Main 1401 Story
by Charles Branscomb
The purpose of this story is to document the origin and development of the 1401 accounting system. I was assigned the responsibility for directing the development of this system in mid -1957. What I have written here are my recollections of a great experience some 50 years ago.
Background
I feel some of my background related to accounting machines is pertinent to the 1401 story. I was a junior at N.C. State in 1945 studying Mechanical Engineering when I turned 18 in October and got the call from Uncle Sam. Since World War II had ended, I was not pleased with getting drafted, taking armored infantry basic training at Ft. Knox, and getting shipped off to Korea. What followed had a profound effect on my life and especially my career at IBM. I was assigned to the 107th MRU (machine records unit) which used IBM equipment and was responsible for all information handling on the 36,000 GIs in Korea. At the time, I had never heard of IBM. I learned to operate and program (via plugboards) punched card machines. During the last 5 months of my 12 months in Korea, I was what in today’s lingo would be called the data base manager. This user experience was invaluable in my career, especially the early years.
People
I hesitate to name a lot of people because I know I would leave out some important ones. There are three that standout, however, and they were crucial to the success of the 1401.
Jim Ingram was responsible for the processing unit development and release to manufacturing. Jim was from the south and a true southern gentleman in the best sense of those words. He was very knowledgeable, soft spoken, always calm but persistent to make sure the right things happened. His steady hand was important since, with everything new, there were unexpected problems all of the time.
Fran Underwood was the creative fire in the 1401 team. The architecture of the 1401 was Fran’s almost completely. He was interested (and knowledgeable) in many things and had a great sense of humor. You almost never saw Fran without a smile on his face and a glint in his eye, chuckling about something or maybe even nothing. His creative acronym, SPACE (Stored Program Accounting Calculating Equipment), for the 1401 struck a responsive chord since Sputnik had gone up in October, 1957. The 1401 development was known as the SPACE machine throughout IBM.
Shel Jacobs was responsible for market planning for the 1401. While the huge planning effort for the Worldwide Accounting Machine (WWAM) provided a lot of general information, it did not get specific enough to show just what was required for success. Shel did a lot of additional work to show the profile of punched card installations, the applications they were running and specifically what it would take to get them to move to a new system. He resisted his management’s push to demand more function, more function to the extent that at times they were not pleased with him. We (the development team) were very pleased, and he was a very important team member.
Marketplace
First we should talk about accounting machines in the 1950’s period. IBM had several such machines – I believe they were actually called tabulating machines – the 402, 405, 407. These machines used in conjunction with collators, reproducers, and keypunches permitted customers to process information stored in punched cards, and do the various accounting operations including billing, preparation of reports, etc. There was very little competition for the IBM installations in the United States, but Europe had begun to receive some competition from firms who were incorporating at least portions of more sophisticated electronics in their offerings. IBM had thousands of these machines installed and of course they represented a large revenue source.
WWAM
It was clear that with the coming electronics revolution there needed to be follow-on systems to move these customers into the new world. In recognition of the need to provide new systems for this function, and also the increasing competition in Europe, IBM had begun the WWAM program in the mid-fifties. There was a huge market planning effort between US and Europe to define the requirements for such a system, and this planning effort resulted in a huge quantity of information contained in three or four volumes. Development activity in response to these requirements was initiated – primarily in world trade. The responsibility for the processor was assigned to the IBM France laboratory, and responsibility for the printer was assigned to the IBM German laboratory. IBM world trade was delighted with these assignments since they had felt that the US laboratories had not been adequately responsive to unique European requirements.
My first assignment as Manager, Accounting Machines, in mid -1957 was to assess the WWAM program and conclude whether or not it was the proper program, and was on schedule. As mentioned earlier, there had been a huge market planning effort for the WWAM program, but I felt it did not provide adequate focus on the specific needs that must be satisfied to have a very successful program. In assessing the development programs in Germany and France, it was clear that there were difficulties. The printer was based upon a complex mechanical approach that used vertical rods called sticks, with a cylinder having engraved type around its surface attached to the top portion of the stick. The stick could be moved vertically, and the cylinder could be rotated, to position the proper engraved type to the print line. In order to get some 300 lines per minute, it was necessary to use 15 of these rods (sticks) across the face of the print line, each printing 8 columns. Of course all of the mechanical and electromechanical support required to drive these sticks was significant.
The engineers in the German lab were very competent, and were convinced that they could build a successful printer based upon this technical approach. My conclusion was that this technical approach was too complex, had limited potential for further development such as higher speed, and would likely require more service in the field than we could afford to have.
