Up until this point, the Nascom system (apart from the CP/M modification to the Nascom 2) had used a "fixed" memory map i.e. different areas with dedicated RAM or EPROM functions. The Gemini system, because RAM was now much cheaper, filled all the addressable area with RAM. EPROMs then "punched holes" in the memory map at any convenient address as required. Because it was designed like this from the beginning, it was far easier to implement CP/M (which requires RAM from address 0000h) than on the Nascom, with its fixed monitor and BASIC areas.
Those of you who know about such things will remember that there were already other computer busses in use. The problem with most of them was the cost. Being "professional" systems they tended to use techniques like active termination, special connectors etc. The Nasbus (later slightly modified and renamed the 80-Bus) allowed simple single-sided 0.1 inch edge connectors to be used, mounted on a passive backplane which could easily be made from Veroboard.
The amount of hardware which became available was quite large when you consider the relatively short lifetime of the system. This was encouraged by the release of a "patch board" which allowed home users to design quite complex cards for their own systems.
nascom.info