ERMA: the starting point for banking automation

by Marius Marinescu / 14 September

Bank of America in the early 1950s decided to automate their rapidly expanding check handling business. Can you imagine a time when you could take a piece of paper of any practical size and color, hand write your bank name, the payee, the amount, and add your signature (maybe legible) and use that as your bank draft or your check?

Eventually this “document” would arrive at your bank for someone to process it. Bank of America had set up a chain of banks in California and their first problem was to determine, from your signature, to which branch the document should be forwarded to. (It was impractical to send the account summaries of each branch to all other branches on a daily basis.)


In any case, the system was large, shaky, error prone, tardy and labor intensive. A number of bank employees figured there had to be a better way, and their ideas were effective and deemed worthy of further study.


The Bank of America was good at banking, had deep enough pockets, but did not claim automation expertise. They hired Stanford Research Institute of Menlo Park, CA to design a system for them. (Stanford Research Institute was “requested” by Stanford University to not use their name, so the name SRI was chosen and is still used).


Among other problems that SRI addressed was the fact that there was no effective machine (computer) method of reading documents (OCR is still not reliable enough for financial transactions). Ken Eldredge of SRI invented the MICR method of encoding and reading data from documents. This method prevailed over other competing methods and the American Banking Association finally adopted it.


At the same time, transistors became generally available for practical computer use, and SRI proposed a system using these new transistors instead of the vacuum tubes of the era. General Electric prevailed in suggesting general purpose computers instead of hard-wired special purpose logic, which it designed, built and programmed.


Machines for encoding documents with MICR, as well as machines for reading/sorting documents had to be developed. SRI made a suitable prototype that was promising enough for the Bank of America to want up to 36 commercial versions.


SRI did not want to get into the manufacturing business, so Bank of America requested major computer manufacturers to bid on making 30 banking systems for them based on SRI’s ideas and prototype.


To everyone’s surprise, the General Electric Computer Department, (a department that was non-existent at General Electric at that time) won the $31,000,000 Bank of America ERMA contract. General Electric corporate headquarters didn’t know of the bid and didn’t know of this new “department”. The same day the contract was signed, the bid team received a stern letter from G.E. president Ralph Cordiner stating that “under no circumstances will the General Electric Company go into the business machine business.”


The General Electric Computer Department chose Phoenix as headquarters, had a manufacturing establishment built, refined the prototype, built and/or OEMed the system elements, delivered the first system and passed acceptance tests on December 31, 1958. Some “tightening up” of the equipment and operating procedures was necessary to reach the design goal of 55,000 accounts/day.


Bank of America “encouraged” its clients and others to choose preprinted checks using the new MICR along the bottom edge. By March, 1959, the machine(s) were processing 50,000 accounts/day and on September 14, 1959, the Bank of America and General Electric presented 4 of the proposed 30 systems running in a transcontinental closed-circuit TV press conference. These 4 systems were capable of processing over 220,000 customer accounts in the Los Angeles area. The machines were using the newly developed standard E13B magnetic ink font that GE had developed which was more human readable. This E13B font is used on the bottom line of your checks today.


This is the E13B font which is the banking standard and first used on the ERMA.



The ink utilzed for the MICR characters can be magnetized, as part of the reading process, to create machine-readable information The alphas “A” to “D” mark the beginning of various fields, such as issuing bank number, customer number and dollar amount. Most fields are preprinted, but the dollar amount is printed after the customer writes it.


The ERMA system served the Bank of America well for 8 years (a long time for a commercial data processing system). Unfortunately, when the time for replacement came, General Electric was no longer providing banking-oriented systems or peripherals. The now obsolete GE-225 series had been popular with banks. The GE 4xx series was not suitable (could not respond to interrupts fast enough to handle the document handlers) and the GE 6xx series was too large and expensive for handling documents. General Electric took themselves right out of the banking business.


In any case, Bank of America went with IBM. Bank of America ordered the IBM 360/65 and took delivery in July, 1966 with conversion scheduled to be completed in December 1966. But conversion was deferred as a result of IBM’s continued delay in providing a multi-tasking operating system and severe tape drive problems. Situation had not improved by spring of ’67. Ray of hope came in late May ’67 with the successful “start” of the demand deposit conversion, the banks largest ERMA application. But damage had been done – delays had a direct economic impact on the bank’s profit amounting to $1,471,000. Total impact was estimated to be on the order of millions of dollars offset partially by IBM’s contribution in the form of paying all equipment costs, providing professional help valued at $2,700,160 and due to the fact that IBM maintained an account balance of more than $14,000,000 in the bank this entire period.


During the conversion, IBM had invested 66-man years of field engineers and 10-man years of tape specialists to make the tape system operable. After the conversion, IBM accepted the GE equipment as a trade in, allowing credit for the remaining book value of the ERMAs. A first for IBM, the allowance was kept confidential to avoid starting a trend. The IBM contract to replace the ERMA systems had a delivery penalty. IBM was to pay the ERMA maintenance until their system was up and running.


Despite the initial high-cost and technological set-backs, MICR was so successful in its design that it was adopted as the industry standard by the American Banking Association (ABA) in 1956. Bank of America made MICR technology available to all banks and printers without royalty charges. In 1984, American Banker stated that “the development of the MICR line, which enabled checks to be sorted and processed at high speeds, has been recognized as one of the great breakthroughs in banking”.


Today, the MICR methodology remains the standard around the world.