I like the idea of using electro-mechanical relays, instead of solid-state switches, for this application. There will only be four relay coils energized at any particular time for a measurement of soil conductivity (or its reciprocal, resistance), with two relay contacts closed on each of two earth electrodes. Relay coil power consumption will be minimal, especially if magnetically-latching bi-stable reed relays are used. Using inexpensive Microchip PIC processors to implement the serial data links and relay selection logic will add almost negligible additional power consumption. I estimate (this is a pure rectal extraction) that 95% of the design effort will be devoted to PIC programming and serial data-link design. However, once the programming is solid (complete and error-free) for just one earth electrode, it will be solid and can be duplicated for as many additional electrodes as required.
The only thing that changes among the earth electrodes is the address of each individual earth electrode. This address could be specified at start-up and downloaded (via the data link) to non-volatile memory on each PIC, or each individual earth electrode address could be "hard coded" into the application software of each PIC, instantiated when each PIC is programmed. There are many variations on this theme, but the LAST thing the OP needs is a circuit board with the added expense of a bunch of slide-switches (or jumper wires) to specify the earth electrode address. Hopefully, this sort of thing disappeared in the previous century, where AFAIK it was most recently used to "program" residential garage door openers.
The earth electrodes are presumably separated by some distance, perhaps ranging from less than a meter to several meters. A shielded multi-conductor (six to eight conductors) cable will be mandatory for connecting each earth electrode to a motherboard containing the analog signal wiring. To be determined is what gauge wire will be adequate to excite a pair of earth electrodes. That will depend on how much current is expected to be drawn by the electrode pair, the amount of voltage drop tolerable in the connecting cables, and the spacing between pairs of earth electrodes. Commercial cable exists that contains a pair of heavier gauge wire for power and lighter gauge wire for low-level, low-current signals. And of course it is always possible to use two cables, with appropriate gauge wire in each cable.
While FETs could certainly be used in place of relay contacts, I think that less design effort is required to implement relay switching, with a PIC microprocessor performing the logic and setup function at each earth electrode. Given how infrequently the relays need to be switched (to configure for a new set of measurements), small
magnetically latching reed relays rated for 1000 VDC and a few hundred milliamperes of contact current would be more than adequate, although perhaps a bit pricey. Ordinary reed relays can also be used because the "latching" function is only used to reduce relay coil power consumption in the field. Reed relays are already low-power devices, so eeking out that last bit of wasted energy to minimize their power consumption could be unnecessary.
It all depends on how the equipment is powered in the field, and how long it remains powered without connection to a mains power source. With re-chargeable LiPo (lithium polymer) cells, a suitably designed, fully-charged, portable system should be able to acquire at least eight hours of unattended data acquisition and ideally a much longer period, perhaps days or even weeks, to allow for measurement changes effected by weather conditions. It's really a matter of how much money you want to throw at rechargeable batteries to determine how much operating time in the field is possible before batteries must be charged again.
You could add "bells and whistles" fairly easily, some inexpensively, others not so much. For example, active earth electrodes could have an illuminated LED on their circuit board to indicate that their relay contacts have been configured. Perhaps a pair of red and green LEDs could be used to designate positive or negative excitation connection to an earth electrode, and another pair of red and green LEDs could be used to designate whether an earth electrode is being measured with the positive or the negative input to the measuring device. Or perhaps this pair of LEDs could indicate the polarity of the measured earth electrode with respect to the other measured earth electrode. These are all simple, inexpensive, possibilities. Most could be configured during circuit board design for later inclusion in the project if desired. A more expensive whistle would be remote RF monitoring of the motherboard data acquisition, with data sent back to a camp site or motel room. Of course it is always better to "nail down" specifications up-front to prevent "design creep" and over-budget additions.
As far as cost is concerned, that would depend on factors that we have not considered yet. Cost of parts for, say, eight or ten pairs of earth electrodes, is perhaps ten percent of the overall cost of parts for the data acquisition and control motherboard, whose cost is influenced by the measurement accuracy and excitation accuracy required. There is a "sweet spot" for analog-to-digital conversion that currently is somewhere between twelve and fourteen binary bits for several thousand conversions (samples) per second. Higher resolution and faster conversions jack the cost up rapidly, so it pays to be sure what is necessary for publishable results, not just what is affordable.
A major cost of this project will be paying for the design effort. I am pretty sure it should require less than US$12,000 and less than a year to complete, but that all depends on the qualifications of the designer. A retired, experienced and interested designer could cost less, overall, than someone who is only interested in the renumeration that is obtained. That all depends on what kind of working relationship is established initially. Good luck finding someone local (ideal) who can do this work, but don't be afraid to look abroad either. There is a lot of good engineering coming out of the former Soviet republics, which is the first place I would look if based in Europe. Don't neglect Asian talent either, although there could be huge communication problems caused by language and cultural differences. Best of all would be if you learned to do the project yourself. There are members here who can help with that.