The IBM French laboratory also had very good, creative engineers, but the processor approach for the WWAM was based upon a plugboard approach rather than stored program as you have with a computer. For those not familiar with plug board machines, the programming of the punch card machines were all based upon use of a plugboard where the customer would wire between "hubs" in the plugboard to move information between different fields for arithmetic operations or to move information into the printing field, or other operations. The plugboard, of course, limited the scope of functions that could be performed, and also, once the plugboard was wired and inserted into the punch card machine, the only way to alter the program or make modifications was to stop the machine, remove the plugboard manually, change the wiring in the plugboard, and manually place it back into the punch card machine. Such a system could possibly meet the minimum requirements for the accounting machine area, but future potential would be extremely limited and would probably be obsolete shortly after its introduction.
In arriving at my conclusion from the assessment, I was sensitive to the fact that IBM World Trade was going to be extremely disappointed, and would probably fight a decision to stop the WWAM program since they, in looking at the market requirements put together, felt that a system to meet those requirements would be very good for their market. I talked to several people regarding this assessment, including my mentor for several years, Larry Wilson, who basically told me what he had told me many times before – "Do what you think is right." The recommendation to terminate the WWAM program was accepted, thanks to help from the lab director, Jim Troy, and the director of engineering, Ralph Palmer.
1401 System
Now, of course, we were under the gun to get a program going that would not only satisfy the current needs of the accounting machine marketplace, but have potential for growth into the future. With the market planning information from the WWAM program and much additional information regarding customer installation profiles, we began to establish the basic objectives for a new system that would provide an attractive move from the punch card plugboard world, into the stored program world. We defined the entry-level system – function/cost/price objective. (Such an entry-level system would satisfy all of the functions currently available in plugboard machines with much potential to go beyond that.) We essentially etched these objectives in stone, and made it clear to all involved that we would not deviate from these objectives. This was not easy since once we decided to go to a stored program system, it was easy to add function. We had a lot of criticism from various corners (particularly computer people) that our memory was too small, processing speed too slow, and our input/output was too limited. (We did not initially have magnetic tape or random access disk equipment.)
We were fortunate to have people coming onto the program with a punch card machine background, because we realized the transition of our customers from the accounting machines of 1950s to the stored program machine was going to be difficult. One absolute must was that the system be a variable field length machine since the plugboard machines had variable field length, and most early computers were fixed word machines. It was obvious that a fixed word machine would not fly in this market area. Since this kind of system did a lot of printing, it was very important to have efficient print editing – things such as floating dollar signs and commas in the amount fields, special symbols, etc. in the name and address fields, and special requirements in European languages. It was also absolutely necessary that we make very efficient use of memory. Memory was relatively expensive in the late 1950s, and in order to meet our cost objectives we were limited in the amount of memory we could afford. In order to achieve variable field length and efficient use of memory, we decided to use "word marks" to determine the beginning and end of a field. We incorporated an extra bit plane in memory such that each character position had an extra bit which would be set or not set, depending on whether that position was the end of a word or a field. This approach also permitted "chaining" of instructions without incorporating memory address information, and this contributed to making the system have very efficient use of memory.
Electronics
I should note at this point that the corporation had made a decision that all future products would be "solid state" machines, with no tubes permitted. We considered other memory technology, but the options were actually quite limited, and we settled on a ferrite core memory operating, as I recall, at 13 microseconds. We established 1400 characters as the memory size for our entry-level machine, based strictly upon what we could afford for cost of memory, and our analysis that almost all of the basic current accounting machine problems could be run in such memory size. IBM was developing three different circuit systems for use, depending upon the performance and cost of the particular system under development. These were TTL (transistor transistor logic), CTDL (complementary transistor diode logic), CTRL (complementary transistor resistor logic). After considerable analysis of cost, performance, heat dissipation, and particularly the projected cost improvements of the technology through time, we chose CTDL for the SPACE system. This turned out to be a good decision since the high-volume use of diodes in electronics drove the cost of diodes down faster than either resistors or transistors.
Electronics Packaging
IBM had under development two basic approaches for packaging electronic technology. One was for small-to-medium size systems (called CUBE). And the other was for large systems (called ROLYGON). The SPACE machine fit the small/medium size system, and thus we initially planned on using the CUBE package. However, as we made specific requests on the CUBE packaging group for modifications to help us meet our cost objectives, and received essentially no commitments, we told them we were leaving them, and would go to the ROLYGON package. They simply laughed at us, and said, "You will be back." They were right. And within a couple of days we were back, making our case to get some changes made. And of course we ultimately wound up with the CUBE package.
