Code_ The Hidden Language of Computer Hardware and Software - Charles Petzold [32]
But another way to look at the UPC is as a series of bits. Keep in mind that the whole bar code symbol isn't exactly what the scanning wand "sees" at the checkout counter. The wand doesn't try to interpret the numbers at the bottom, for example, because that would require a more sophisticated computing technique known as optical character recognition, or OCR. Instead, the scanner sees just a thin slice of this whole block. The UPC is as large as it is to give the checkout person something to aim the scanner at. The slice that the scanner sees can be represented like this:
This looks almost like Morse code, doesn't it?
As the computer scans this information from left to right, it assigns a 1 bit to the first black bar it encounters, a 0 bit to the next white gap. The subsequent gaps and bars are read as series of bits 1, 2, 3, or 4 bits in a row, depending on the width of the gap or the bar. The correspondence of the scanned bar code to bits is simply:
So the entire UPC is simply a series of 95 bits. In this particular example, the bits can be grouped as follows:
The first 3 bits are always 101. This is known as the left-hand guard pattern, and it allows the computer-scanning device to get oriented. From the guard pattern, the scanner can determine the width of the bars and gaps that correspond to single bits. Otherwise, the UPC would have to be a specific size on all packages.
The left-hand guard pattern is followed by six groups of 7 bits each. Each of these is a code for a numeric digit 0 through 9, as I'll demonstrate shortly. A 5-bit center guard pattern follows. The presence of this fixed pattern (always 01010) is a form of built-in error checking. If the computer scanner doesn't find the center guard pattern where it's supposed to be, it won't acknowledge that it has interpreted the UPC. This center guard pattern is one of several precautions against a code that has been tampered with or badly printed.
The center guard pattern is followed by another six groups of 7 bits each, which are then followed by a right-hand guard pattern, which is always 101. As I'll explain later, the presence of a guard pattern at the end allows the UPC code to be scanned backward (that is, right to left) as well as forward.
So the entire UPC encodes 12 numeric digits. The left side of the UPC encodes 6 digits, each requiring 7 bits. You can use the following table to decode these bits:
Table 9-1. Left-Side Codes
0001101 = 0
0110001 = 5
0011001 = 1
0101111 = 6
0010011 = 2
0111011 = 7
0111101 = 3
0110111 = 8
0100011 = 4
0001011 = 9
Notice that each 7-bit code begins with a 0 and ends with a 1. If the scanner encounters a 7-bit code on the left side that begins with a 1 or ends with a 0, it knows either that it hasn't correctly read the UPC code or that the code has been tampered with. Notice also that each code has only two groups of consecutive 1 bits. This implies that each digit corresponds to two vertical bars in the UPC code.
You'll see that each code in this table has an odd number of 1 bits. This is another form of error and consistency checking known as parity. A group of bits has even parity if it has an even number of 1 bits and odd parity if it has an odd number of 1 bits. Thus, all of these codes have odd parity.
To interpret the six 7-bit codes on the right side of the UPC, use the following table:
Table 9-2. Right-Side Codes
1110010 = 0
1001110 = 5
1100110 = 1
1010000 = 6
1101100 = 2
1000100 = 7
1000010 = 3
1001000 = 8
1011100 = 4
1110100 = 9
These codes are the complements of the earlier codes: Wherever a 0 appeared is now a 1, and vice versa. These codes always begin with a 1 and end with a 0. In addition, they have an even number of 1 bits, which is even parity.
So now we're equipped to decipher the UPC. Using the two preceding tables, we can determine that the 12 digits encoded in the 10 ¾-ounce can of Campbell's Chicken Noodle