A Short History of the Barcode

Solving the checkout problem

In 1948, a supermarket executive visited the Drexel Institute of Technology in Philadelphia and asked a dean if the university could help develop a new system for encoding product information in a machine-readable format that checkout systems could read. However, the dean said the proposed project was outside the school’s curriculum and declined the executive’s request.

The executive was Sam Friedland, founder and president of the Food Fair supermarket chain headquartered in Philadelphia. Friedland had firsthand knowledge of the limitations of the existing supermarket checkout system, which explains why he went out of his way to ask for help in devising a better one. At the time, checkout systems had no way of reading product information and identifying items at the point of sale. This created issues:

• Employees had to put price labels on individual items to tell identify how much items cost. For hard-to-label items, such as fresh produce, price lists were used.
• Cashiers had to read each item’s price label and manually type in the price into a cash register when checking out a customer. There was no way to automatically pull the information into the checkout system.
• Retailers had poor real-time visibility into inventory and relied on manual counts, supplier invoices, and estimates to track inventory.

The first two issues increased labor costs and reduced customer satisfaction in supermarkets. For instance, customers had to wait longer to complete purchases because of how long the checkout process was. Moreover, manual price entry was prone to errors, but detecting those errors was difficult. Receipts were generally unitemized, so customers only saw a final amount and never saw how much they were paying for each item.

Friedland wanted a solution that could enable checkout systems to automatically identify products. If a product could be automatically identified at checkout, using something like a product identifier, cashiers wouldn’t have to key in prices, nor would employees have to place price labels on items. The product’s information could be looked up electronically using the identifier and pulled from the retailer records into the register during checkout.

Such a system would slash labor costs, improve worker productivity, and make shoppers happier. It could also potentially inventory management for retailers and provide real-time sales data that could be used to optimize pricing and stocking decisions. However, this system could only come to life if product information was encoded in machine-readable format, i.e., a format that a machine (or more accurately, an electromechanical system) could interpret.

From Morse code to barcodes

The dean declined Friedland’s request to build a solution to the supermarket checkout problem, but the story doesn't end there. A student named Bernard Silver overheard the conversation and decided to work on the problem. He later got a colleague Joseph Woodland, involved in the project and they spent the next couple of months searching for a solution.

The first prototype hey created used patterns printed in special ink that glowed under ultraviolet (UV) light. This prototype was abandoned because the ink was too expensive and difficult to print. However, Woodland had a conviction he could solve the problem and kept working on it.

Shortly after, in late 1948, Woodland decided to work on the project full time and give it full attention. He dropped out of Drexel, sold some stocks to tide himself over, and left Philadelphia to live with his grandparents at Miami Beach. It was during this period that Woodland had the breakthrough moment that led to the invention of the barcode.

The breakthrough moment unfolded in the most unlikely setting: at the beach.

Woodland was drawing in the sand when he noticed that the shapes he drew, dots and dashes, were similar to the symbols used to represent signals in Morse code. In Morse code, dots represent short signals and dashes represent long signals. To communicate in Morse senders transmit signals, using physical media like sound, light, radio waves, and electrical current, following specific timing rules. Timing rules determine how senders encode and receivers decode messages based on the duration and spacing of signals.

Woodland extended the physical marks downward by running his finger from each one and drawing parallel, albeit uneven, lines. The lines were uneven because the marks had different sizes: dashes were wider than dots, so dashes created wider lines when extended downward and dots created thinner lines. The spaces were similarly uneven: larger spaces between the marks created large spaces between the lines and and thinner spaces between marks created thinner spaces between lines.

Because the lines and spaces were formed by extending the visual representations of short and long signals, Woodland essentially found a way to encode information as it was done in Morse code. However, where the original Morse system encoded information temporally, using differences in the duration of each signal and the length of the time gap between consecutive signals, the new system encoded information spatially. Specifically, it used the physical space occupied by each line and the space of the gap between the lines to encode information.

