UPC The Real Story

In early 1972 Joe Woodland was transferred to IBM Research Triangle Park as a planner for the development of a super market scanner and bar code label.  The job of a planner is to survey the potential customers and determine the product requirements to assist the engineers in their development.  Woodland was introduced as having the original patent for a super market barcode.  Three IBM engineers had been developing a proposal for the UPC code and were curious about Woodland’s work.   When asked what his patent covered, he simply replied it was a printed barcode.

Today, a Google search for “Joe Woodland” yields a number of descriptions of his early work.  A History of the Bar Code by Tony Seideman, a freelance writer who lives in New York City, is one of those articles.

See: http://www.barcoding.com/Information/Barcode_History.shtml

The various articles for the most part tell the same story.  Rather than repeating the details, which can easily be found on the Internet, I will only state the results as they relate to the invention of the UPC.

                                 Woodland’s Invention

In 1949 at Drexel University, Bernard Silver learned that the president of Food Fair wanted a means to automate the process of grocery checkout.  Silver and Joe Woodland, after many failed attempts at their own expense, patented a bar code to be used for grocery identification.

There are at least two versions of what the code consisted of.  One version claims the code was a combination of the movie soundtrack and Morse code were the dots and dashes of Morse code were elongated into bars to be scanned.  Another described a code were each character consisted of a first bar to mark the start of a character followed by the presence or absence of three data bars.  These three bars represented three “bits” of data for an alphabet of seven characters.  They suggested that more bars/bits could be added to form a larger alphabet.

Woodland and Silver had attempted to build a scanner but the technology of the day did not offer a usable solution.

Woodland and Silver sold their patent to Philco, for a modest price.  It eventually found its way to RCA who in the spring of 1971 demonstrated a bulls-eye bar code and scanner.  There is some question whether Woodland did or did not propose the bulls-eye label, which could be scanned by a straight-line scanner from any angel.  The demonstration failed because the print quality available on the various grocery products was too poor to provide the required reliability at the necessary density.

                                The IBM Effort

In late 1970, IBM Research Triangle Park, NC had a large development effort called Consumer Transaction Systems or CTS.  It consisted of two separate efforts to develop point of sale terminals (POST).  One was for the Super Market industry, the other for Retail (all other retail stores).  The major efforts of each was to development Point of Sale or POST terminals which IBM engineers were well equipped to do.  Both efforts were managed by Paul McEnroe.

It was well understood that the super market product would require some form of checkout scanner something no one knew how to do.  As a gesture of good faith to upper management George Laurer, who had a good patent record, was placed in an office at the end of a hall with the assignment to solve the scanner problem.

The retail effort needed a hand held wand to read a bar code label containing much more information than the super market requirement.  For this, Jack Jones was brought in to manage the effort.

                             The Retail Bar Code

At the time there was a bar code known as Delta A.  Delta A was a delta distance code consisting of narrow bars printed with two different distances between them.  A wand could be passed across the bars while measuring the time between marks.  Each time was compared to the previous to determine if it was similar or different (example: about ½ or 2 times the previous time or about the same).  If the times were different it could be called a “1”, if the same a “0”.  With this code a gradual change in scan speed, hand speed, would have little effect on the reading as opposed to a method that required a precise speed.

In 1970 E. G. Nassimbene at IBM on the west coast and Jack Jones at IBM Research Triangle Park, NC independently invented a new “double density” bar code that became known as Delta B.  In Delta B the bars were replaced by bar edges, that is rather than using the distance from one bar to the other, Delta B used the distance from the leading edge of a bar to the trailing edge and then the distance from trailing edge to the next leading edge.  This, they believed, eliminated the “wasted” space for the bar width of Delta A which contained no information.

The Delta B code Patent Application Serial No. is 31,959.

Jones proposed the Delta B code to the National Retail Merchants Association or NRMA claiming it to achieve the twenty alphanumeric characters per inch they required.

