Automatic Screw Machine

List, in order, the operations to turn one part: feed stock, turn to diameter, form the head, cut the thread, part off. Five actions, always the same order.

Cut a 150 mm disc of plywood. Mark its centre and drill it to fit the dowel rod.

With the protractor, divide the rim into five sectors sized to your five operations — not equal, but proportional to how long each one takes.

Cut the rim away in each sector to the depth that action needs. The rim height is now a graph of the cycle.

Push the dowel rod through the centre hole and rest the ends in two notched blocks so the cam turns freely.

Pivot a card-stock lever so its tip rides on the cam rim. The lever is your cutting slide.

Rotate the cam once slowly. Watch the follower run through all five positions and return to the start.

Record the follower's height at the centre of each sector with the steel rule. Those five numbers are the program.

File one sector 3 mm deeper. Turn the cycle again and record the new travel. You have just reprogrammed the machine without touching the machine.

Wrap string around the dowel in two opposite directions and pull each in turn. Spencer's claim replaces your two hands with a clutch that picks one of two counter-rotating pulleys.

Turn the cam ten times at a steady rate and time it. Divide by ten for the cycle time — the number a shop actually buys.

Now perform your five operations as five separate hand movements, reading the list each time. Time ten of those. The ratio is the machine's real argument.

Clamp the brass rod in the lathe chuck, mark it, release and re-clamp five times, checking run-out each time with the dial indicator. Automation is worthless if holding is not repeatable.

US 147,291, “Improvement in Machines for Making Metal Screws”, Christopher Miner Spencer (1833-1922), Hartford, Connecticut. Filed 17 December 1873, granted 10 February 1874. The claim is a clutch drum that shifts drive between two oppositely rotating pulleys, reversing the screw rod automatically so the dies engage and withdraw without a hand on a belt.

What this patent is not. Spencer's celebrated “brain wheel” is a cam drum carrying an entire machining sequence, fitted to the automatic turret lathe he built in Hartford in 1873. Popular accounts fold the two into one invention. They are separate mechanisms and only one of them is the subject of this document. Where sources merge them, the document wins.

Why the screw came first. Screws are small, are needed in enormous numbers, and each one requires an identical multi-step sequence — the ideal first case for automation. The class of machine Spencer's work founded is still called a screw machine even when it is making something else entirely, and its descendants are today's CNC lathes with bar feeders.

The deeper idea. A cam is a physical program: a shape that stores a sequence and replays it exactly, forever, with no memory and no operator. Every argument later made about numerical control — that the instruction should be separable from the machine and cheap to change — is already visible here, where changing the part means filing a new cam.

Spencer before machine tools. He patented the Spencer repeating rifle in 1860; several hundred thousand were made during the American Civil War. The interchangeable-parts discipline he learned in armouries is exactly what he brought to the screw machine.