Industries / Industrial equipment and services / Industrial machinery / Bearings
Making rings and balls
This work turns steel into rings, rollers and balls and gives them their final size, shape and surface, mostly by grinding, then measures them to fractions of a micrometre (a thousandth of a millimetre).
6
Milestones achieved, from 1915 to 1975.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1915
Rollers ground without centers
Lewis R. Heim of Danbury, Connecticut, filed a patent on 6 March 1915, granted on 2 January 1917 and assigned to the Ball and Roller Bearing Company, for a machine that grinds rolls fed between a fast grinding wheel and a slower "regulating wheel" that sets how fast they turn, so the rolls need no centers to hold them. This is centerless grinding. The American Precision Museum says Heim needed rollers true to 0.00025 inch when the technology of the day held only 0.0005 inch, and that Cincinnati Milling Machine acquired the grinder in 1922; a 1921 advertisement shows a battery of them grinding rolls for roller bearings.
Credit
Lewis R. Heim; Cincinnati Milling Machine Company
1917
A ball factory taken over
The Hoffmann Manufacturing Company was formed in 1898 at Chelmsford, England, to buy Ernst Gustav Hoffmann's patent rights for making steel balls. In 1917 the Ministry of Munitions took it over as a National Factory because its owners were reluctant to expand for the demands of the aircraft industry, and by 1918 it made 230,000 ball bearings a month, according to Grace's Guide.
Credit
Ernst Gustav Hoffmann; the Hoffmann Manufacturing Company; the Ministry of Munitions
1943
Bombing the ball plants
About half of Germany's bearing output came from plants around Schweinfurt, and from 17 August 1943 US bombers dropped about 12,000 tons of bombs on the industry. The United States Strategic Bombing Survey found from industry records that Schweinfurt's September output fell to 35% of the pre-raid level, at a cost of 36 of 200 bombers in the first raid. Germany dispersed the plants, redesigned equipment to use other types of bearings and drew on stocks, and output was back to pre-raid levels by autumn 1944. The Survey found "no evidence that the attacks on the ball-bearing industry had any measurable effect on essential war production."
LessonMachine tools were damaged far less than the factory buildings around them, so a concentrated industry recovered once it was spread out.
Credit
the United States Strategic Bombing Survey, which reported the raids' effects in 1945
1949
Measuring how round
Roundness is how far a ring, roller or ball departs from a true circle. Taylor Hobson of Leicester built the Talyrond 1 in 1949; the company timeline in Grace's Guide calls it the world's first roundness measuring instrument. Only one was made at first, and customers sent their parts to Taylor Hobson to be measured until demand put it into production in 1954.
Credit
Taylor, Taylor and Hobson; Richard E. Reason, credited with the company's advances in measuring surface finish and roundness
1952
Gauging during the grind
Mario Possati founded Marposs in Bologna in 1952 to make what the company calls the first in-process equipment for checking parts during grinding: a gauge that measures the part while it is being ground, not afterwards.
Credit
Mario Possati, Marposs
1975
Ball grades agreed worldwide
ISO 3290, first published on 1 September 1975 by the International Organization for Standardization's committee on rolling bearings, gave steel bearing balls numbered grades. In grade 3, the best, each ball's diameter may vary by no more than 0.08 micrometres and its shape may depart from a true sphere by no more than 0.08 micrometres; grade 10 allows 0.25, and grade 200 allows 5. Member bodies from Australia to the USSR and the United States approved it, and the United Kingdom voted against it on technical grounds.
Credit
ISO Technical Committee 4, Rolling bearings
Watch for
AI cameras sort part defects
K2TECH told a bearing industry conference that its AI optical inspection system for bearing parts measures within 0.1 mm, wrongly rejects fewer than 1% of good parts, cuts false detections by up to ten times by telling real defects from harmless marks such as oil films or heat-treatment spots, raises overall equipment effectiveness by more than 8% and saves about four operators per line. It named no customers.
Achieved whena named bearing maker or an outside party publishes detection and false-reject rates over months of production, compared with human inspection of the same parts.