Heard Baumeister, a mechanical engineer of extraordinary analytical skills, working in Advanced Technology, Personal and Item Identification, learned of the proposal and suggested to management that Delta B, in a real world, was likely not as good as Delta A.

In late 1970, to settle this question management formed a taskforce consisting of Heard Baumeister, Earl Miller, Albert Ruocchio, and Jack Jones.  They setup a lab to evaluate printer tolerance, and scanning accuracy.

Their findings were that printers had the obvious errors of placing bars slightly to the right or left of ideal.  They also found that the bar width could vary.  If a print bar struck a ribbon extra hard it would cause the bar to widen, if too lightly the bar would become narrower.  Similarly if the ribbon had too much or too little ink the bar would become wider or narrower.  In addition they determined that when bars widened the spaces got narrower and vice versa.

Baumeister generated a set of equations to apply the measured tolerances of printer and scanner to yield the maximum density allowable for a reliable product.  (Previous densities were determined intuitively.)

The conclusion of the taskforce found Delta B to be extremely sensitive to the bar width error since it compared the bar width to the space width.  Delta A was relatively insensitive to the bar width error.  As a result, for a reasonable printer, Delta A could achieve about seven alphanumeric characters per inch while Delta B could achieve only about five.  (NRMA required twenty.)

In late February 1971 the Advanced Technology, Personal and Item Identification department was disbanded and three of its engineers were transferred to CTS.

Heard Baumeister was placed in an office at the end of the hall next to George Laurer.  The Super Market Scanner effort now had two prolific inventors making upper management happier.

William Crouse and Glen Strickland were placed in Jack Jones’s department to develop the retail wand.  Crouse had a high level of patent achievement, was an exceptional circuit designer, and was well aware of Baumeister’s taskforce findings.

Crouse was surprised to find Jones was still pitching Delta B to NRMA.  When Jones persisted, Crouse, returning from a trip to the west coast, decided he had to do something.  On the plane, he invented a new bar code, Delta C, which compared the distances of leading to leading edges and trailing to trailing edges.  This was immune to the bar width problem and could achieve about fifteen characters per inch.  Crouse presented this to Jones the next morning but he had not made the actual character set assignments.  Later in the day Jones came to Crouse’s office with an invention disclosure in both their names.  Jones had drawn the symbols for the character assignments, a task that is rather straight forward, and felt he should be a co-inventor.  Crouse, more concerned with disposing of Delta B, signed the disclosure, confident Delta B was over.

Later, Crouse learned Jones was still pitching Delta B to NRMA.  Again, on a plane coming back from the west coast, he made an improvement in his code, using a fixed leading to leading edge that spanned most or all of the character as a reference to be compared to each of the other leading to leading and trailing to trailing edges.  This longer reference was less sensitive to printer tolerance allowing the code to achieve about twenty-one alphanumeric characters per inch, more than required by NRMA.  This time Crouse also recorded the full character set so as to leave nothing to doubt.   The next morning he filed an invention disclosure and informed Jones.  For “political” reasons, not explained to Crouse but approved by possibly two levels of management above Jones, Jones continues to pitch Delta B to NRMA.

The latest code improvement was combined with the first Delta C patent and finally issued as US Patent 3,723,710 in the names of Crouse and Jones.  Only the latest improved code, containing none of Jones’s ideas, was ever used.

See: http://v3.espacenet.com/textdoc?DB=EPODOC&IDX=US3723710

Shortly there after, Crouse arranged a transfer to an office at the end of the hall next to Laurer and Baumeister.  Now Paul McEnroe had three prolific inventers contemplating a super market scanner.

                              The Super Market Scanner

The trio was aware of the RCA bulls-eye scanner and its short falls as well as other published attempts.  Baumeister suggested the bulls-eye could be cut in half but even this left a label that was too large if readable.

Baumeister was a genius at analyzing mechanical devices but was very pessimistic about inventing new things.  Only when personally challenged did he become creative.

The small group at the end of the hall had very little support from management and offered no solutions.  Crouse had been thinking about a bar code wand that could be worn like a ring.  He modeled the device but was discouraged from buying ten dollars worth of parts to continue the effort.  The original parts had been purchased when Crouse was still in Advanced Technology.

