Cash Register

The patent names James Ritty AND John Ritty. James is usually credited alone; his brother John is on the document.

Write down what an owner can verify at day's end without a register. Almost nothing — and that, not arithmetic, is what the machine addresses.

Make one disc with ten positions and a pawl that advances it by exactly one. Everything else is copies of this.

Add a spring-loaded pawl into notches. A wheel that stops half way is unreadable, and an ambiguous total is worse than no total.

Fit a single long tooth on the units wheel that engages the tens wheel once per revolution. That one tooth is the carry, mechanised.

Cycle the units wheel through zero fifty times. Carry is where every mechanical counter fails, and it fails intermittently — so test it a lot.

Two wheels must advance from one key press. Either allow enough travel for a cascade or fit a spring-loaded carry that stores and releases — this is the hardest mechanism in the build.

Make the amount key move the wheel a proportional number of steps. A five-cent key advances five; the key IS the number.

Link each key to a numbered flag that rises where the customer can see it. Ritty's first machine had no drawer and no receipt — the display was the audit.

Trip a striker on every registered sale. The sound tells everyone in the room that a transaction was recorded — a public, unfakeable signal.

Make the drawer release only when a key completes its stroke. This is the enforcement: no recording, no money.

Put the cumulative counter behind a locked cover the clerk cannot open. An adjustable total defeats the entire purpose.

The tally must not run backwards. A one-way pawl is what turns a counter into a record.

Ring up a known sequence, then compare the locked total against the cash. Any mismatch is a mechanism fault — find it before trusting the machine.

The patent. US 221,360, "Improvement in Cash Register and Indicator", granted 4 November 1879 to James Ritty and John Ritty of Dayton, Ohio. James ran a saloon and was losing takings to employee theft. The widely repeated account is that on a steamship voyage he saw a mechanism counting the revolutions of the propeller and realised the same idea could count sales. The brothers' first machine was nicknamed "Ritty's Incorruptible Cashier".

The first version had no drawer. It rang a bell, raised numbered flags the customer could read, and advanced a locked internal tally. Money still went into an ordinary till. The register did not physically guard the cash — it produced a record the clerk could not alter and a public signal the customer could not miss. Linking the drawer to the key stroke came later and is the enforcement half of the same idea. Framing it as a counting machine misses the point: it is an accountability device, and the customer is deliberately recruited as the witness.

The carry is the hard part. Any wheel-based counter is straightforward until a digit rolls past nine. Then one key press must advance two wheels, and at 999 it must advance four, all from a single finger's energy. Mechanical calculators solved this with cascading pawls or with spring-loaded stored carries released in sequence. It is the same problem Pascal and Leibniz faced two centuries earlier, and the reason mechanical adding machines stayed expensive and heavy right up to electronics.

What became of it. Ritty found running a saloon and a factory at once too much and sold the business in 1881. It was bought in 1884 by John H. Patterson, who renamed it the National Cash Register Company. NCR's real innovation was commercial — an aggressively trained direct sales force, sales quotas and territories — and its training methods shaped twentieth-century selling; Thomas J. Watson learned his trade there before leading IBM. The receipt-printing register followed, adding a copy for the customer, and the audit trail became something both parties held. Everything after that, up to the point-of-sale terminal on a modern counter, is the same idea with better storage.