I'm behind, but I finally managed to do some reading.
The early IBM tabulators used 'accumulators', which a collection of decimal wheels. Typical widths were 8 or 9 digitals. The wheels had relays and contacts for reading. The position of the wheels in the accumulator determined the value. The displays would look exactly like a digit based counter display.
In the late 30s there were two types of computing going on. There were 'difference engines', which can best be described as analog devices. Voltages were used to represent numbers, angles, etc. Setup time was long and difficult, and the precision of the machine was limited by the physical attributes of the machine. These machines were the electronic equivalents of slide rules, although with the setup capability they could do more complex things.
There was also 'digital computing' - scientists using accounting equipment as fast calculators. The accounting equipment of the day used punch cards and was decimal based.
IBM did a few custom machines for the government and universities. One of the more advanced ones built on the electromechanical accumulators was the ASCC. It had 72 accumulators, each holding 24 digits. (The highest order digit actually held the sign of the number.) There were 60 'registers' which had no math capability - they would be used for storing contants.
The machine had multiplication and division capability. It took some creative algorithms to implement true multiplication and division, not just quick adding and subtracting. And since this was all counter and relay based, you can imagine the racket it would have made when it was working.
In the early 40s people knew that vacuum tubes could be combined into flip-flops to get digital storage. Three 'flip flops' would get you to 8, and four would get you to 16. But nobody was interested in 8 or 16 at the time - they were still thinking in decimal, and to be a drop-in replacement for the counter wheels they needed base 10.
One solution to the base 10 problem was to use 10 flip flops connected in a ring, each one representing a digit. The hardware would 'pulse' the unit the correct number of times to select the correct digit. This system was used by ENIAC. Tiny neon lamps were used to show an operator or engineer which digit was selected in the unit.
It was a wiring error in one of these units that lead to the realization that 4 flip flops (bits) could be used to represent a decimal number in a more compact manner. The patent for the 'wiring error' was granted in 1952.
The change to vacuum tubes from counter wheels made things much faster, and also caused people to redo the logic that implemented multiplication, division, etc. That is generally when the world started to go binary too .. going binary reduced part counts.
Anyway, for the purposes of your project, the 10 flip flops in a ring where each one represents a digit seems to be the most interesting. You have to design a circuit such that you can set it to 0 or clock it a few times to select the right digit. That most closely resembles the ancient tabulator wheels from the punched card equipment of the 20s and 30s.