An invention too early for its time

The second prototype Woodland and Silver developed was based on Woodland’s Morse code-inspired “bar code”. It consisted of:

• A barcode: A rectangular pattern of parallel lines and blank spaces that encoded information in the width and arrangement of physical marks.
• An optical scanner: A device that converted light reflected by physical objects into to electrical signals for machines to process.

The barcode pattern encoded a product identifier that could be printed on a product and used to identify it at checkout. To make the information machine-readable, the barcode pattern was printed in contrasting colors: black for lines and white for blank spaces. Because white reflects more light than black, a scanned barcode pattern produced different amounts of reflected light. The scanner had a photosensitive component that measured the differences in reflected light and converted those fluctuations into electrical signals. Those electrical signals would then be decoded into usable information by a machine or electromechanical system, such as a computer.

By the time Woodland and Silver patented the design in 1952, they had switched to a bullseye barcode pattern. This version used concentric circles of black lines and white spaces, encoding information in the size of the rings and length of the spaces between each ring. It was also easier to scan from any direction compared to the rectangular barcode that required scanning in a particular direction.

However, even this prototype proved impractical and too early for its time. For instance, the light source used to illuminate the barcode during scanning was a heavy 500-watt incandescent bulb, and the entire apparatus was the size of desk according to historical accounts. More importantly, the prototype couldn't be used in stores because computers were still too large (early computers like ENIAC weighed several tons and occupied the equivalent of several rooms) and extremely costly to purchase and operate.

Woodland and Silver used an oscilloscope to prove that scanning a barcode created signals that could theoretically be decoded into a product identifier and used to read information associated with the scanned item. However, they could not prove it would actually make it possible for a checkout system to automatically pull a product’s information by searching digitized supermarket records. That required a computer could transform electrical signals created by scanned barcodes into product identifiers and perform electronic record lookups to find information associated with scanned items, such as prices and descriptions.

No supermarket could afford a massive computer system, so the proposed barcode-powered system was essentially dead on arrival. Woodland and Silver sold the patent to Philco, a major manufacturer of consumer electronics for just $15,000, though the patent changed hands again when Philco sold it to Radio Corporation of America (RCA). By then, Woodland had joined International Business Machines (IBM), hoping to leverage IBM’s technical resources to bring his vision for the barcode to life.

The Universal Product Code

In the decades following World War II, particularly through the 1950s and 1960s, the retail industry experienced increased revenues, with sales exceeding $100 billion annually, as the combination of strong postwar economy, growing suburban population, and increase in consumer credit boosted consumer spending. However, increased shopping activity worsened a problem that bedeviled retailers for years: lack of product identification.

There was still no way to identify products being bought by customers at checkout. As such, employees still had to attach price labels on thousands of items and cashiers still had to manually enter prices. The problems of manual checkout that we identified previously, such as higher labor costs, customer dissatisfaction, and poor inventory tracking, became more visible and pronounced as more people shopped.

Desperate for a solution, a consortium of retailers and packaged-goods manufacturers created a committee to develop and standardize a product identification system. The committee came up with the Universal Product Code (UPC), a 10-digit string that identified a product’s classification and manufacturer. Each item would have a unique UPC number that checkout systems could use to identify it and pull relevant information from a database, such as the item price and description.

The committee invited companies to submit proposals for a machine-readable label that would carry the UPC number and be scanned at checkout to identify the labeled item. The label had to be:

• Small and neat: It must be no more than 1.5 square inches.
• Cheap to print: It must be printable with with existing technology
• Omnidirectional: It must be scannable from any direction at high speed.
• Highly accurate: It must have fewer than one in 20,000 errors when scanned.

The barcode was the natural format for the Universal Product Code. In fact, Woodland and Silver had designed it precisely for the purpose of encoding a product identifier that could be decoded and read by an electromechanical system such a computerized checkout. Woodland now worked at IBM and became involved with the company's effort to develop the UPC barcode.