Printing
Printing is a crucial function of an accounting system. We were extremely fortunate that the Endicott laboratory had been working on a new printer concept in their advanced development area. This printer approach minimized the amount of mechanical hardware in the printer itself, and relied heavily on some good electronics to drive hammers against moving type. There was a fundamental difference between their printer approach and prior printers in that the moving type was attached to a band that moved horizontally in front of the print line, and this meant that the vertical alignment of the print characters was excellent. It also meant, however, that consistent firing of the hammers – print magnet hammers – against the engraved type was crucial, and that the hammer action must get in against the type and out very rapidly to prevent smearing.
This printer was initially intended to be used with large systems in an offline mode such as a tape-to-print operation. Use of this printer – which became known as the "chain" printer – with the SPACE machine gave the printer program much larger quantities than initially visualized. Even though this printer technology approach minimized mechanical hardware, the hardware that was needed was very challenging to design. The band used to attach the engraved type and move it horizontally across the print line at a speed such that the printer could achieve 600 lines per minute meant fatigue problems with the band itself. As a matter of fact, late in the development cycle of both this printer and the SPACE machine, the printer program was having great difficulty, breaking bands (chains) and had several task forces to try to understand and correct the problem.
The program was being managed by a wonderful senior engineer, Jonie Dayger, who had been in place a long time, and who did not have great analytical knowledge, but had great experience and a great feel for what is right and wrong. Nevertheless, as bands continued to break, the division started a task force with some highly-educated mechanical and materials people, to determine the fix for this fundamental problem. After a few weeks of analysis, this group came forth with a proposal to the senior engineer on how to correct the problem. The senior engineer sort of chuckled with them, and said, "We tried that months ago." They looked at what he had tried, and decided there was a slight variation. So, the new approach was built and tested, and lo and behold, broke print bands even faster than the current technical approach. While this was going on, the senior engineer continued his own exploratory work on fixing the problem, and fortunately found small changes to his original approach that corrected the problem and gave long life to the chain band. I do not know all the specifics, but from my perspective the division vice president of engineering was about to replace the long-successful senior engineer because of the chain breakage problem. Fortunately, this was not necessary.
More should be said about the paper handling "carriage" used with the 1403 printer. This hydraulic driven carriage could accelerate and decelerate paper forms very rapidly—up to 80 inches per second and back to zero without damaging the paper sprocket holes. As you can imagine this required sophisticated research, engineering and testing, and the result was so advanced that the knowledge gained was used in other printer products well into the future. This played a big part of the "throughput", since most printing involves printing a few lines on an invoice, a statement, a check, etc., and then quickly skipping to the next document. The 1403, with the higher speed version later, was the workhorse printer for the industry for many, many years.
Printer Electronics
I noted the requirement for high-speed solenoid operation to drive the print magnet hammers. This activity required a circuit spike of some 7 amperes. I think I’ve already mentioned the corporate decision to go solid state, and it seemed clear initially that we were not going to be able to find print magnet drivers to handle the peak current load required. Thus, we and the printer program appealed to the director of engineering in our division for an exception to the solid state decision, just for the print magnet drivers. We were refused, and hence the significant effort pursued to find a power driver that was consistent enough to provide the precision required, and also robust enough to handle the peak 7 amp currents. Fortunately, we were able to find such a device, or rather participate with the vendor in getting it developed. The power dissipation in this device led to some excitement in the SPACE program engineering model.
All of our print magnet drivers were on one "gate" of the CUBE package, and one day during lunch hour when no one happened to be in the model maker room, and our prototype model of the SPACE machine was being worked on, smoke was seen to be coming over the partition wall by the people next to the model room. Sure enough, the print magnet gate was essentially on fire, although very little blazing, but I guess we came quite close to having our only engineering model of the SPACE machine go up in flames, which of course would have been a profound blow to the program at that time.
Card Reading/Punching/Collating
We were also very fortunate that there had been card reader technology development in the Endicott lab, mostly under the direction of Larry Wilson, for some time. We used that technology to build the 1402 card reader punch which had opposing card feeds on each end of the machine, and stackers in the center to provide the functions needed for reading cards, collating cards, and punching cards.
Tape Systems
The key to the 1401 success was to establish the cost effective base to move the thousands of punched card users to a new world. Tape attachment for announcement was not needed to achieve that. But I assure you that we did not "stumble" on tape for the 1401.
We knew tape and other I/O would be attached but I refused to let the expanded 1401 deflect us from our fundamental focus. That sounds like we couldn’t do more than one thing at a time but adding things early could easily have put us on a slippery slope. As Fran’s notes have established, we did want to do enough technical work to make sure that we were not doing things that would make future I/O attachments particularly difficult. You may recall that the MAC program was going to be "everything for everybody" and wound up being "nothing for anybody".
