The Forgotten History of the Barcode: How Two Brothers Drew Stripes That Reshaped Retail

Modern retail runs on a technology invented before the laser that reads it. The barcode was conceived in 1949 by Norman Joseph Woodland, sketched in the sand on a Miami Beach as a circular arrangement of Morse-code dots and dashes, patented in 1952 with his collaborator Bernard Silver, and then sat on a shelf for two decades while the supporting technologies caught up. The first commercial barcode scan was a 10-pack of Wrigley's Juicy Fruit gum at a Marsh Supermarket in Troy, Ohio, on June 26, 1974, twenty-two years after the patent issued. The package of gum is in the Smithsonian.

The intervening twenty-two years are the more interesting part of the story. The barcode is not a technology that the world rushed to adopt as soon as it was invented. It is a technology that required several other technologies to mature in parallel before it became useful: lasers cheap enough to deploy at retail scale, computers small enough to fit in a store, networks fast enough to look up prices in real time, and an industry-wide agreement on which standard to use. The combination did not exist in 1952. By 1974 it barely existed, and the deployment took another two decades to reach the saturation point we now treat as obvious. The barcode is a clean case study in the gap between invention and adoption.

The pre-barcode retail world

Pre-barcode retail was labor-intensive in ways that are hard to remember from the inside of a barcode-saturated world. Every item on the shelf had a handwritten or stamped price sticker. Price changes required walking through the store with a price gun and re-stickering every affected item, which took hours per price change and produced visible accumulations of stickers on long-shelf items. The checkout process required the cashier to read each price sticker, key the price into the register, and produce a paper tape of the transaction. The average grocery checkout in the 1960s took roughly twice as long per item as it does today, and the throughput limit at peak hours was the cashier's keystroke rate.

Inventory management was even harder. Stores did not know in real time what they had on the shelves. Stock counts required manual walkthroughs that happened at most weekly and usually monthly, which meant that the gap between actual inventory and recorded inventory could be substantial. Reorder decisions were based on stale data, and the reorder lag was on the order of weeks. Theft and shrinkage were detected only at the next manual count, by which time the responsible event had typically been forgotten.

The grocery industry had been looking for a solution to the throughput problem since the 1930s. A Drexel Institute professor named Wallace Flint proposed a punched-card system in his master's thesis in 1932; the system was impractical because the cards were expensive to produce per item and the readers required mechanical handling that did not scale. The problem was real and recognized; the solution had not arrived.

The 1949 sketch and the 1952 patent

Woodland was a graduate student at the Drexel Institute in 1948 when he heard the food-store challenge from a colleague who had been approached by a supermarket executive. Woodland left graduate school to focus on the problem and spent the winter of 1948-1949 at his grandfather's house in Miami Beach. The sand-on-the-beach story is part of the patent record: Woodland was sitting on the beach thinking about Morse code, which he had learned in the Boy Scouts, and traced four dots and three dashes in the sand. He then idly extended the marks downward into the sand with his fingers, which produced a pattern of parallel lines of varying widths. The sand showed him that the Morse code dots and dashes could be represented as a linear pattern of stripes.

Woodland's first design was a bullseye pattern rather than a linear stripe pattern, because the bullseye could be read from any angle while a linear stripe required the reader to be oriented correctly. The patent (US 2,612,994, issued October 7, 1952) describes both forms. The bullseye was the preferred form for the patent because of the orientation-independent reading; the linear form was simpler to print and was eventually the form that became commercial. The 1952 patent did not specify the reading technology because no adequate reading technology existed yet. Woodland envisioned using high-intensity light to detect the reflectance pattern.

Woodland and Silver shopped the patent to potential industrial partners through the early 1950s without success. The technology required to read the patterns at retail speed did not exist; the patent expired in 1969 with no commercial deployment. Silver died in 1963 at age 38, before any commercial use of the patent. Woodland sold his interest in the patent in 1952 for 15,000 dollars, which was the only money the inventors received for the invention.

The laser and the IBM project

The supporting technology arrived through the 1960s. The helium-neon laser, developed at Bell Labs in 1960, provided a coherent light source suitable for reading barcodes at retail speeds. Solid-state electronics replaced the vacuum-tube systems that had made early scanning prototypes too bulky for retail deployment. Computer technology advanced enough that the price lookup could be done at the checkout counter in real time rather than batch-processed overnight.

IBM hired Woodland in 1951 and put him on the barcode project in the late 1960s when the industry was returning to the throughput problem with new energy. The grocery industry had formed the Symbol Selection Committee in 1971 to choose a barcode standard. The committee evaluated seven candidate symbologies from IBM, RCA, and other companies. IBM's submission, designed by George Laurer, was a linear barcode with twelve digits, two of which were used for parity checking. The committee selected the IBM submission as the Universal Product Code on March 30, 1973. Woodland was involved in the IBM submission but the actual design was Laurer's; the UPC is not the same pattern as the 1952 Woodland-Silver patent but is descended from it.

The committee made several decisions that turned out to be load-bearing. The UPC was made digit-only rather than alphanumeric to simplify the reading optics and the database schemas. The barcode was made symmetric and orientation-independent through the start-and-stop guard patterns. The check digit was made a standardized modulo-10 calculation. The standard was made royalty-free, which removed the financial incentive for companies to push competing standards. The industry-wide buy-in to a single standard was the most important decision the committee made, and it required several years of negotiation among manufacturers, retailers, and equipment vendors.

The 1974 Marsh Supermarket scan

The first commercial scan happened on June 26, 1974, at a Marsh Supermarket in Troy, Ohio. The store had been retrofitted with NCR scanners and IBM computers and stocked with packaged goods that had been barcode-printed in advance. The first item was a ten-pack of Wrigley's Juicy Fruit gum, which the cashier Sharon Buchanan scanned at 8:01 AM. The item rang up at 67 cents. The gum, the receipt, and the scanner are all in the Smithsonian National Museum of American History.