Claimed by
K2TECH (Sean Fang), at the Bearing & Power Transmission World Meetings 2025, reported in November 2025
Open
01
Grinding burn, measured openly
An open reference set of hardened 52100-type (100Cr6) bearing steel specimens ground on purpose at stepped feed rates, so that some are sound and others lightly or heavily burned, each measured by nital etching, Barkhausen noise, microhardness depth profiles and X-ray residual stress, with all raw data and photographs published and a blind test of which method catches which level of burn. Grinding burn is heat damage from grinding that changes the hardness of the surface layer, can leave internal stress and shortens fatigue life; nital etching is a chemical etch that makes burned areas visible; Barkhausen noise is a magnetic signal that changes with the steel's microstructure and stress.
The briefCredit · unclaimed02
Ball grades, checked by buying
A public test of loose bearing balls sold online by grade: lots labeled G10, G25 and similar, bought from many sellers and measured for diameter variation, departure from a true sphere and surface roughness against the ISO 3290 limits at a university or accredited metrology lab.
The briefCredit · unclaimed
Open 01 · the brief
Grinding burn, measured openly
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open reference set of hardened 52100-type (100Cr6) bearing steel specimens ground on purpose at stepped feed rates, so that some are sound and others lightly or heavily burned, each measured by nital etching, Barkhausen noise, microhardness depth profiles and X-ray residual stress, with all raw data and photographs published and a blind test of which method catches which level of burn. Grinding burn is heat damage from grinding that changes the hardness of the surface layer, can leave internal stress and shortens fatigue life; nital etching is a chemical etch that makes burned areas visible; Barkhausen noise is a magnetic signal that changes with the steel's microstructure and stress.
Why it is still open
Pitoňák, Mičietová, Neslušan and colleagues at the University of Žilina (Materials, July 2026) ground 100Cr6 at different infeed rates and measured Barkhausen noise, but their data are "available on request" because the files are too large to share easily. A review by Guo and colleagues at Harbin Institute of Technology (Materials, August 2026) reports that Barkhausen noise picked up mild burn on the raceway of a 1,650 mm wind turbine main bearing that nital etching could barely show, and says which areas need more study. Ducharne and colleagues (Sensors, 2023), who treated specimens of two other steels to give increasing levels of burn, found burn level, surface stress and hardness only weakly correlated, and call nital etching efficient but polluting. Searches of GitHub, Hugging Face and Zenodo in November 2026 found no public data set of graded grinding burn.
Built with what exists
A surface grinder and a metallography lab at a university, run by its own staff; specimens hardened to bearing specifications; a commercial Barkhausen noise analyzer or a lab-built sensor; a microhardness tester; X-ray diffraction residual stress measurement; Python for analysis; Zenodo for the data.
First step
Ask the Žilina group whether its raw Barkhausen and microhardness data can be released as the first block of the set, and list the specimens it already holds.
Done looks like
At least 60 specimens across at least five burn levels, each measured by at least four methods, with raw signals and images published; two inspectors rate the nital-etched specimens without knowing the grinding settings, and their agreement and each method's detection rate at each burn level are reported.
Open 02 · the brief
Ball grades, checked by buying
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public test of loose bearing balls sold online by grade: lots labeled G10, G25 and similar, bought from many sellers and measured for diameter variation, departure from a true sphere and surface roughness against the ISO 3290 limits at a university or accredited metrology lab.
Why it is still open
ISO 3290 ties each grade to measured limits, 0.25 micrometres of diameter variation and of departure from a sphere for grade 10 in its first edition. Sellers state grades with no measurements: VXB, for example, sells packs of 100 quarter-inch chrome steel balls labeled G10 and says the grade "ensures excellent roundness", with no measurements on the page. A search in November 2026 found standards, sellers' claims and hobby guides, but no independent measurement of balls bought online against their stated grade.
Built with what exists
Lots of balls from at least ten sellers; a roundness instrument such as a Taylor Hobson Talyrond or a Mahr MarForm; a laboratory length comparator in a temperature-controlled room; an optical profiler for roughness; Python for the statistics.
First step
Buy five lots of quarter-inch G10 balls from five sellers and ask a university metrology lab for a measurement plan and a quote.
Done looks like
At least 20 lots from at least 10 sellers, with 25 balls per lot measured for diameter variation and 5 per lot for form and roughness; each lot reported as passing or failing its stated grade, with all raw readings published.