Once we decided to use the chain printer, I had discussions with my good friend, Jonie Dayger, about using the 1401, assuming we met our objectives, as the "control unit" for his off line tape to print application. At that time, Jonie said there was no way he was going to stop his TAU based upon our lofty goals. I agreed with him. But I told him that tape would get attached to the 1401 and take much of the market from his TAU. We agreed to wait until costs and function were much clearer before having serious discussions about using the 1401.
I never looked to see how many of the 1401’s had tape attached but my guess would be less than 30%. Since this is just a guesstimate it may not even be close. Of course that percentage would be very high for the 1460. Also, many of the 1401’s with tape would have been used for tape to print and card to tape as opposed to a true tape processing system.
Once we were quite sure of meeting our cost objectives, I was more receptive to seeing tape with the initial announcement (provided that didn’t cause a meaning delay) since the company needed the chain printer for off line printing in large system installations and I did not want us to announce the TAU and then a few months later essentially wipe it out with tape on the 1401. The TAU announcement would have misled IBM customers. Also, Bob Evans clearly pushed for tape announcement for his 7070 as well as other large systems (just before October, 1959, I starting reporting to Bob in Endicott instead of Don Spaulding in White Plains). Somewhere I read about an interview with me where I said we attached tape to the engineering model before entering Product Test. That is not my recollection now but again I may be wrong. I’m sure Product Test put us through the wringer before supporting tape but I do not recall just how we achieved that. My guess is that we relied heavily on the testing experience of the TAU/chain printer and critical review by Product Test of our design for attachment to the 1401.
Reliability
There was a heavy reliability thrust throughout IBM in the second half of the fifties, and there was a "reliability group" in the Endicott lab that analyzed our designs and continually made suggestions. Many times their suggestions would have added cost to the machine, and we heavily resisted those items unless they could redesign their suggestion or show us other areas of the machine where we could get offsetting cost savings. This turned out to be a very helpful relationship because we wanted better reliability, of course, and they wanted to help us incorporate it into the machine. We were able to do most everything they recommended without actually increasing the cost of the machine.
Sterling Arithmetic Application
Throughout the development cycle, we were continually working with market planning to ensure that the SPACE machine met all the customer requirements, and would provide a relatively easy transition from the old punch card machine world. We of course also worked closely with world trade Europe since they had been one of the originators of a new accounting machine focus. From that activity, we were challenged at one point by IBM UK that the system would not satisfy one of their most fundamental applications, sterling arithmetic. There was a particular program used by large numbers of customers that their analysis and work showed would not fit within the maximum memory size of the 1401, which was at that time 4000 characters of memory. We asked them to bring all of the information on the program to Endicott, and we would pursue programming the application in the 1401, and determine whether or not in fact it would fit. Fortunately, when we finished that activity we found that the application would fit within 3300 characters of memory, leaving us some 700 characters to spare. Both we and they were delighted.
San Jose Challenge
During development we got another one of the many challenges to the program. The 305 RAMAC program had been developed in the San Jose laboratory, with primary focus on random access processing using disk files. They proposed to our director of engineering that a 305 follow-on program would be a better vehicle for the marketplace than the SPACE machine. We were ordered by the director of engineering to evaluate together with the San Jose people their proposal versus the SPACE machine, and make a joint report as soon as possible.
The charge to direct this activity actually went to the laboratory directors in Endicott (Jim Troy) and San Jose (Lou Stevens) labs. Jim Ingram, Fran Underwood and I went to San Jose for an intensive four- or five-day review, with a plan to present conclusions to the laboratory directors at the end of that period. As you might expect, the people from our program and their program could not reach an agreement. We felt strongly that the word mark approach was much more efficient, and provided lower cost of the electronics as well as excellent print editing; and that we were focused entirely upon moving the large volume of accounting machines in the field into stored program processing. They of course had strong convictions that the future was random access processing, and that the 305 base would be a better vehicle for the future.
Since Jim Troy had a cold, he decided to take the train across country to meet with the San Jose lab director for our final review. When Jim arrived in California, we were all going out to dinner at a place picked by Lou Stevens that was many, many miles from the train station. This was not too popular with Jim who had been on the train for five days, but we went. Unfortunately, when we got there, the restaurant was closed that day. We then went back to downtown San Jose, to the Chuckwagon Steakhouse, and it had been one of those times when it had rained heavily in San Jose. When Lou Stevens drove us up to the restaurant to get out, he unfortunately parked where Jim stepped out of the car directly into a deep puddle of water. Jim was very senior to Lou, and those of us from Endicott felt the battle was now over before we had ever made any presentation. We of course did make presentations but thanks to not only having the right system for the marketplace we were going after, but also having a strong lab director, we were able to eliminate that challenge.
