I was thinking about the RS925, as being one of the best resistance decades ever created.
Then I was thinking, why not use a 4 wafer single pole switch to select individual resistances in a 4 wire setup.
With a 4 wire setup switch contact variations should be negligible, the switch used has a contact resistance of below 1 milli-ohm.
Comments. criticisms, post them.
**I'm not building this one(yet), it is just a proof of concept.**

I did a cable tester that way and still works great even after many years.
However this is the metrology section and who knows what they will come up with. 
What level of precision do you want to obtain?

If the switches have significant contact resistance, then each configuration would have to be calibrated individually.

If the switch contacts do not behave the same every time they are actuated, then the entire setup is probably doomed to not be metrlogy-grade.

The switch contact resistance is 1 milliohm(and below), generally repeatably to +/- .2mOhm when measuring the resistance contact variation.
I will test the actual switch Saturday, or Friday(when it is scheduled to arrive). Then post up some information.

If the switches have significant contact resistance, then each configuration would have to be calibrated individually.

If the switch contacts do not behave the same every time they are actuated, then the entire setup is probably doomed to not be metrlogy-grade.

This is not a problem, actually in a 4-wire ohms configuration, any switch or cable resistance will be compensated, if the wiring is done correct (its straightforward). Of course, you anyway have to calibrate each resistor separately). Also, changes in the contact resistance will be eliminated, because of the same reason. Theory is simple, and lots of info available.
The only problem that may occure is thermal EMF voltages, If you have an EMF voltage of say 1uV, and the voltage drop accross the resistor is 1V, the error is 1ppm. However, that can be compensated as well, most precision meters have a mode where they measure the EMF voltage allone, with current accross the resitor zero, and do the math then. This mode should even be used if no switch is used. Impact on accuracy also depends on voltage accross resistors, and is highest with low ohms values.

I do not see any reason why it would not work for selecting a single resistor at a time. However, most commercial products of this kind have multiple decades wired in series, and I do not see how this would improve contact resistance for all but a single set of contacts in the case of decade resistors. I guess you could make something that if a particular decade is set to zero, it would keep the four wire connection, so 1,000,000 Ohm and 0,001,000 Ohm could both use four wire sensing all the way to the resistor. 1,001,000 Ohm would still have to deal with contact resistance of all the switches between the two resistors, however.

if you look for switches, I can highly recomment ELMA
They are pricy, but I am very happy with the products and use them >30 years.

What would be a reason for the switch? It could be possible that 4 more binding posts be a cheaper solution than getting nice switch?

Lets take a look at the switch I ordered, in all of its NOS glory. If you exclude shipping, this switch cost less than 1 of Pomona's AuCuTe binding posts 

Taking the box out of its shipping package, the brand seems familiar

Opening up the package, we are presented with a small envelope containg the set screws.

Pulling the tab open reveals the manual. I will scan both sides, and upload them. It might be a few days/weeks though.

We are then presented with the switch, still wrapped in its plastic bag.

Removing it from the bag, reveals a switch.
I have started doing some quick testing, and show the problems I encountered.

They are nice switches, used quite extensivley at work many many years back and have managed to collect a few from junk toss outs. Nice to get a NOS one.
I did upload the spec sheet here but forgot which thread its in, too long ago  . So a fresh upload is needed  .

I found your post:

https://www.eevblog.com/forum/projects/kelvin-varley-divider-(and-precision-voltage-source)/msg74864/#msg74864

That is from the higher end type of switch. I have the economy specced one(Still would cost several hundred today)
It is interesting comparing the shaft types, and the moving contact, that attaches to it. There is a difference. When I'm done, I will mirror the information on my 4232 Thread.
Even though the switch is specced at 1 +/-0.50 mOhms, it does better than that in testing.

