Cover of The Barcode: Line, Light, Ledger, a book by Ketineni RamaraoFrom the book The Barcode: Line, Light, Ledger
Chapters 9 and 18. Read the full book free as a PDF.

Ninety-five modules

A UPC-A symbol is a rectangle exactly ninety-five modules wide, the module being the width of the thinnest bar. It opens and closes with a guard pattern, two thin bars bracketing a space; a center guard of five elements splits the symbol into left and right halves of six digits each. Every digit occupies seven modules, encoded as two bars and two spaces. The digits on the left use one family of patterns, the digits on the right another, chosen so the scanner can tell from the pattern alone which direction the label is travelling and which side it is reading.

What the twelve digits mean

In the original allocation: one number-system digit, five for the manufacturer, five for the item, and a final check digit. Modern GS1 company prefixes vary in length, so the boundary between company and item is not something a reader can reliably infer by splitting every number in the middle; the register at GS1 is the only authority on where a given number divides. Uniqueness is an administrative achievement, not a technical one: somewhere, someone has to keep the register of who owns which digits, and that register-keeper, GS1 and its national member organizations, is the least known and most load-bearing institution in the system.

The check digit, worked through

The twelfth digit is arithmetic, not identity. Take the number 012345678905 and ignore its final digit. The digits in positions one, three, five, seven, nine and eleven sum to 20; multiply by three to get 60. The other five digits sum to 25. The weighted total is 85; five brings it to 90, the next multiple of ten, so the final digit is 5. Every scan recomputes that sum, which is how the committee could write its requirement of fewer than one undetected error in twenty thousand scans and mean it. The limitation is honest: a single misread digit is caught, but swapping two adjacent digits whose values differ by five changes the weighted sum by exactly ten and passes.

The margins are part of the message

Two invisible parts of the symbol do as much work as the bars. The quiet zones, the mandatory blank margins on either side, tell the scanner where the sentence begins and ends; printing in them is among the most common causes of failure in the field. And the human-readable digits below are not decoration: they are the fallback that lets a clerk key the number when the label is torn, the moment the whole system admits that the database, not the mark, is the point.

What the symbol refuses to hold

Just as deliberate is what the UPC does not contain: no price, no description, no expiry date, nothing that can change. It is a license plate, not a dossier. The price lives in the store's computer, which is why the same can can cost different amounts in different towns without a single package changing, and why the symbol could spread into hospitals, warehouses and baggage halls without changing its shape.

The European thirteenth digit

Europe wanted the convenience without the dependence: a numbering system administered solely from the United States left no room for European companies to hold their own prefixes. George Laurer, the UPC's own designer, showed that the symbol had unused capacity: the left-hand digits could be printed in two parities, and the pattern of those choices could smuggle a thirteenth digit into the same bars. The thirteen-digit EAN was approved in December 1976, deliberately upward-compatible, so that wherever the European symbol was scanned, American UPCs would read as well; full reverse compatibility arrived only in 2005. Twelve countries founded the European Article Numbering Association in Brussels in 1977, with the Dutch grocer Albert Heijn as first chairman, and prefixes allocated by country organization solved the politics with arithmetic. The family kept the same grammar: UPC-E compresses eligible twelve-digit numbers by suppressing zeros; EAN-8 carries a separately assigned eight-digit identifier for small packages. When the committee chose the symbol in 1973 it was not choosing a drawing; it was choosing an alphabet.

Quick answers

What do the numbers under a barcode mean?

On a UPC-A, traditionally one number-system digit, five for the manufacturer, five for the item, and a computed check digit. Modern company prefixes vary in length, so only GS1's register can say where a given number splits.

What is the difference between UPC and EAN?

EAN-13 is the thirteen-digit European form of the same symbol grammar, created in 1976-77 so non-US companies could hold their own prefixes. It was designed so American UPCs read on EAN scanners; full mutual compatibility was formalized in 2005.

Can a barcode have a wrong check digit?

Only by error or fraud. The twelfth digit is computed from the other eleven by a fixed rule (odd positions times three, plus even positions, to the next multiple of ten), and every scan recomputes it to catch misreads.

Sources

Symbol anatomy per Chapter 9 of the book and the GS1 General Specifications. Check-digit example worked per GS1's manual method. EAN-13 parity trick, December 1976 approval and 1977 founding per the European historical record; 2005 compatibility per the GS1 timeline.

More from this book

Mechanism · From the book The Barcode: Line, Light, Ledger

How barcodes work

Everything a linear barcode says is said with one trick: time.

History · From the book The Barcode: Line, Light, Ledger

History of the barcode

The barcode was invented in 1949 and mattered in 1974.

Mechanism · From the book The Barcode: Line, Light, Ledger

How QR codes work

The most famous barcode in the world was invented as an internal tool for a car-parts company, in a country that needed something the American stripes could not give it.