My Life With Technology - Chapter 1.3

Check Processing

by Bill Holmes


Have you ever written a check? Have you ever wondered how it went through the system and eventually got back to your account?  This is my second attempt to document my experiences with check processing circa 1970. I got several paragraphs into this chapter before I realized that it was going to be too long and too involved. My objective with this series is to document dinosaur technology from the stone age, not explain how the banking system works. There will be some processing details documented, but not all the steps. Maybe someday I'll go into greater detail because I find it fascinating. I'll warn you before I post that article so the faint of heart can avoid it. So, let's begin.

As I mentioned in a previous chapter, I walked into the computer room of The First National Bank of Atlanta (FNB) for my first day (night) of work in 1968. Actually I walked into the fourth floor IBM computer room that day. I soon found out that there was a whole other computer room and supporting departments on the third floor. The fourth floor was mostly about IBM computers, tabulating (Tab), production control and keypunch. The third floor housed the check processing computer room and their supporting departments. Although called the check processing department or more commonly the transit department (transit) they obviously processed more than just checks. There were deposit slips and internally generated debits and credits. A few other applications, such as installment loan payments, were also processed. Regardless, the vast majority of items were checks.

In a way, we in the IBM environment were the client department or customer for some of the output from the third floor. They actually did most of the messy front end work for the checking account processing. At a bank, checking accounts are the biggest, highest volume application. Checks touched many departments from the front line tellers to the loading dock employees.

The third floor computer room and support departments was a whole different culture and environment than the fourth (IBM) floor. Their computer room was crowded, cluttered, noisy and dusty. Checks and forms and tapes were stacked everywhere. The fourth floor computer room was pretty clean with extra room and although often noisy it was nowhere near the decibel level generated by the check sorters. Only the tapes, disks and forms being used were out on the floor. The rest were either in the tape library, forms room or designated staging area. Dress codes differed too. The IBM computer operators wore company issued navy blue blazers with a department patch on the breast pocket. White shirts, ties, dress slacks and dress shoes completed the outfit. The transit guys didn't wear ties, too many places for them to get caught and cause strangulation, they didn't wear suit coats or company blazers and many wore tennis shoes. Wrangling checks all shift was a physical job. The sorters had to be fed and emptied constantly and the operators were on their feet most of the time.

Checks, deposits and other paper documents were processed by machines using Magnetic Ink

MICR Characters

Character Recognition (MICR). Those are the funny looking numbers and symbols on the bottom of checks, deposit slips and other documents. The ink used to print those characters contains iron particles that can be magnetized. Machines can read those magnetized characters just like machines can read magnetic tape or disks or the strip on the back of your credit/debit card. In 1970 the technology to read MICR characters was only about a decade old.

At FNB we used Burroughs MICR reader/sorters and computers to process checks. There were other vendors making similar equipment but Burroughs had a very large part of the market. I would say that Burroughs was the dominant company when it came to document processing. Those sorters were connected to Burroughs B 300 computers. In 1970, MICR reader/sorters could read, sort, endorse and capture items to disk or tape (online). They were also capable of sorting checks into numerical order without being attached to a computer (offline). The sorters large banks used for check processing ran in the 1,500 to 2,000 documents per minute (DPM) range. They all had at least 13 pockets and some a few more. I think there were two computers, four sorters and an assortment of tape and disk drives, card readers and printers in the third floor computer room.

Check Reader/Sorter on left

The transit department was all about getting checks in and out as fast as possible. A short and very abridged tutorial on check processing follows. Banks receive checks from all over the country and world. You might deposit a check from an employer or stock dividend or pension or even that Christmas check from Grandma that is drawn on another bank. Those checks have to get back to the issuing bank for the funds to be collected. The faster the check got to the issuing bank, the faster the funds were collected. It's true, time is money. Most cities and towns with more than one bank would exchange checks directly among themselves. For most out of town checks, member banks used the Federal Reserve Banks (Fed). There are 12 Fed districts some of which have branch locations in addition to the main bank. The Fed was the major check clearinghouse for the banks. Atlanta is a Federal Reserve city and the Fed was a couple of blocks away from FNB. Many smaller banks or banks outside Fed locations would send their checks to us to be cleared rather than deal directly with the Fed. There were many reasons for that arrangement. FNB would combine all those checks and include them with items we received in deposits or other transactions.

Microfilm Spool

Before any items were processed by the transit department they had to be microfilmed along with any adding machine or other listings that went with them. These were stand-alone manual units that could only be fed a handful of items at a time so it was labor intensive. The items were recorded on rolls of film that had to be developed just like your Kodak snapshots. Once the negative was produced that was the end. Prints were only made when needed either for internal bank reasons of if a customer requested a copy of an item. It was not easy to find the right spool of film to do research or print a copy. Any prints were made by microfilm reader machines and were actually more like a Xerox copy than a photograph print. FNB had an in-house department that developed the film. In later years the check reader/sorters could microfilm the items while they were being processed.

