How machines learned precision
glinscott.github.io31 points by glinscott a day ago
31 points by glinscott a day ago
> After repeated copying and correction, Maudslay made a master screw five feet long and two inches across, with fifty threads to the inch and a foot-long nut that engaged six hundred of them at once.
That's the stuff I like to read about the historical endeavors. A fifty-threads nut doesn't average the distance between the grooves quite accurately enough? Then let's make a nut with six-hundred threads on it, and so what if it's foot long.
Author here. This is a sequel to my beam engine article: https://glinscott.github.io/beam-engine/.
I started digging into Watt's challenges getting an accurate cylinder bore for his pistons, and it spiraled out from there :).
The article covers some of the key techniques developed during the industrial revolution to measure and make incredibly precise parts. There are a lot of animations and interactive figures to explore how it works.
Very cool! Though my ChemEng degree obligates me to say that the animation in the "steam" section should really show boiling all the time if the pressure gauge is indeed showing the pressure in the container. Though in reality you'd never heat water at below atmospheric pressure in a real engine, if you did it would be boiling at any temperature at or above the triple point, and below the critical point, so long as there is no air in there.
Interesting that many of the early steam mills used a steam engine to pump water into a water tower, which then drove a conventional water wheel to power the millworks.
Yes, until they were able to get smooth rotation out of the engines, the water wheel was a much smoother source of power. Especially for looms and weaving machines, stable, smooth power was critical. "Power in the Industrial Revolution" by Richard L Hills is a deep exploration of the history there :).
How was AI used when writing the article? Not implying anything, I haven't started reading it (yet).
Certainly a fair question!
The text was hand written, but reviewed by AI and light edits were made. The CAD models that drive a lot of the animations were either sourced or built based on pictures of historical machines. As called out in the credits, the interactive figures were built by myself and Claude Fable 5 (mostly - a bit of Opus 5.5 in there now).
The article itself was built over about 4 months.
Armchair metrologist here. Did metalworkers back then use 16ths, or 12ths and 24ths? I'm familiar with the use of 1/12th lines ''' and multiples of 1/144th points ''''.