Forecast/Market Reaction
We continued work with market planning, and the forecasting group was showing a large volume forecast for the SPACE system. The first forecast for the US was 2200 systems and a guesstimate of 4700 systems worldwide, in a time frame where most systems were in a few hundred range. Both the SPACE machine and the printer program were really put through the wringer by "product tests," which was a separate line organization directly to the president of our division. We would not be able to announce the 1401 until they put their stamp of approval on it.
We finally received approval and the 1401 was announced in October of 1959. Now the real challenge began – namely, moving from an engineering prototype to production models, and to full production. We had large numbers of orders after announcement, and it was clear that we had a very successful system unless we stubbed our toe somewhere along the line, including putting it into production.
Production Start-up
Within a month or so after announcement, I was offered an opportunity to go to the division president’s office as an administrative assistant. I turned down the opportunity because I knew it was going to be a real challenge to put the 1401 into production, and I was going to see that through before I moved anywhere. The transition into production turned out to be at least as difficult as we suspected. We were dealing with new technology, new packaging, new power supplies, new architecture, and both the development lab and manufacturing organization did not have a lot of experience in this kind of technology. Several people on the team in the development lab would actually work the first shift in the laboratory, and then go over to manufacturing to help them for a significant part of the second shift. As a matter of fact, for a few people that second shift sometimes went on into the third shift as well. I can remember one instance where I was over there at probably 3:00 to 3:30 in the morning with one of our guys (Bill Murray), and actually ordered him to go home to get some sleep, and don’t get to the laboratory before 9:00 or 10:00. I went home and did not get to the lab until about 9:30 the next morning, and of course I saw him there when I first came in, and asked him what time he came in. He said, "Oh, about 7:00 or a little before."
Overall, the manufacturing organization did an outstanding job. They were "asked" to double production for the first year while we were still in the release to production stage. Many of the manufacturing people were skeptical of these "engineers" from the lab over in their organization, presuming to help. I won their respect through some work on the 1402 card reader punch, which was manufactured in Rochester, Minnesota. At about 11:30 at night they were trying to test out the connection of the 1402 to the mainframe, and could not get satisfactory operation.
After working with them to diagnose the problem, we decided that the 1402 had been timed incorrectly. They said, "We will have to get someone in from Rochester or get another 1402," since they could not fix the problem. I told them to get an oversized taper pen, get someone to work with me to remove the current taper pen, set the timing to the proper setting, and clamp it in position. Then we would drill slightly oversized – for an oversized paper pen, and pin the shaft to the correct timing position. They initially would not let me do that. I don’t know if it was safety reasons, or they just didn’t think I could do it, but after getting their management involved, we proceeded along the path outlined. Shortly, the 1402 was performing perfectly and the test out of that system proceeded normally. I found they were more receptive to listening when I had comments in the future.
Customer Demonstrations
While we were in the "release to manufacturing" stage, the marketing organization was moving ahead full steam with the marketing of the 1401. They of course wanted demonstrations, et cetera. We were very limited in what we could provide since we only had one engineering model when we announced. We were building additional engineering models to give the engineers more access to evaluate engineering changes, to correct difficulties, as well as make some significant changes for cost reduction, since technology was continuing to move ahead.
IBM Europe was in general delighted with the 1401, and they were demanding demonstrations for their customers. They got the company’s agreement to demonstrate the system at the Hanover Fair in the spring of 1960 – the Hanover Fair was already a huge industrial fair, and was a key place to initially demonstrate the 1401 in Europe. It turns out that we used the initial engineering model for this demonstration, and IBM Europe sent over excellent people to train on the system and to service the system while it was in Europe, to reduce the demands on our engineering team since we were quite busy. The demonstration in Hanover went very well, and was well received by customers. IBM Europe had also gotten agreement to demonstrate in a couple more of their locations before it was returned to the US.
You might imagine my shock when I got a call from the customer engineering manager of IBM Netherlands, telling me that they needed help to solve a problem on the 1401. I asked him when the next demonstration was scheduled, and he proceeded to tell me that the 1401 was installed in a customer’s office. We had no idea that any such installation was in the works, but of course we had to try to respond. We got one of our best people headed for the Netherlands on the next airplane. Probably while he was less than an hour out of New York City, I got another call from the Netherlands customer engineering manager who said they had found problem, fixed it, and the system was performing as planned.
