Nearly every warehouse, store, and distribution center in the country runs on barcodes, and almost nobody knows where they came from. The short version involves a graduate student overhearing a conversation he wasn't part of, a man drawing lines in beach sand, and about twenty-five years of the idea going nowhere.
Counting things, the hard way
Before there was a way to read a product automatically, taking inventory meant closing the doors and counting by hand. A store shut down for a day or two, paid its staff overtime to walk the aisles with clipboards, and reopened with numbers that were already going stale. Grocery chains felt it worst, because they carried thousands of items that moved fast and were worth very little each.
In 1948, an executive from a Philadelphia food chain went to the dean of the Drexel Institute of Technology and asked whether anyone there could build something that captured product information automatically at the checkout counter. The dean said no.
Bernard Silver, a graduate student, overheard the conversation on his way past. He brought it to his friend Norman Joseph Woodland.
Morse code, pulled downward
The two of them tried ultraviolet ink first. It worked in a demonstration and failed everywhere else. The ink faded, and printing it was expensive.
Woodland left his teaching post at Drexel and moved into his grandfather's apartment in Miami Beach so he could work on the problem full time. What happened next is unusually well documented, because he told the story for the rest of his life. He was sitting on the sand thinking about Morse code, which he'd learned as a Boy Scout. He pushed four fingers into the beach and pulled them toward himself, and looked at the four lines they left. Dots and dashes, stretched downward into wide and narrow bars.
Then he swept his hand around in a circle. A bullseye could be read from any direction, which mattered if a cashier was going to drag a can across a reader without thinking about which way it was pointing.
They filed for the patent in 1949 and received it in 1952.
Twenty years of not much
Reading the code turned out to be the hard part. Practical lasers didn't exist yet, so Woodland built a reader out of a 500-watt incandescent bulb and a photomultiplier tube, the kind used to read the optical soundtrack running alongside movie film. It worked. It was also blindingly bright, hot enough to be a hazard, and it took up a desk.
That's roughly where the barcode sat for the next two decades. Woodland joined IBM in 1951 and kept pushing the idea internally without much luck. In 1962 he and Silver sold the patent to Philco for $15,000. Silver died the following year at 38, having never seen a barcode scanned in a store. Philco later sold the patent on to RCA.
The railroads got there first
The first industry to actually deploy automatic identification at scale wasn't retail. It was the railroads, who had a version of the same problem in a harder form: railcars travel the entire country, get loaned between lines, and routinely go missing on somebody else's track.
Sylvania built them a system called KarTrak, using colored reflective stripes on the side of each car. The Association of American Railroads adopted it in 1967. It worked well enough in a lab and poorly in the weather, because a label caked in road grime doesn't reflect anything, and the railroads abandoned it in the 1970s.
It wasn't a dead end, though. The engineer behind KarTrak, David Collins, left to start Computer Identics, and by 1969 his company had installed some of the first working laser barcode scanners, at a General Motors plant and a distribution operation in New Jersey. Lasers solved the reader problem that had stopped Woodland cold in 1949.
The rectangle that beat the circle
In 1971, RCA demonstrated a bullseye scanning system at a grocery industry convention. IBM was there, noticed the interest, and realized the man who had invented the thing was already on its payroll. Woodland was brought onto the effort, and an IBM engineer named George Laurer was given the job of designing a symbol that could survive a real printing press.
RCA and Kroger ran an eighteen-month bullseye pilot in Cincinnati starting in 1972, and it surfaced the flaw that decided the whole thing. Presses smear ink in the direction the paper is travelling. A smear across a bullseye cuts through every ring and destroys the code. Laurer's answer was a rectangle of vertical bars, printed so the smear runs along the bars rather than across them. The code degrades and stays readable.
That design was accepted as the industry standard on April 3, 1973, and it's the UPC still printed on nearly everything you buy.
On June 26, 1974, a ten-pack of Wrigley's Juicy Fruit was scanned at a Marsh supermarket in Troy, Ohio. It was the first commercial barcode transaction anywhere.
Slow, then all at once
Adoption crawled at first. By 1977, only about 200 stores in the United States had scanning systems, because the equipment was genuinely expensive and no single grocer benefited much until manufacturers printed codes on everything as a matter of course.
Then it tipped. Once enough stores had readers, printing a UPC became the default rather than a favor, and by 1980 thousands of stores a year were converting. The chains that went early found the savings weren't really at the register at all. They were in knowing what was on the shelf without shutting down to count it, which was the thing that Philadelphia executive had walked into a dean's office asking for in 1948.
Endpoint's own work with barcodes starts in 1989, putting scanners and labels onto production and warehouse floors for manufacturers and distributors, and connecting what gets scanned back to the books.
Woodland lived until 2012. He never made much from the patent, which had expired long before anyone found a good use for it. He did live long enough to watch four lines he'd drawn in the sand get scanned several billion times a day.
