The 7203 is most likely among the last of Burroughs' calculators. Its PCBs have design dates from 1972-73 and the ICs have manufacture dates from 1975.
It is a print-only calculator with programming facilities including loop, branch and subroutines. It has 16 data memory locations, 204 (optionally 408 or 816) program steps and a magnetic card unit that could read and write programs and data. It does not have scientific notation for numbers, nor any further inbuilt mathematical functions beyond square root.
The calculator was built in a Burroughs factory in France. The PCBs have English designations and the ICs are American:
The feature set of this machine is a bit peculiar and one wonders what the target market was. It has limited mathematical ability with no inbuilt higher functions so its appeal to scientific users would be limited, far less than a Wang machine of the early 1970s. The program store was generous and the manual states that function libraries (eg trigonometric functions) were available on magnetic cards, to be loaded when needed. The data storage capacity was only 16 locations and this would have placed a severe limit on overall problem size, particularly if some locations were used by function libraries. The card unit could read and write all 16 locations (or a subset of 4) so perhaps the machine was most suited to performing basic processing on small data sets. The manual suggests that a payroll could be stored and maintained on cards, for example.
It is interesting to compare this American/French machine with the Soviet Soemtron, produced and sold at almost the same time. The gulf in technology available to ordinary users in these countries is enormous!
There is a very large programming space available but to date no original programs have been discovered to give examples of the type of programming that was anticipated. As noted above, there are severe limits on the storage of data and intermediate variables but clever programming or use of magentic cards may help to overcome this. A further limitation is that input/output is limited to simply waiting for a number or printing the contents of a storage location. There is no way to prompt for a specific input or label the output with anything other than its register number. This rapidly becomes a fruitful source of errors in more complex programs as very close attention has to be paid to the progress of the program so as to supply the correct input at the correct stage, and care has to be taken to transcribe the correct printout at the correct stage in processing. It is easy to imagine how the primitive interfaces of early desktop computers must have appeared to be a major improvement on this situation.
It would be interesting to obtain some original programs to see the types of problems that were tackled on this machine and the manner in which the storage and user interface limitations were handled.