Scaling is a powerful analytical technique(*), but in some cases it can
be a little trickier than meets the eye.

Consider, for example, scaling a piece of TV twinlead to twice the
frequency so it'll behave exactly the same (both as a transmission line
and as a radiator) at the new frequency.

The wire diameters have to be reduced by a factor of two.
The spacing between the wires has to be reduced by a factor of two.

Luckily, if the scale model and the original are both in free space, then

The dielectric constant of the insulator remains unchanged.

And, almost always overlooked,

The wire conductivity has to be increased by a factor of two.
The dielectric conductivity has to be increased by a factor of two.

Fortunately, these last factors are usually unimportant. If the original
is made from copper, it isn't possible to scale to a much higher
frequency. But it's something to be kept in mind if loss is significant
and an accurate assessment of loss is necessary.

Permeability, incidentally, remains unchanged with frequency when scaling.

But even if you can neglect the conductivity scaling, you wouldn't be
able to run down to the store and buy a piece of the scaled twinlead to
use in your antenna for another band.

(*) Antennas are often scaled to higher frequencies for testing because
the scale model is a more convenient size. When I was involved in the
development of very high-speed sampling circuits, we often made scale
models of various structures (for example, coax connector to microstrip
transitions) at *lower* frequencies, so they'd be large enough to
measure and physically adjust.

Roy Lewallen, W7EL