However, George Laurer, who was heading the UPC barcode project, refused to use the bullseye barcode Woodland and Silver had patented years earlier. According to IBM’s own corporate history, Laurer opposed the bullseye pattern because it was difficult to print on products; plus, the ink was prone to smearing, which made it difficult for scanners to read the code properly. Instead, he opted for a rectangular pattern of parallel lines and spaces, which was what Woodland and Silver used originally before adopting the bullseye pattern.

Nevertheless, the new rectangular barcode design preserved the underlying idea from the prototype Woodland created with Silver decades ago. Lines and spaces were printed in black and white colors, respectively, with information encoded in the widths and arrangement of the printed shapes. The team also built a scanner that used a laser to illuminate the pattern and a light sensor to convert the reflected light into electrical signals that a computer converted into a 95-bit binary pattern representing the UPC number encoded into the barcode.

To prove the barcode format’s accuracy, Laurer had a softball pitcher on the team throw items with barcodes printed on the packaging at high speed over a scanner. The scanner read every barcode correctly, convincing any reminding skeptics inside IBM. The design met the other requirements set by the committee, too: it could be read from any direction and was small enough to be printed on most items without becoming unreadable.

On March 30, 1973, the Symbol Selection Committee tasked with selecting the UPC barcode format chose the design submitted by IBM. Laurer’s rectangular barcode beat out proposed formats from six other companies, including the bullseye barcode submitted by Radio Corporation of America (RCA), which now owned the patent and developed a functional version that it trialed at various stores in the early 1970s.

On June 26, 1974, the first commercial UPC barcode scan was completed at a Marsh Supermarket in Troy, Ohio. The scanned item was a pack of Wrigley’s Fruit Gum, bought by Marsh’s head of research, Clyde Dawson and scanned by employee Sharon Buchanan. It was the first time a cashier had checked out an item without keying the price into a cash register.

It was also a moment of vindication for everyone involved in the barcode’s development, from Joseph Woodland, Bernard Silver, and George Laurer to Sam Friedland, the supermarket executive whose frustration with the lack of product identification inspired Woodland and Silver to work on what would become the barcode. More importantly, it was a moment that signaled that the future of retail, particularly automated checkout, was here.

Adoption and takeoff

Retailers and manufacturers were initially slow to adopt the barcode due to the costs involved and uncertainty around industry adoption. Barcode scanners were costly to install and operate, and the computerized checkout systems they worked with, costlier still. Investing in barcode-powered checkout was justifiable only if enough items carried barcodes, since increased employee productivity and lower labor costs would offset the initial cost of installing the equipment and training staff to use it.

On their part, manufacturers would spend upwards of $200 million each year to print barcodes on consumer goods. The investment only made sense from a manufacturer’s perspective if most supermarkets and grocery stores had barcode scanning equipment and most goods, including those supplied by other manufacturers, carried barcodes. In such a situation, the cost of printing barcodes could be interpreted as a necessity to remain competitive.

However, this changed in the following years. Companies like IBM developed better, cheaper scanners and offered point of sale (POS) systems that integrated computerized checkout and reduced equipment costs. The introduction of the handheld barcode scanner made it possible to scan barcodes anywhere and led to retailers using barcodes outside of checkout; for example, barcodes were used in warehouses to track movement of goods through distribution networks and track inventory in real-time.

Barcodes would eventually go beyond retail and proliferate through every corner of commerce and industry. They were used in manufacturing, healthcare, transportation, libraries, and every sector where identifying and tracking physical objects was valuable. The barcode also inspired the development of other automatic identification and data capture (AIDC) technologies, including the Quick Response (QR) code, Radio Frequency Identification (RFID), and Optical Character Recognition (OCR); many other innovations, including biometrics, mobile imaging, and more, are also downstream of the barcode’s success.

Today, more than 10 billion barcodes are scanned per day across the world. An invention that came from an eureka moment at the beach has become one of the biggest and most valuable technological products of the last century. This is the power of technology: to harness the human potential for creativity and innovation to create tools and systems that solve problems, create immense value, and unlock progress.