Dave, when I got the assignment to study the "small scientific market for IBM" in 1958 I assembled a small team. Initially there was me, Robert C Jackson, William H Rhodes. Quickly others were added. Anne Deckman (died tragically before project was completed), Kelly B. Day, William Florac and James Brenza. I have contacted Bob Jackson. I think he has some artifacts or documentation. Kelly Day is deceased. I do not know the whereabouts of the others.

The group was small and we all pitched in on most aspects of the program. Competing machines were the LGP 30 by Librascope and the Bendix G-15. We quickly concluded that IBM could offer nothing really new with drum machine so I looked elsewhere. I thought that IBM should be able to take advantage of their bigness and use technologies developed for larger machines in the smaller machine arena. I do not recall being given a set of hard objectives.

Human memories are notoriously unreliable, particularly 40 years later. The laboratory was organized such that a memory group designed memories for various projects. Likewise there was a circuits group which designed standard circuits and cards. The basic packaging design (SMS) was designed in the IBM Endicott Lab. Machine groups were expected to draw on these speciality groups if they could do so and meet their objectives. Projects were assigned by the Laboratory Director (in this case H. Tyler Marcy) according to resources and demands. Of course many of the best projects were the result of engineers working a related area and kind of grew.

The basic objective was to provide a useful machine at the least possible cost. That meant restricting the instruction set (someone can always suggest a nice additionally instruction for just a little more circuirtry.) It meant being clever in design. There were no arithmetic curcuits in the original design and releasee. That was all added by the San Jose group. It meant finding the least expensive Input/Output. You can imagine the flak I received for putting out an IBM machine which did not have a CARD reader and punch.

I cannot tell you the origin of the idea to use table look up. I do recall some general suggestion that we ought to begin thinking about using the logical functions in a machine to replace hardware. We completed the engineering model sometime in the spring followed by a debugging session. Then a model from release drawings. This was submitted to the Product Test Lab (a separate organization) for testing against specs.

At the completion of that test the machine was announced. It was a fully operational machine. It was not a "partially working machine" as you have apparently been lead to believe.

A patent was filed and issued for the Table Lookup Multiplying Computer. We later receied an IBM Patent Award for the content of that patent. Later the San Jose group made improvements and used the machine as a basis for process control functions.

In the fall of 1959 there was a major re organization within th IBM Co. Small machines were asigned to the General Product Division and large scale machines were wssigned to the Data Processing Division. Clearly the 1620 did not belong in the Poughkeepsie area of responsibility. The decision was made to release it to San Jose Manufacturing. A Product Engineering group was formed to support the project in the San Jose Lab. As I recall John Kyffin headed that group and was responsible for the San Jose end of the transfer.I personally do not know the people you have listed in the San Jose group. Perhaps they came along later? I really didn't follow the project closely after release as I was busy with other things. I do know that everyone on the project looks back with fond memories of that project and what we accomplished.

More next time,