Cabling
I don’t think I have made clear the importance we placed upon meeting the initial goal for cost for the 1401 system, and how tenacious we were in doing everything possible to meet those costs. A decision we made, which in retrospect was clearly wrong, was to bring cables from the 1401 processor out of the system without bringing them through a connector. They were simply long cables like tails that came out of the system for attachment of the 1402 card reader punch and the 1403 printer. We did this for both cost reasons and reliability reasons since connectors were one of the most unreliable elements of this low-level signal technology in the early days. Our approach worked well functionally, and probably gave us improved reliability. These dangling cables were an eyesore, and also limited customer flexibility on placing the 1402 and 1403. Adding connectors was a part of the big engineering change that we introduced to manufacturing during fairly high-volume production, somewhere around the second or third quarter of production. This was not an easy task for anyone.
Initial Shipment
When we announced in October, 1959 we committed for shipment to occur in September of 1960. This time frame between announcement and shipment was not unusual in those days. We didn’t, however, really anticipate that the initial shipment would occur at 11:30 p.m. on September 30th, 1960. Of course I was on the manufacturing floor as the system was buttoned up, and moved to shipment. This system went to Time-Life in Chicago, as I recall, for the printing of labels for magazines.
Costs and Profitably
I have mentioned (probably a dozen times) that we set solid objectives for costs early in the program, and we were very happy when the final announcement level cost information came in. We were only $50 over the $25,000 target for the entire system: 1401, 1402, and1403. My boss in headquarters had bet me a felt hat that we would not make our objective. When I next saw him I was surprised that he insisted on me buying him a felt hat, even though we only missed our objective by $50. He marched me to downtown White Plains to one of the more expensive stores, and I bought him a felt hat. (All executives in IBM in those days wore hats.)
The follow-up to this cost story is that the new cost estimate after we had entered into production actually showed that the system overall program costs were $100 million lower than the initial estimate. Much of this, of course, was due to the cost reductions going on in electronics in those days. I had known that we had excellent profitability in the 1401, based upon the announcement information, and you can imagine I was shocked when the word came through that the profitability of the 1401 was reduced following the production level cost estimate, even with the big cost reductions. I insisted on finding out what went on, but the financial organization in the division refused to share any information with me. Fortunately, I was now an administrative assistant in the division president’s office – told him the situation, and he agreed that when finance came in to present the profitability of the 1401, I could participate in the meeting. When the meeting occurred, finance attempted to get me kicked out of the room, but the division president told them to go ahead, and thus I found the basis for the presumed reduction in profitability.
In those days most systems were rented as opposed to outright purchase, and a key element in the profitability was the forecast life for the systems in the field. Since the profitability would have looked too good (if there is such a thing), they had arbitrarily reduced the estimated life of each system in the field by over three months. Naturally, this showed a reduction in profitability of the system.
Diversion from 1401 Focus
During the time of development of the 1401, I also had another department working for me, responsible for the development of a 2000 card per minute sorter. We had designed the machine, been through announcement level testing, announced the machine, and were in what was called production level testing. We found some punch card stock that would cause jams in the 2000 card per minute machine, and you can imagine what a jam would look like, with cards being sorted at that rate.
Customer commitments on ship dates were sacred in IBM (and I knew that), but I was not going to put a product in the field that would very likely get a black eye. So we proceeded to delay initial customer shipment by, as I recall three or four months, to give us time to fix the problem. And, it just so happens that in that period of time IBM had a big push on reliability and customer satisfaction.
After word was sent to headquarters, I got a call saying Vin Learson would be in Endicott the next day to visit me. He was, as I recall, Executive VP and had a reputation as a tough guy, but I had not had any real problems with him. I met him at the airport the next morning, and we had small talk on the way to the laboratory. We then proceeded to a conference room where the vice president of engineering for our division was there, the laboratory director was there, the direct manager of the product working for me was there, and I was in the meeting with Mr. Learson. All of a sudden, small talk was gone and things got very serious. Mr. Learson looked me directly in the eye and said, "I am here today for you to tell me how we’re going to ship machines in less than any three months late, period." I was young and naïve, and after a few seconds I said, "The only way we’re going to ship machines in less than 90 days late is for you to agree to ship at a lower reliability level than we think we should, period." Things were very quiet for what seemed like a long time to me – the vice president of engineering said nothing, the laboratory director said nothing, and things were a little tense. Then Mr. Learson said, "All right. Let’s work out the program." I think we finally shipped about 45 days late, and the 084 turned out to be a very successful 2000 card per minute sorter.