                                   The UPC Label

On one occasion when Laurer was pleading with Baumeister to offer an idea Baumeister suggested an “X” scanner that could scan a bar code if the bars were printed with a length greater than the distance across the bars.  If a label, of any orientation, was not rotated as it moved across the “X” scan and the scan was fast enough, including the extra bar length and label speed, the “X” would make at least one pass across all lines.

Laurer returned to his office to document the idea.  After several days he returned to Baumeister to ask how to make an “X” scanner.  Making a straight line scanner using a laser reflected off spinning mirrors as used by RCA was well understood but an “X” scanner was something new.  Baumeister told Laurer a straight line scan could be made into an “X” by reflecting it through a book mirror.

Again, Laurer returned to his office but after some time went to Crouse for help.  How could a “book mirror” make an “X” from a line scanner?  After some thought he came up with two mirrors held at right angles to each other.  A straight-line scan when reflected from this mirror would trace an “X” on a flat surface.

Before Laurer completed his latest documentation Baumeister, with no outside prompting offered another improvement.  The bar code could be split into two half tickets which would allow the line lengths to be cut to nearly half the length since the length needed only to be longer than the distance across the bars that must be read in a single scan pass or half the label.

Baumeister’s first proposal reduced the label size from a bulls-eye of radius r (the distance across all the bars or marks) to a rectangle of width r and a height r+.  The bulls-eye has an area of 3.14 times r x r (pi times r squared).  Baumeister’s first label had an area of slightly over r x r or about one third the area of the bulls-eye.  His second proposal cut his label about in half or about one sixth the area of the bulls-eye.  Laurer now had a very attractive proposal to document.

About this time CTS was moved from Research Triangle Park, fifteen miles to a Raleigh facility.  Baumeister, using a technicality, was reassigned to another department and out of CTS.

Laurer needed a coding protocol to complete his proposal for the Super Market Committee.  He went to Jones, the local authority on bar codes, who recommended Delta B and provided the taskforce document that explained tolerance calculations.

After some time Laurer found that Delta B yielded a label still far too large to be acceptable.  Laurer again went to Crouse for help.  Not surprisingly Crouse suggested his Delta C code.  Delta C yielded four times the character density of Delta B and three times that of Delta A.  In area that was sixteen times smaller than Delta B and nine times smaller than Delta A.

Later, after the label had been accepted by the Super Market Committee, Laurer was pressured to extend the alphabet by adding the necessity to read one bar width on each of two characters, a violation of the Delta C teachings.  Some years later, Laurer admitted that those two characters even with exaggerated bar width differences, were far more likely to fail than all the others.  (Laurer may have mentioned that those characters were involved in a test label that contained the numbers 666. )

On another occasion Laurer asked Crouse for help in defining how the scanner could locate the label.  Crouse suggested a method using the wide margin of the label as compared to the fixed narrow marks at the beginning of the code.

Around this time, late 1971, Joe Woodland came on as the planner for the super market scanner.  His knowledge of commercial printing helped obtain sample labels for the formal proposal and demonstrations.  He also helped Laurer with the proposal effort.

                                The Ring Wand Demo

Crouse finished his model of the ring wand, which read the first UPC labels and demonstrated the robustness of the code and label during the demonstration to the Super Market Committee.

Crouse’s ring had a single, large diameter, plastic fiber optic fiber imbedded through it so that when the knuckles were curled under, the tip of the fiber could be drawn across a label.  The other end of the fiber terminated in a bracelet with the fiber end divided by a thin razor edge metal strip to separate a small “grain of wheat” light bulb on one side from a photo diode on the other.  The razor kept the light from reflecting off the end of the fiber back onto the photo diode.  The light from the bulb entered the fiber, traveled down and out the other end onto the label.  The reflected light reentered the fiber and traveled back to the photo diode.

