We (well, Peter S and I) had some discussion a while back about riverine
borders - whether if the river shifted its course the border stayed put or
moved with it. I was for the border moving (on no evidence at all, I must
confess).
My copy of Prescott's Maritime Political Boundaries of the World finally
arrived today, and while I was browsing for information about high seas
enclaves (of which considerably more later, I'm afraid) I came across a
riverine boundary that is acknowledged to move - the section of the
USA-Mexico border marked by the midline of the Rio Grande.
Here's Prescott:
"The two countries have used a neat solution to the problem posed by shifts
in the mouth of the Rio Grande. Since the centre of this feature is the
landward origin of the maritime boundary it would obviously be inconvenient
if the alignment of the the territorial sea boundary moved with the river.
The problem was largely overcome by fixing a point which acts as a hinge in
the boundary. This point is located 2000 feet east of the centre point of the
river's mouth as it existed at the time the agreement was signed. Seawards of
the fixed point the boundary is shown in a map attached to the agreement ...
Landwards of the fixed hinge the boundary will consist of a straight line to
the centre of the river's mouth, wherever it is located. This means that the
short landward section of the maritime boundary will pivot about the hinge."
This arrangement makes no sense at all unless the US-Mex land border is
deemed to stay with the midline of the river whenever it shifts.

Grant