It does fix the problem, although I don't think I understand the rest of the way fixed_point_format is supposed to behave.  According to the documentation,

The scaled format prints a scaling factor on the first line of
output chosen such that the largest matrix element can be written
with a single leading digit.

It then gives the example

fixed_point_format (true)
logspace (1, 7, 5)'
ans =

  1.0e+07  *

  0.00000
  0.00003
  0.00100
  0.03162
  1.00000

which seems alright.  But,

octave:12> fixed_point_format (true)
octave:13> x = [1e1; 1e3; 1e5; 1e7]
x =

       10
     1000
    1e+05
    1e+07

In this case, why isn't the 1e7 factor extracted and presented at the top as before?

It seems to require that one of the variables must have at least 8 significant digits.  For example,

x(3) = 1.234567e5
x =

  1.0e+07 *

   0.0000
   0.0001
   0.0123
   1.0000

Just seems to be the display routine.  I can edit variables in the Variable Editor and they are correctly changed in Octave.

Confirmed with 7.1 and 8.0 (on Linux) as well.