Crouse also designed the electronics for the ring, amplifier, fast response threshold circuit, and digital counter.  A technician built the electronics, and a skilled young associate engineer programmed a computer to perform the decode function for the Delta C code.  The ring never became a product but played a key role in presenting the robustness of the UPC proposal.

The design of the scanner was moved to IBM Rochester, Minnesota for various reasons.  As a result, the presentation of the UPC proposal took place at that location.

Crouse, with his programmer, took his ring, hardware and software to Rochester to be connected to their computer.  The next day the rest of the IBM RTP (Research Triangle Park) group, Laurer, Woodland, McEnore and others arrived, as did the representatives of the Super Marked Committee.

December 1,1972 Laurer presented the UPC document, an impressive booklet with a large centerfold (two pages 8.5 by 11) of the label.  In addition to the text it also had a page showing a tabletop full of grocery items, all with UPC labels.

Following some of the presentation the group was moved to the lab for a demonstration.  Crouse scanned a few labels with his ring then scanned the large, two page centerfold label.  All with accurate scans.  The committee was amazed that the code could handle such a large range of label sizes.  Then Crouse flipped the page to the photo of the grocery items.  The labels in the photo were small and flawed due to the resolution of the photo and print quality of the fine lines.  Crouse scanned several of those labels with many scanning successfully.  Several committee members tried the ring with success.  A loop was offered for a closer look at the photo labels.  The extent of the line flaws became more evident and the robustness of the code was appreciated.  The UPC was born.

                                   The Aftermath

January 1, 1973 Bill Crouse transferred back to the Triangle to pursue a career developing satellite communications, speech processing and other activities related to signal processing.

About the same time Jack Jones transferred out of CTS and management to pursue other interests and about a year later left IBM.

Paul McEnroe later became the Lab Director of RTP.  When his main development effort experienced a major failure due to a “Bi-Modal Jitter” problem Crouse played a major part in identifying the solution but not before McEnroe accepted an executive position with another company.

Joe Woodland later moved on to other activities, proposing several leading edge technologies not related to bar codes.

George Laurer followed through with the UPC proposal and continued his career working in that activity.  He received a large monetary award from IBM for his contribution to the UPC.

                                   Final Notes

The efforts of Heard Baumeister and Bill Crouse were pretty much forgotten.

An IBM document titled: SOME “MOST VALUABLE” PATENTS/IVENTIONS dated 12/18/84 listed two patents for Rochester, Universal Code Structure by Crouse and Jones and Label Detection By Optical Scanner by Laurer and E. Moore.  It is not known, by this author, whether the Label Detection patent contained any of Crouse’s suggestion.  Baumeister probably never submitted invention disclosures for his inventions.  It is highly likely he would have received a patent for the “X” scannable  bar code proposal and his two half label improvement.

There were many responses submitted to the Super Market Committee from literally all over the world.  Nearly all were not viable.  The most feasible were probably optical character recognition of human readable labels but the technology was not available at that time, neither the scanner speed and resolution nor the computer power for the speed required.  The bulls-eye was feasible except for the size of the label required to achieve the reliability.

Without the inventions from IBM, both Baumeister and Crouse, and the persistence and proposal effort of Laurer there may not have been a UPC for many years.  Baumeister reduced the area by about 6 and Crouse reduced it by another 9 (a little less after margins etc.) for a total of nearly 50 to 1.  The following table shows the workable labels, available in the early 1970’s, with their sizes.

LABEL	DIMENSIONS	AREA
Bulls-eye with Morse Code	Large	Large
Bulls-eye with Delta B	12.0” diameter	113.10 sq. in.
Bulls-eye with Delta A	9.0” diameter	63.62 sq. in.
Baumeister 1st w/ Delta B	6.0” x 5.8”	34.80 sq. in
Baumeister 2 halves w/ Delta B	6.0” x 3.0”	18.00 sq. in
Baumeister 2 halves w/ Delta A	4.5” x 2.3”	10.35 sq. in
Baumeister with Delta C	1.5” x 0.9”	1.35 sq. in.