser melting, builds a part by spreading thin layers of metal powder and melting each layer's outline with a laser. Towe Norlen described making gold jewelry with a selective laser technique in 2006. Gold reflects almost all of the infrared laser light, so parts came out porous. Ulrich Klotz, Dario Tiberto and Franz Held of the Research Institute for Precious Metals and Metals Chemistry (FEM) in Schwäbisch Gmünd, Germany, cut porosity in 18-karat yellow gold from about 2.5% with fresh powder to 0.7% with surface-treated powder, and to 0.1 to 0.3% with alloy additions. Printed surfaces measured 10 to 13 µm in average roughness (Ra) and needed blasting and polishing.
LessonPrinting gold took new powders and alloys, not only a new machine.
Credit
Ulrich Klotz, Dario Tiberto and Franz Held, FEM; earlier work by Towe Norlen
Watch for
Printed platinum goes on sale
Platinum Guild International unveiled Tùsaire, a collection designed by Maeve Gillies and printed directly in platinum by laser powder bed fusion at ProGold in Italy. The trade site VoxelMatters calls it "the world's first commercially launched platinum-focused collection utilizing direct metal 3D printing" and gives no production numbers or outside tests.
Achieved whenan independent lab reports the density and flaws of production pieces, or the makers publish volumes that show printing in routine production.
Claimed by
Platinum Guild International and ProGold, June 2025
Printing gold without loose powder
Legor, an Italian precious-metals company, added Tritone's MoldJet printing process to its 3D Metal Hub and became Tritone's manufacturing partner for jewelry. The article says the process avoids the safety and metal-recovery problems of handling loose precious-metal powder and makes complex gold and platinum parts with high definition and surface quality, and gives no figures.
Achieved whenan outside lab publishes density, surface roughness and alloy measurements for MoldJet parts beside cast parts of the same design.
Claimed by
Legor and Tritone, June 2026
Open
01
An open casting-defect atlas
A public, labeled image set of flaws in cast jewelry, cut open and photographed under a microscope: gas porosity, shrinkage porosity, inclusions of mold material, incomplete fills and cracks. Tag each with the alloy and casting conditions where known, and publish a baseline classifier that names the flaw type from a photo.
The briefCredit · unclaimed02
Robot polishing by demonstration
A low-cost robot arm that learns to pre-polish cast rings from a few dozen human demonstrations, working at a fixed polishing motor inside a guarded enclosure, with surface roughness measured before and after each ring.
The briefCredit · unclaimed
Open 01 · the brief
An open casting-defect atlas
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public, labeled image set of flaws in cast jewelry, cut open and photographed under a microscope: gas porosity, shrinkage porosity, inclusions of mold material, incomplete fills and cracks. Tag each with the alloy and casting conditions where known, and publish a baseline classifier that names the flaw type from a photo.
Why it is still open
Flaws in jewelry castings are still diagnosed case by case by experts. Dieter Ott of FEM traced severe gas porosity in one white-gold ring to polluted raw material, and FEM measures porosity by image analysis of polished sections. Searches of Hugging Face and Crossref in November 2026 found jewelry image sets made for image generators and try-on, porosity data for other processes, and casting handbooks and papers, but no labeled image set of jewelry casting flaws.
Built with what exists
Rejected castings donated by casting houses; a metallography service, or a bench saw and polishing papers; a USB microscope; labeling in Label Studio; an open detector such as Ultralytics YOLO, or a vision-language model such as Claude Fable 5.2 describing each flaw (see the Note "Three working devices") with a second model flagging doubtful labels for a person to check (see the Note "Seven more flagged").
First step
Ask one casting house for 50 rejected castings with notes on alloy and flask temperature, then section ten of them and photograph the cut faces.
Done looks like
At least 3,000 labeled flaw images from three or more casting houses, released under an open license, with the classifier's accuracy for each flaw type reported on castings from a casting house held out of training.
Open 02 · the brief
Robot polishing by demonstration
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A low-cost robot arm that learns to pre-polish cast rings from a few dozen human demonstrations, working at a fixed polishing motor inside a guarded enclosure, with surface roughness measured before and after each ring.
Why it is still open
The robotic polishing cell found in jewelry uses special fixtures and tool changers for each design. Christian Tse Design & Manufacturing's VDK-2300 system from Flexible Robotic Environment (2022) uses two industrial arms, one holding the piece and one the tool, and the company's release gives no measured roughness or time. OTEC's EF-Nova (March 2025) replaces polishing with an electrochemical process and claims up to 90% less manual polishing, by its own figures. A 2026 review of robots that learn deburring and polishing from demonstration mapped 24 studies across industry and still lists the gap between simulation and real parts as a barrier; none of the jewelry systems found publishes a measured comparison with a bench jeweler.
Built with what exists
A low-cost arm trained with LeRobot (see the Note "No download"), or an open robot model fine-tuned from a few demonstrations, such as NVIDIA GR00T 2 (see the Note "Five demonstrations per task") or Runway Praxis-1 (see the Note "Open robot model weights"); a bench polishing motor or flex shaft with guards, dust extraction and eye protection; brass or silver castings in place of gold to keep costs down; a contact profilometer or a lab service to measure average roughness (Ra). As a reference, FEM measured 10 to 13 µm Ra on as-printed gold and about 1 µm after fine glass-bead blasting.
First step
Buy or cast 20 identical plain brass bands, measure their roughness, and record 10 teleoperated demonstrations of pre-polishing one band.
Done looks like
20 filmed attempts on 20 identical cast rings, with Ra and time for each ring published beside the same figures for a bench jeweler, and the robot matching the jeweler's Ra on at least 15 of the 20 with every dropped or marked ring counted.