meet USPS performance specifications in 1980, when USPS initially decided to use single-line OCRs, and that test results on the prototype multi-line OCRs were not available until April 1983, after USPS had decided to purchase single-line OCRs.
OTA did not itself investigate and has reached
no conclusions on the OCR that, as of May 1984, the
preponderance of evidence indicates that multi-line OCR performance is essentially equivalent to that of single-line OCR performance for processing 9-digit ZIP mail, and that multi-line performance is substantially better for processing 5-digit ZIP mail to the
For 9-digit ZIP (ZIP+4) mail, USPS performance data indicate that the Burroughs and Pitney Bowes single-line OCRs and the REI multi-line OCRs all correctly read, code,
and sort 98 to 99 percent analysis, USPS assumes 100
of OCR-readable ZIP+4 percent correct reading of
For 5-digit ZIP mail, both single-line and multi-line OCRs correctly read, code, and sort 98 to 99 percent correctly to 5 digits. However, only the multi-line OCR can read, code, and sort 5-digit ZIP mail to 9 digits.
USPS has estimated that the multi-line OCR can read, code, and sort 60 percent of
5-digit ZIP mail to 9 digits.
USPS believes that this 60 percent estimate may be high,
since USPS assumed that the total local metropolitan area would be included in the OCR computer directory. If the local directory has less than total coverage, the read-code-sort rate would be reduced. Also, USPS notes that adequate test data are not available on how the multi-line OCR performs two-stage encoding (e.g., placing a 5-digit bar code on a non-local letter at an originating post office and subsequently placing the