Industries / Industrial equipment and services / Industrial machinery / Fasteners
Threading
Threading puts the spiral ridge on a bolt or screw, today mostly by rolling the blank between hardened dies that press the thread into it, to standards that fix each thread's angle, pitch (the distance from one ridge to the next) and diameter so that parts from different makers fit together.
7
Milestones achieved, from 1841 to 2021.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1841
One thread for Britain
Joseph Whitworth, a Manchester machine-tool maker, proposed a standard system of screw threads in 1841, with a 55° thread angle. Grace's Guide calls it the first nationally standardized thread system and says that the railway companies, which had all used different threads, took it up and so made it widely accepted.
Credit
Joseph Whitworth
1864
A 60° thread for America
In April 1864 William Sellers of Philadelphia set out a simpler system in a paper at the Franklin Institute, with a 60° thread that ordinary machinists found easier to cut and rules for pitch and nut sizes for bolts from 1/4 in to 6 in. An Institute committee endorsed it on 15 December 1864. It did not win out until the 1880s, when it became the United States Standard thread.
LessonA thread standard spreads only as fast as precise taps, dies and accurately sized bar stock become available.
Credit
William Sellers; Franklin Institute
1948
Allies agree on one thread
TIME reported in October 1948 that mismatched British and American threads had been "nearly a calamity" in the Second World War: a B-29 bomber held 225,000 bolts, and a nut lost from one in Britain had to be replaced from the US. Under the new Unified thread, the US gave up the Sellers profile for one modified from Whitworth's, and Britain gave up the 55° angle for 60°. Representatives of Canada, the UK and the US signed the accord on 18 November at the US National Bureau of Standards.
LessonMismatched standards cost most when parts must cross borders in a hurry.
Credit
the standards bodies of the United States, the United Kingdom and Canada
1965
Screws that form their threads
Harvey F. Phipard Jr. of Research Engineering and Manufacturing (REMINC) patented a screw with a rounded three-sided cross-section (US 3,195,156, filed 1961, granted 20 July 1965) that presses a thread into a plain hole instead of cutting one, so it makes no chips and needs less torque to drive. The design is sold as TAPTITE by licensees such as ITW Automotive.
Credit
Harvey F. Phipard Jr.; REMINC
1970s
America's own metric system fades
In the early 1970s Ford, GM and Chrysler committed to metric designs, and US fastener committees answered with their own "Optimum Metric Fastener System", which allowed only coarse threads and proposed a spline head in place of the hexagon. Most of its ideas faded. The Industrial Fasteners Institute, where the work began, wrote in 2011 that it "just added another metric fastener standard option," making things more complex, not simpler.
LessonA national variant of a world standard adds a third choice instead of replacing the other two.
Credit
Industrial Fasteners Institute; the ASME B18, ASTM F16 and SAE fastener committees
2001
Germany retires its own standards
Germany's standards body, DIN, withdrew its fastener standards in 2001 in favour of ISO's, as recorded in DIN 918 Supplement 3; the Industrial Fasteners Institute calls this the biggest endorsement of ISO fastener standards. Parts made to the two are mostly interchangeable, but DIN hexagon heads on M10, M12 and M14 bolts are 1 mm larger across the flats and need a different spanner.
Credit
DIN; ISO technical committee TC 2
2021
Rolling order enters the code
Rolling a thread leaves its surface under compression, as shot peening does, which slows fatigue cracks. Johan Maljaars of Eindhoven University of Technology and TNO and Mathias Euler of Brandenburg University of Technology evaluated a few thousand fatigue tests and proposed separate design values for bolts in tension: 71 MPa for threads rolled after heat treatment, 56 MPa for threads rolled before it, and 50 MPa for hot-dip galvanized bolts or cut threads, each the stress range that 95% of bolts survive for 2 million cycles. The values went into the draft revision of the European design standard EN 1993-1-9.
LessonThe same steel earns a different design value depending on whether its thread was rolled before or after hardening.
Credit
Johan Maljaars; Mathias Euler
Watch for
Rolling threads on hardened bolts
Videx Machine Engineering of Israel showed thread rollers for pre-hardened fasteners at wire 2026. It says they monitor feed to 0.01 mm, set the alignment of the two dies by servo to "99.9% accuracy," and that computer-controlled feed alone can extend die life by 40% to 50%. No outside test is cited.
Achieved whena fastener maker or test lab publishes die-life counts on the same parts with and without the feature.
Claimed by
Videx Machine Engineering, March 2026
Open
01
Thread size from a photo
A blind test of the apps and AI models that claim to identify a loose bolt's diameter and pitch from a phone photo, using at least 500 fasteners, metric and inch, coarse and fine, each measured with thread gauges and photographed next to a printed scale card.
The briefCredit · unclaimed02
Bolt fatigue data, opened up
An open table of individual fatigue test results on threaded fasteners, each row giving steel class, diameter, pitch, whether the thread was rolled before or after heat treatment, preload, any later baking temperature, stress range and cycles to failure, with a link to its source, plus a model fitted to it that predicts the gain from rolling after heat treatment.
The briefCredit · unclaimed
Open 01 · the brief
Thread size from a photo
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A blind test of the apps and AI models that claim to identify a loose bolt's diameter and pitch from a phone photo, using at least 500 fasteners, metric and inch, coarse and fine, each measured with thread gauges and photographed next to a printed scale card.
Why it is still open
App listings promise this without figures; Screw Identifier – Bolt ID, released in February 2026 and since removed from the US App Store, claimed "high accuracy". The one research test found, by Toshit Jain and colleagues (2023), used a fixed camera rig and reported 98% correct identification on its own set, with pitch unreliable on small bolts. No public test set exists for comparing tools.
Built with what exists
Phone cameras; a printed reference card; screw pitch gauges and go/no-go thread ring gauges; OpenCV and open vision-language models.
First step
Buy a mixed bolt assortment, gauge 100 bolts and photograph each one with three phones.
Done looks like
Photos and gauge readings for 500 fasteners released openly, with accuracy for diameter and for pitch reported separately, by size band, for at least two apps and one open model.
Open 02 · the brief
Bolt fatigue data, opened up
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open table of individual fatigue test results on threaded fasteners, each row giving steel class, diameter, pitch, whether the thread was rolled before or after heat treatment, preload, any later baking temperature, stress range and cycles to failure, with a link to its source, plus a model fitted to it that predicts the gain from rolling after heat treatment.
Why it is still open
Maljaars and Euler's appendix lists test series only by source, grade, diameter and preload, not the individual results. Studies that vary preload (University of Iowa, SAE paper 2005-01-1321, class 12.9 bolts) and heating after rolling (Metalac, SAE paper 2003-01-3657, 140 to 230 °C, as coating cures and hydrogen-relief bakes require) sit behind paywalls, and API sells its own fatigue data for low-alloy steel bolts as a technical report (API TR 21TR2).
publications.tno.nlsaemobilus.sae.orgsaemobilus.sae.orgknowledge.bsigroup.com
Built with what exists
The sources Maljaars and Euler cite, many of them public reports and theses; WebPlotDigitizer to read curves from figures; an AI document reader to pull out tables, with every number checked by hand (the Note "Seven more flagged" covers an AI reviewer of long documents); Python.
First step
List the sources in Maljaars and Euler's Table A.7 and find which ones are free to read.
Done looks like
At least 1,000 individual results from at least 15 sources published with code, and a model that predicts the fatigue strength at 2 million cycles of test series it was not fitted to, within a stated error, compared with the 71, 56 and 50 MPa design values.