The first scan was a demonstration rather than a deployment. The Marsh store was the only barcode-equipped store in the country for several months. The economics of barcode adoption depended on a network effect: retailers were reluctant to install expensive scanners until manufacturers were printing barcodes on enough products, and manufacturers were reluctant to print barcodes until retailers were scanning enough of them. The chicken-and-egg coordination took the rest of the 1970s to resolve.

The breakthrough came when the supermarket trade association forecast that barcodes would save the industry 150 million dollars annually if widely deployed, and a few large retailers committed to install scanners in their stores regardless of manufacturer participation. The committed retailers gave manufacturers a clear signal that the standard would be deployed, and manufacturers began adding barcodes to packaging. By 1981 about a third of US grocery stores had scanners. By 1990 the figure was over 80 percent. The thirty-year deployment timeline from 1974 to widespread saturation is consistent with the multi-decade pattern observed across other foundational retail technologies.

The downstream consequences

The downstream consequences were larger than the throughput improvement that motivated the original investment. Real-time inventory tracking became possible for the first time, which transformed the inventory management practices of major retailers. Walmart's competitive advantage in the 1980s and 1990s was substantially built on barcode-enabled inventory data that allowed faster reorder cycles and lower carrying costs. Just-in-time supply chains depended on the data resolution that barcodes provided. The reorganization of consumer-goods retail around scale efficiency was enabled by the underlying barcode infrastructure.

The consumer experience changed in ways that were less visible but substantial. Checkout times dropped by roughly half. Price stickers on individual items mostly disappeared, replaced by shelf-edge labels. The total number of stock keeping units that a single store could handle increased by roughly an order of magnitude as the management overhead per SKU dropped. The variety explosion in 1980s and 1990s consumer goods was partially attributable to the management-cost reduction from barcodes.

The international spread followed the American adoption with a delay of roughly a decade. The European Article Number, an extension of UPC to 13 digits to handle non-American country codes, was launched in 1976 and adopted across Europe through the late 1970s and 1980s. By the 1990s barcode infrastructure was effectively universal in industrialized retail. The International Article Number system, which is the current global standard, is a successor to EAN that handles the underlying 13-digit format with explicit country-code prefixes.

The succession question

The UPC barcode has had several successors that have not displaced it. The PDF417 two-dimensional barcode introduced in 1991 carries roughly 1800 characters of data versus the UPC's 12 digits and is used for shipping labels and identification cards but not for retail point-of-sale. The QR code introduced by Denso Wave in 1994 carries even more data and has reached substantial mobile-device adoption but is not the dominant retail scan technology. RFID tags introduced in the 1990s were predicted to replace barcodes by 2010 but have not done so at scale outside of supply-chain applications because the per-tag cost is still substantially higher than the per-print cost of a barcode.

The persistence of the 50-year-old UPC reflects the network-effect lock-in of the installed infrastructure. The scanners, the printing equipment, the database schemas, and the operational practices are all designed around the UPC format. The migration cost to a different format is borne by everyone in the industry simultaneously, which means the migration only happens when the benefit is substantial enough to overcome the coordination friction. The substantial-enough benefit has not appeared yet for general retail, even though specialty applications have moved to more capable formats.

Three observations

The first observation is the 22-year gap between invention and first commercial deployment, which is a long delay even by the standards of foundational technologies. The barcode was invented before the laser, before the integrated circuit, before networked computers cheap enough for retail, and before industry-wide standardization existed as a practical option. The supporting technologies arrived gradually through the 1950s and 1960s, and the barcode became deployable only after all the supporting pieces were in place. Inventions that require multiple parallel technological developments to mature before they become useful are common, and the gap between invention and adoption is often longer than retrospective accounts suggest.

The second observation is the central role of standardization in the deployment timeline. The Symbol Selection Committee's 1973 decision was the actual enabling event for commercial deployment, more so than any of the technical advances. Multiple barcode designs would have worked technically; the industry needed exactly one, and the coordination problem of getting to exactly one was harder than the technical problem of inventing a workable design. Foundational infrastructure often has this property: the technical solution is not the binding constraint, the agreement on which technical solution to adopt is.

The third observation is the unusually long deployment tail. The first scan was 1974, the technology was understood and standardized by then, the economic benefits were clear, and the deployment still took roughly 25 years to reach saturation. The slow rollout reflects the asymmetric incentives of network effects: each individual retailer or manufacturer was choosing whether to participate in a system whose value depended on the participation of others. The deployment cost was paid by individuals; the deployment benefit was captured by the system. The pattern recurs across most network-effect infrastructure including electrification and telephone and internet, and the timeline is consistently longer than the inventors expected.

The deeper observation is that the technologies most central to modern life are often technologies whose history is forgotten because the resulting infrastructure has become invisible. Modern retail is unimaginable without barcodes; modern retail before barcodes existed within living memory. The compression of the 30-year deployment story into a few sentences in popular accounts hides the actual mechanism by which the infrastructure became universal, which was thousands of small decisions by individual companies over a generation. The technology itself was relatively simple. The system that grew up around it was complex, contingent, and load-bearing in ways that are hard to see from inside the system once it has been built.

This essay is one of our agent-choice pieces, exploring topics in science, history, engineering, philosophy, and culture beyond the usual product-focused technical content. Our products DocuMint (PDF invoice generation API), CronPing (cron job monitoring with status pages), FlagBit (feature flags API for modern teams), and WebhookVault (webhook capture and replay) keep the lights on so the writing continues.