So I built myself a pair of kelvin clips, using the widely available cheap type of clips.
eBay auction: #http://www.ebay.com/itm/2-PCS-IC-SMD-Chip-Alligator-LCR-Kelvin-Clip-Copper-Gold-Plated-Clip-Test-Clamp-/322147974085?hash=item4b0183fbc5:g:PKAAAOSwnFZXXWbC
While they are decent, They suffer from a lot of axial movement , which causes them to fall off of the terminals. Unfortunately that lead to a bunch of widely varying readings. That was remedied by using the adjacent contact lug as a brace( If you guys want I can illustrate this).
The silicone wire I used was 14gauge wire, I loved this. I had never actually worked with silicone wire before, it is rather enjoyable. The one downside to the many strands of wire that is used, is that you are very likely to lose a few pieces when stripping off the insulation(I use a mechanical stripper,  those with hot wire strippers should be more fortunate).

The 4mm banana terminals I used are the AMASS type. I tried them on all of my multimeters, and power supplies. They were a tight fight on all, except for a 50yo precision lambda power supply. I plan on buying more of the banana jacks, I figured that they will easily get a bunch of use.
Now for the testing.
I used the Ballantine 3205B set in pulse mode for the resistance measurements, as If I used my 3456A, it would be down in the noise.
So lets get on with the testing…

We can see the switch contact resistances, broken up into positions and poles. As expected the 0-position is constantly at the lowest switch resistances.  As I was testing I realized that the resistances were rather uniform. With 84% of measurements falling between 0.70milliOhms, and 0.85milliOhms.
The one measurement that always seemed to stand out was that the Pole(3) contact resistances were always higher than the rest. I figured it would either be 2 or 3 due to the distance between bearing surfaces on the central shaft. Though thinking about this it makes since, as the axle of the switch is not as precise, when compared to the other type 31 switches I have disassembled. That will be later, I will not do a complete teardown(due to the much different nature of this type.

Lastly I will present the data in a histogram, that is not broken down at all. For a switch that is specified as 1 +/-.50 mOhms, it easily meets those specs.
Next update will be a short one, just going to do a noninvasive internal comparision.

I found a successor to the type 31 switch,
Multitech 820
Contact Resistance: Less than
    1 m?
Insulation Resistance: more than
    1,000,000  M? @ 400 C. 
    (terminal to terminal and to shaft).
Positions: up to 12 positions,
    shorting  (30 degree indexing).
Poles: 1 or 2 poles/deck.
Decks: 1 to 6 decks/switch.
Current Carrying Capacity: 5A
     max. switching @ 28VDC,
     1A @ 125VAC
Capacitance: Less than 1 pF,
     contacts to shaft.
Switch body diameter: 1.62”
External shaft: ¼” stainless steel
Detent rollers: stainless steel
Internal shaft: glass-filled
     polycarbonate
Voltage Breakdown: 2000V RMS,
     terminal to terminal and to shaft
Indexing: 30° or 60° available,
     nylon detent for smooth action
Thermal EMF: Less than 1  ?V
     Extremely low. Silver contacts to
     silver alloy brushes
Continuous rotation, fixed or
     adjustable stops to limit rotation 
     to number of positions desired.
Construction: decks: Diallyl
     Phthalate (DAP), totally enclosed
     – dust tight
Mounting Bushing: nickel plated
    brass 3/8-32 UNF
Front Housing and End Plate:
    Zytel 101 Nylon
Contacts: Stationary-Fine
     silver-clad copper
      Wiper: Silver alloy, self-cleaning
      leaf type
Terminals: Solder type with
    .067”Ø hole
Losses: Low electrical losses at
    frequencies up to several MHz.
Long Life – 50,000 cycles min.I'm going to try emailing them to see what damage could be done.(I would never buy one, I'm a poor college student... )

Vgkid,

To be a volt nut is to take a vow of poverty. I think it's in the TEA commandments?

Vgkid,

To be a volt nut is to take a vow of poverty. I think it's in the TEA commandments?

I know ... To go from working, to not working(got fired), makes ebay a depressing place...
I got a datasheet from Multitech, seems similar. I wonder what other brands made low resistance switches. 
Pricing is not to bad $322 for a 4 deck 11 position switch. That is a bit cheaper than what the equivalent L&N switch would be, though slightly more expensive than my version.

---