The Barcode That Waited Twenty-Five Years

At just after eight o’clock on 26 June 1974, a cashier in Troy, Ohio, passed a packet of Wrigley’s Juicy Fruit chewing gum over a glass window. A red beam flickered across the label. Somewhere inside the checkout machine, a pattern of black bars became a number, the number became a database entry, and the till made a sound that now belongs to every shop on earth.

The packet cost 67 cents. It was not chosen at random. Nobody knew whether a code could be printed small enough to fit on chewing gum, so the gum was a useful test. Its passage through the scanner was staged, but the problem it solved was ordinary: a cashier no longer had to recognise the product, find its price, and type it in. The shop had learned to let the object identify itself.

That little rectangle had already existed for twenty-five years, though not in its familiar form. It began as a question from a supermarket manager, became a bull’s-eye of concentric circles, spent years waiting for the right light, and only then turned into the stripes we take for background.

The problem hiding in the checkout queue

In 1948, Bernard Silver, a graduate student at Drexel Institute of Technology, overheard a grocery executive ask the engineering department for help. Supermarkets were growing, but the checkout still depended on a person reading and entering information for every item. The same labour that slowed the queue also made stocktaking laborious. Silver brought the problem to Joseph Woodland, a friend and former Drexel student who was interested in mechanisms that could make marks readable by machines.

Woodland went to Miami Beach, where he was staying in his grandfather’s apartment, and thought about Morse code. Dots and dashes already represented letters as combinations of two marks. He imagined doing something similar with wide and narrow lines. According to Woodland’s later account, he dragged four fingers through the sand, then swept the lines into a circle. The circle mattered because a cashier should not have to orient every packet in exactly the same direction.

The story has the neatness of an origin myth, but the surviving patent shows the idea was more than a flash of beachside imagery. Filed on 20 October 1949 and granted on 7 October 1952, the patent described an optical classification system: patterns of light and dark marks would be attached to objects, illuminated, read by a photosensitive device, and translated into instructions for sorting or identifying them. One drawing shows a package travelling on a conveyor. Another shows the circular target, a bull’s-eye whose centre helped the machine find the code even when the package was badly positioned.

The inventors had designed an answer before the world had built a convenient question-reader.

A brilliant idea with a terrible lamp

Woodland and Silver made a crude apparatus, but it was not something a supermarket could place beside a till. The prototype used a 500-watt incandescent bulb and an oscilloscope. It occupied roughly the scale of a desk, generated a great deal of heat, and still had to turn changing reflections into usable information. An assessment that the invention was about twenty years ahead of its time was less an insult than a diagnosis.

The patent itself is striking because it treats the future as if it were already waiting in the stockroom. It describes labels on cans and packages, photoelectric sensing, conveyors, and the possibility of classifying goods into price categories. The code could contain more combinations simply by adding more possible line positions. The machine did not need to understand a tin of beans. It needed only to recognise a pattern and return the classification assigned to it.

That distinction is the quiet invention inside the invention. The label did not need to carry a price in ordinary human language. It could carry an identity, while a computer kept the changing price elsewhere. This made the same printed mark useful when the price changed, when the item moved to another store, or when the shop wanted to use the record for inventory rather than checkout.

But the system remained commercially stranded. Woodland and Silver joined IBM in 1951, and sold their patent to Philco in 1961 for $15,000. The idea survived; its first owners did not find a working market for it.

The circle loses to the stripe

The missing pieces arrived separately. Lasers supplied a narrow, intense beam that could sweep a label without the hot bulk of the original lamp. Smaller and cheaper computers made it practical to turn a pulse of reflected light into a product lookup. Printing also mattered. A code that worked in a laboratory could fail when ink spread, paper warped, or a label was printed quickly on an uneven package.

By 1970, grocery retailers and packaged-goods manufacturers had formed an industry committee to choose a common, machine-readable symbol. A standard was essential. If every manufacturer used a different code, a scanner would be a collection of private dialects rather than a language. Several companies proposed systems. RCA had a working bull’s-eye design, descended from the Woodland and Silver patent, and tested it in a Cincinnati store in 1972.

The bull’s-eye had one obvious virtue: it could be read from different directions. It also had a dangerous weakness. Imperfect printing could deform the circles and make the pattern unreliable. At IBM, engineer George Laurer argued for a linear symbol instead. His design used alternating black and blank vertical bars, arranged so a laser could read them from either direction while the whole mark stayed small enough for ordinary packaging. It was not a rejection of Woodland’s idea so much as its translation into the limits of mass printing.

The committee selected IBM’s proposal in 1973. The Universal Product Code was born as an agreement before it became a familiar object. Manufacturers had to print it, retailers had to install scanners and computers, and shoppers had to accept a machine interrupting the old ritual of putting a price on every item. The system faced a classic chicken-and-egg problem: a shop would not buy a scanner until enough goods carried codes, and a manufacturer had little reason to print codes until enough shops could read them.

The beep becomes a memory

Marsh Supermarket in Troy became one of the places willing to go first. On the night before the ceremony, staff marked hundreds of items while the scanner and computers were installed. The device now preserved by the Smithsonian shows what the beep concealed: a helium-neon laser, a rotating mirror, a glass plate, a photodiode, and a cash register connected to a stored database. The scanner did not know that the first item was gum. It recognised a signal; the database supplied the name and price.

The first scan was therefore less a triumph of eyesight than a new arrangement of memory. The product carried a key. The store carried the facts. Once enough stores and manufacturers accepted the arrangement, a packet could be counted as it entered a warehouse, sold at a till, or found missing from a shelf without anyone writing its identity down.

Woodland lived to see his sand-lines return in a form he had not drawn. Silver died in 1963, eleven years before the first supermarket scan. Neither man invented the modern checkout alone: the result required a grocery industry willing to agree on a standard, printers able to reproduce it, lasers able to read it, and computers able to remember what the pattern meant.

That is why the barcode is easy to overlook. It looks like information, but it is really a promise about where information lives. The stripes on the packet are almost empty. Their power comes from the invisible agreement behind them, the decision that an object may carry only a name while an entire network keeps its story. The beep at the checkout is the sound of a physical thing being admitted to a digital memory.