In a few days the laboratory director informed me that I was fortunate to have the 1401 program going so well. Otherwise, I would have been at least removed from management, and possibly fired. I obviously did not show the proper respect for the company executive, and would have handled it differently with a little more experience. As it turned out, Learson liked people to be honest and direct with him, and I had a great relationship with Vin throughout the rest of my career in IBM.
Recognition
Sometime after announcement five of us were invited to a luncheon with the division president (O. M. Scott) – Fred Demer, who was the inventor of the chain printer concept; Jonie Dayger, who managed the development of the 1403 printer; Shel Jacobs, who was the key market planning manager, Fran Underwood who was responsible for the architecture; and myself. This was a very nice recognition lunch which was all we expected, but at the end of the lunch the division president said, "I would like for you guys to take your wives and review the laboratories in Europe. By the way, do not spend much time in the laboratory." A trip to Europe today is somewhat incidental, but in 1960, for a young guy who had not been around very much, it was great. I found out what great hosts the Europeans were, and I found how appreciative they were for the 1401.
I don’t think I have mentioned that I was in Paris, as part of my WWAM review, in October of 1957 when Sputnik went up. I was meeting with the product planning manager of IBM France (Jacque Maisonrouge who later became Executive VP of IBM), who loved to jab the Americans. He said that Sputnik up there goes round and around the world, going beep-beep-beep-beep, and when it goes over the USA it says, "Ha, ha, ha, ha."
1401 Related Follow-on Activity
The 1401 program was, by any measure, extremely successful. There were extensions for lower entry, more memory (1440, 1460), tape attachment, disk file attachment, multiplexer front-end, and many, many other input/output devices. The origin for the multiplexer front-end is quite interesting. While I was Administrative Assistant to the Division President, John Haanstra was Division Vice-President of Engineering. He asked me to join him on a visit to the Kingston, New York laboratory—a drive of some 2 hours. All the way up and back, he picked my brain about how the 1401 handled input/output operation. Our trip was on a Friday, and the following Monday his secretary called me to ask what I had done to her boss. She said he showed up Monday morning with 14 IBM Executary tapes to be transcribed. Over the weekend, John had architected and designed a multiplexer front-end for the 1401. As I recall, a product was pursued basically as he had designed it.
One interesting follow-on to the 1401 came about in an unusual way. A group in the Endicott lab was developing, as I recall, a follow-on to the IBM 650 (a fixed word machine). Their system was a fixed word machine and would provide very good scientific computing. The "product line manager" in Headquarters (Don Spaulding) questioned whether this shouldn’t be a variable field machine compatible upward from the 1401. This, of course, would make it much more of a business-oriented machine and dramatically reduce its scientific computation ability. The 650 follow-on people were very upset with the possible change in their program, so Don took a group of their people and a group of the 1401 people to Rocky Point Lodge in the Adirondacks to "study this area and conclude what was right." (It was clear to me that Don had already made up his mind for a 1401 follow-on). Sure enough, after two or three days of intensive study with reports from several groups, Don concluded that the program would be changed to be upward compatible from the 1401 and, of course, have variable field length. One brilliant young mathematician (Bill Hanf) working on that program who really wanted a fast scientific computation machine was very upset. He said, "This is the beginning of the end of IBM—I have one share of IBM stock, and I am going to sell it when I get back to Endicott." The system became the 1410, upward compatible from the 1401, and, as far as I know, was quite successful. I personally believe that this early decision by Don provided the spark for the 360 system compatibility throughout the product line. I am sure others have different views of what provided such a spark.
With the success of the 1401, it is not surprising that several competitors were taking a shot at it. As I recall, the key ones in Europe were ICL and Bull-Gamma. In the U.S., I think UNIVAC and RCA were the expected ones, while the most surprising was Honeywell with its "liberator" program to liberate customers from the 1401. They claimed that this system would run 1401 programs essentially as is with very little manual intervention required. Our Endicott lab engineers working on the low end of System 360 did not believe it. Jim Desell, who was Systems Engineering Manager for the Eastern Region of IBM, managed to get to a demonstration of the machine by Honeywell, and was impressed that it really did work as advertised. Jim went to Endicott to present the story and generated much more serious interest in emulators of the 1401. Such emulation of the 1401 had been suggested before (by me and I’m sure others), but after some initial work, the Endicott people said the emulators were too slow and would still require too much manual intervention.
Fortunately, the Honeywell 200 provided additional emphasis for some serious work on emulators and, using read-only store to do hardware emulation, the Endicott people found that they could in fact almost completely emulate the 1401 programs. Thus, the Model 30 of the 360 development program incorporated emulators in their plan.
