While all vendors of local library systems offer data entry using the scanning of barcode labels, a number of smaller libraries and libraries developing systems locally continue to rely on keyboarding. In this article we discuss barcode technology, including benefits, options, and costs.
Basically, reading barcodes is fast and accurate as compared with the keyboarding of data. Data entry using a keyboard ranges from one character per second (cps) to several cps, depending on the speed of the operator--which in turn is affected by the complexity of the data. It takes several seconds to enter a typical item identifier. In contrast, a barcode of up to 20 characters can be scanned in a fraction of a second. As to accuracy, a study by Bell Telephone Laboratories documented that uncorrected keying errors in keyboarding range from .42 to .48 percent of the total keystrokes; that's about one error to every 208-230 characters typed. This compares with one error for several million characters entered by barcode scanning. A controlled comparative study by Datalogic documented errors per 3 million entries to be 10,000 with keyboarding, 300 using OCR, and one using barcodes--specifically Code 39 symbology.
Among the different codes available are the following:
2 of 5 Codes--A code originated in the late 1960s for use in warehouse systems. The code also is used to identify airline tickets. This is a very simple code in which the information depends on the width of the bars. Spaces between the bars are there merely to separate the bars. Since the bars alone contain the coded information, the 2 of S code is categorized as a discrete code.
Interleaved 2 of 5 Code--A code similar to the 2 of 5 codes except that the spaces between the bars do contain information. Warehousing and heavy industry, especially automotive, use this code widely. Bars represent odd-numbered digits, and spaces represent even-numbered digits. It is a self-checking code since every character has a built-in check to avoid errors due to printing defects, It is continuous rather than discrete since there is information in the spaces. The width of the wide elements ranges from two to three times that of the narrow.
Code 39--A code which provides for 44 data characters. Three of the nine elements are wide and the remaining six narrow. Each character consists of five bars and four spaces (nine total characters) in which two bars and one space are wide. Digits zero through nine are represented in the same way as in the 2 of 5 code. This code is also discrete and self-checking. This is a popular code with many applications, including the health industry and the U.S. Department of Defense. It is the second most widely used code in the library automation industry.
Codabar--Libraries and the health field put Codabar codes to wide use. It is the standard for use on blood bags. Discrete and self-checking, Codabar codes consist of four bars with three spaces. The complete barcode symbol consists of a stop/start character, the data characters, and another stop/start character. Since it is a variable-length code, it is versatile but limited to 16 different characters--the 10 digits; the period, hyphen, and colon; and the plus, slash, and dollar signs.
In order to read barcodes one needs a fixed or portable scanner and a decoder--usually a hardware/software combination that converts the barcode into ASCII characters. The scanner emits light which reflects back from the code to a photosensor inside the wand. The voltage produced by the photosensor and related electronics is proportional to the code's pattern. The black regions absorb light and the light areas reflect it.
Two factors are critical to a successful scan. First, there must be high contrast between the light and dark areas of the code, preferably contrast ratios of 80 to 90 percent. One of the major problems in printing barcodes on a library's own printer is that the contrast may not be great enough--thus read errors occur. The second critical point is the widths of the code segments. Wide bars and spaces are two, two and a half, or three times the narrow bars. For a successful read, the decoding unit must be able to distinguish a narrow bar or space from a wide one. Again, locally produced labels may not have the precision to assure accurate reads.
Most barcode scanning devices can, with only slight modification, interpret Codabar, Code 39, and most other barcodes. Libraries commonly use a portable lightpen or stationary laser scanner to read the barcodes. While laboratory tests have demonstrated that the laser scanner is much faster, the complete transaction tends to take almost the same amount of time as with a lightpen because the physical handling of the library materials and the interaction between the staff member and the patron are the most significant factors in determining the "throughput" of a charging device. Staff who have not been taught to use a light-pen properly often have to scan a label more than once. The action must be smooth go the entire length of the barcode label, and the lightpen must be at no more than 40 degrees from the perpendicular. Also, the lightpen points must not be allowed to bounce on the counter or floor.
Once the code is read, the software determines how to handle the data. If the reads are not accurate and the quality of the labels is known to be high and even new lightpens properly used require multiple scans, the most likely source of the problems is the software. Virtually every off-the-shelf library automation software product supports Codabar and Code 39. Off-the-shelf software is rarely the source of problems with Codabar and Code 39 because there are so many users. While other codes also may be supported, it is not uncommon for a vendor to levy a "foreign barcode" charge of up to $10,000, and there is a greater chance of the software not properly interpreting the data.
All libraries which automate circulation, regardless of their size, should seriously consider using barcodes on materials and patron cards. They should be concerned about quality control. Therefore, a high quality printer should be used, or labels should be purchased commercially. (The cost for single labels is usually less than $.02 each.) Finally, the editors strongly recommend the use of a barcode which is widely supported by library automation vendors--Codabar or Code 39. Then if a library decides to move from one vendor to another or from a locally developed system to a commercially marketed software, it will not be necessary to change the code, nor to pay a "foreign barcode" charge.