Single Pocket Proof/Encoding Machine

One of the check processing supporting departments was the Proof and Encoding Department. They were somewhat equivalent to the keypunch department. They put the dollar amount with MICR ink onto those items that didn't already have it. That was called encoding the item. Typically the first bank that handled an item was responsible for encoding them. The clerks in proof had a machine that encoded the items, kept count of debits vs credits and produced an adding machine listing. Large banks used single pocket proof machines. Smaller banks had multi-pocket machines since they typically didn't have high speed sorters to separate the items into the various categories. The proof operator would key the transaction amount and then a key to indicate the item type or pocket. The machine would print (encode) the amount, appropriate symbols and a transaction code if needed. A transaction (tran) code is a two or three digit number printed to the left of the amount to indicate the type of transaction. Tran codes were not put on checks from other banks. At FNB all debit tran codes were in 40's and all credits were in the 20's. A normal check was a 45 tran code or no tran code. A normal deposit was 25. Other numbers were used for items like non sufficient fund (NSF) charges or service charges. Once the items were encoded and balanced (proofed) they were sent to the Burroughs computer room.

Items would be held and aggregated until it was time to process them. That time was determined by various deadlines and schedules both internal and external. When it was time to make a run, the appropriate program was loaded into the B 300 and the sorter was fired up. As the items were run through the reader/sorter it would read all the MICR fields on the bottom. The items were also endorsed on the back with an FNB endorsement. Depending on the program the items were directed to different pockets. All the deposits would go to one pocket, all the checks drawn on FNB (on-us) to another, all checks drawn on the local bank down the street another, checks drawn on banks in the state another and so on. Of course there was a reject pocket for unreadable items. There were 15 pockets on the Burroughs sorters so there could be 14 different categories on each run plus rejects. If there were more than 14 categories, items for multiple categories or destinations would be combined into one pocket then a second pass would be needed to further break out the items. There were dozens of different programs depending on the type of items and time of day and/or day of the week. Item listings and totals were printed for each pocket and a master listing of all items. The master listing was used to balance the run. If you were lucky the sum of the master listing totals plus the manually added rejects matched the totals going into the run. We weren't lucky very often. If the run was out of balance you figured it out. There were numerous reasons to be out of balance. The input totals could be wrong, there could be missing or extra items, documents might stick together (piggyback), the machine could misread an item, a sorter operator might not clear a jam correctly and a thousand other reasons. The people who balanced the runs developed a real knack for it and could usually find the errors very quickly. At this juncture of my career it wasn't my job but sometimes those of us in Production Control would go downstairs to help out when work got backed up or if there was a particularly nasty out of balance problem. In my next career stop I got very good at balancing and will detail some of the tricks of the trade when I get to chapter 2.

Like I mentioned in the last chapter about credit card receipt processing, check processing also dealt with original source documents. They had to be returned to the customer in their statement or sent on to the issuing bank. Because of that, there were many methods used to repair damaged items or items with erroneous MICR encoding. We had stickers that would cover and erasing fluid that would erase the MICR numbers. Neither worked great. The stickers sometimes came off which caused jams or items to stick together and piggyback. The correction fluid worked best on newly encoded items. Even then, once the applicator got dirty you wound up with a magnetic smudge rather than a clean surface. For more seriously damage items there were envelops. The correction envelopes had a front that was made out of the same material that address windows in business envelopes is made out of. They had a regular paper back and bottom strip on the front. The problem here was that you just turned a single ply document into a three ply document. Check sorters were built to separate a stack of paper items. They had belts and rollers spinning in opposite directions and air jets to do this so sometimes they would crumple the three ply correction envelop item or rip the transparent front off.

The check sorters of 1970 were rather primitive compared to today's document processing machines. They were a kluge of rollers, belts, gears, springs, lights, electronics, air jets and several other parts and pieces. They were very high maintenance machines. Many parts were made to wear out and were replaceable by the operator. Other parts were more complicated to replace and required a repair technician on a scheduled maintenance time. Still other parts just broke on their own schedule. Despite all that, they were amazing machines. They could read, endorse and sort paper items of various sizes, thickness (paper weight) and condition at a speed of 2,000 DPM. This was a much tougher engineering challenge than manipulating punch cards. Checks got a lot more abuse than cards. They weren't just folded, spindled and mutilated. They were washed & dried, taped, torn, crumpled, stapled, glued and numerous other things some of which are unmentionable. We found blood, food, ketchup, gum and dozens of other foreign items on checks. Some of the substances we really didn't want to identify.

Burroughs made a card reader that had a similar feed mechanism to their check sorters. It could read cards that were in pretty bad shape. One night I was standing near it talking to a transit guy and mentioned that the cards were a mess. He said they weren't too bad and that you could run a sock through the card reader. After the cards were read, he switched the reader offline and ran his sock through it. He was right although I'm not sure how you originally discover that kind of thing.

