I don't know of a "site", but the principles were pretty simple.
Two terminals supplied a <high> current to the object under
test D.U.T.  The magnitude of the current was the controlled variable
so contact resistances (and the induced voltages)didn't perturb that
value. The other two terminals measured (with high impedance)
the voltage resulting from the applied currents. Since this
'took' negligible current from the D.U.T. it did not
steal current from that applied, and (more importantly) didn't
add in its own (ohmic at least) voltage errors. I don't recall
there being much 'magic' to all this, other than some attention
to thermal  (dissimilar metal thermocouple effects if not
balanced) and in extreme cases chemical effects on the voltage
probes.

Probably the 'last word' on this is the "Review of Scientific
Instruments"  R.S.I. Journal at your local university, where
sensor and measurement issues are carried out a few more decimal
places than elswhere. Also N.I.S.T. pubs (formerly Nat. Bureau
of Standards) which are in part online.