The team in Endicott continuing to enhance 1401 did not believe the System 360/Model 30 even with emulators would provide a satisfactory answer to the Honeywell 200. They, instead, proposed a 1401 Super that became known as 1401S. The Vice-President of Engineering for the division, John Haanstra, reviewed the 1401S program. He agreed with the 1401 team and put his weight behind the 1401S.
This caused a big battle between John Haanstra and the management of the 360 development, which went all the way to the top of the business and, of course, management at the top supported the 360 line. John did not accept this conclusion and went to the mat with the Chairman of IBM over the 1401S versus the 360 Model 30. This, in essence, cost John his job, and I don’t recall the timing, but it was not long before he was to leave IBM and go to General Electric to become head of their computer effort. After John had been with GE a short period of time, I got a chance to see him and get his reaction to what he found at GE. He told me that they had some very good people and they could, in fact, have a real impact on the computing industry if they would pull together as a team and put forth the resources that would be required. He was not too optimistic that such effort would be put forth, since he was being challenged heavily by other areas of GE who needed resources and were producing a profit. At the point of time I talked to him, he had not even been able to get the semi-conductor division of GE to support him since they were measured on revenue growth in the commercial semi-conductor industry. Unfortunately, we never got a chance to find out because John, who flew his own airplane was flying across the country toward Phoenix, Arizona (headquarters of GE computing activity) and the plane crashed somewhere in the southwest, I believe in New Mexico. Sadly, John and some other members of his family were killed in the crash and, to my knowledge, they never determined the cause of the crash.
My recollection is that only the Model 65 of the 360 program had serious work underway on emulation prior to the announcement of the Honeywell 200 Liberator. Emulation using read-only store became a major thrust on several models of the 360 under development, and I believe frankly that the Honeywell 200 announcement turned out to be a significant plus for the 360 line because without good fast emulators, 360 would have had a very difficult time. The problems with the 360 operating systems only emphasized the importance of very good emulators of the then current product line.
It turns out it would probably have been wise to do both—some version of the 1401S, as well as, all of the models of the 360 line. The reason I note this is that while the 360 was very successful, it was mostly in equipment rental range above that of the 1401 program. As a matter of fact, in April 1964, when the 360 was announced, IBM had 87 percent of the business for systems in the rental range of $2,500 per month to $10,000 per month. Five years later, in 1969, IBM only had 47 percent of the business for equipment in the rental range of $2,500 to $10,000 a month. This led Frank Cary to start the General Systems Division with the focus on the low end of the line, and it became the only computer division in IBM to be integrated with its own development, manufacturing, marketing, and service. By the way, GSD was very successful in the low end of the business line with the System 32, System 34, System 36, and System 38—which later with enhancements became the AS/400. The division was less successful with its Series 1 entry into the mini-computer business, but did develop the IBM PC shortly before the division was integrated back into IBM.
Two anecdotes about John Haanstra Added July 29, 2014
Robert
I assume you recall my story about John from the "1401 story". In early 1961 when I had moved to division HQS, John was division VP of Development. He asked me to go with him on a Friday visit to the Kingston, NY Lab (about 2 hour drive) and all the way up and back he probed me on how the 1401 handled Input/Output. Monday AM his secretary called me what I did to her boss on Friday. After the weekend John had showed up with some 14 to 18 Executary "belts" (The IBM OPD recorder used belts instead of tapes giving random access to recordings) to be transcribed. They represented the design of a multiplexor front end for the 1401.
John was not only a powerhouse intellectually but he was powerful physically as well. I still vividly remember being in a meeting with John and the VP of Manufacturing for the division (Bud Thue) where John was demanding that 1401 production for the first year be doubled. He had a habit of cupping his hand and pounding on the conference table - I believe the room actually shook when he did that. John won and production was doubled.
You have got some good insights into the Development area activity after Honeywell announced the H200. It is unfortunate That Jim Desell is no longer living. Jim was Systems Engineering Manager in the Eastern Region when H200 was announced and he could have personally provided the Marketing division activity. When he found out that Endicott engineers were skeptical that Liberator would work, Jim found a way to get into a Honeywell customer demonstration and saw that he did in fact work. He then went to Endicott to make sure there was no misunderstanding and demand that development respond. I was on Vin Learson's staff while this was going on and remember interacting with Jim about H200 and Liberator but I do not remember the specifics.
About a dozen years later Jim and I became very good friends in the General Systems Division where he was head of Marketing and I was head of Development and Manufacturing. As I think you know GSD re-established IBM leadership in the middle to low end of the product line. I read recently that IBM world wide sold over 250,000 System 32/34/36/38s.
I did not mean for this to get so lengthy - I hope some of it is of interest to you.
Chuck