The Fed had numerous deadlines throughout the day and night. They were also very strict about those deadlines. A minute late was the same as a day late or in the case of Fridays, three days late. Banks always wanted to get as many checks as they could in under those deadlines. It could mean the difference in getting credit for the checks a day or two earlier. That was money they could make interest on or save interest expense. Because we were only about two blocks down the street it was faster to take smaller bundles of checks to the Fed on foot. There were deadlines all through the 2nd and 3rd shifts. Now there were a couple of important deadlines that we always pushed as close as possible. The transit department would always have a couple of fast runners on staff. They were the designated foot messengers who ran to the Fed. It wasn't unusual to see one of these guys running through the lobby and out the door. For various reasons the doors to our building lobby that were closest to the Fed were locked after hours. The doors that were open all night were a block farther away on the other side of the building. That extra block could make the difference in a made or missed deadline. The lobby guards were instructed to unlock one of the doors on the side close to the Fed five or ten minutes before certain deadlines then lock it once the courier cleared the building. One night either there was a new guard or he was asleep at his post (not unusual). The courier hit the door at a full run and it was still locked. After some expletives he ran across the lobby and took the long route. We missed that deadline which cost the bank some money. The guards were much more conscientious after that night.

All during the day, checks drawn on FNB (on-us) and all checking account deposits were captured on tape. Those tapes were sent to the fourth floor IBM computer room to be used as input to the nightly checking account applications. Any unreadable items were sent to keypunch to create cards that were also input to the account processing. Of course there were checks and balances at each step. Eventually all the items for that business day had been processed and transit would send us final totals and give us the OK to start posting the checking applications. We couldn't start until the whole day's work had been proofed. That was called balancing the bank. It had to be to the penny. Now we were talking about a million checks a day totaling in the hundreds of millions of dollars. There were nights when that last dollar or dime or nickel or penny were damn hard to find, all while the clock was ticking. Everybody would pitch in to find the problem. We'd all go down to the third floor. I remember a few times when we were out of balance by a certain amount and someone would find an error for that exact amount. Everybody would give a cheer or give a sigh of relief only to soon discover that the error was going in the wrong direction. Now we were out of balance by twice as much. We always managed to find the error(s). As time went on the powers that be realized that we were holding up expensive machines, paying people overtime and missing deadlines for a penny or two. Eventually the guidelines were relaxed and we could start processing even when slightly out of balance. We still had to find the errors but could fix them with adjustments the next day. The irony of being required to balance to the penny is that it didn't guarantee that everything was correct only that there were counterbalancing errors. With the volumes we dealt with there were always errors.

Once the bank was in balance the check sorters had another job, they had to sort the checks and deposits for FNB into account number order. This was called fine sorting. Those items had to be sorted so they could be sent back to the customers with their bank statements. They would use whichever of the sorters were available and not being used for initial sorts. This process could be run with the sorters disconnected (offline) from the computers. They had various methods to break up work so they could use all four sorters if they were available. I don't remember how long our account numbers were but they were at least eight digits. That's a lot of passes through the sorters. For some of our larger customers like Georgia Power of Southern Bell, we would sort their checks into numerical order before we mailed or delivered their bank statement(s).

The Burroughs or third floor or transit or check processing department was a vital part of the automation of banking at FNB in the 1960's and 70's. They were mostly a data capture department that had big powerful computers (for the time) but did very little computing as we think of it. The computers were mostly used to control the MICR reader/sorters which in those days was pretty amazing. How can you read some printed characters on the bottom of a small piece of paper, capture the dollar amount, check the account number and determine which bank the check was drawn on. Then direct the sorter to put the item into the appropriate pocket. At the same time, print every item at least twice. Once for the category (pocket) the item was sorted to and once for the master list. If it was an on-us item write the needed information to tape. Check processing was not the most glamorous part of the emerging computer age but it was necessary. The transit operators were the blue collar segment of the data processing workers. Checks and cash were the primary methods of payments until recently. Check volumes grew every year until the mid 1990's. We still write billions of checks annually. An old banker once told me that without automation the bank would have had to use most of the 41 stories in their new building for people to manually run the business. The bank occupied fewer than 10 floors because there was a computer room on the third floor and another one on the fourth floor.

Check processing is a complicated undertaking. I've always been fascinated by the technology involved and the financial implications. Parts of it are obvious and logical but parts of it are art. Even one day's interest on a million dollars can be hundreds of dollars depending on interest rates. The prime interest rates in 1970 ranged from 7% to 8%. A bank that deals in hundreds of millions of dollars can make or lose a small fortune every day if they are good or bad at clearing checks. I was fortunate to be around and involved when banks stepped outside the century old box and began using the emerging technology to gain an advantage.

I will revisit check processing from different vantage points in other environments and circumstances in some of the following chapters. I will next be leaving the big city of Atlanta for the hinterlands but will not be abandoning FNB, technology or banking. Join me next time about 250 miles south of Atlanta after a short detour a little north. I hope you join me on my road trip.

wjh