metrology
You give no details, but It's likely that your problem is MUCH more complicated than which coax to use.
Unless your divider is buffered, the considerable capacitance of the coax may or
may not significantly alter the divider ratio...depending on which coax,
how long, the cap of the scope probe attached to the other end.
Terminating the coax in 75 ohms likely renders your divider useless...maybe.
Not terminating leaves you open to significant transient response fidelity issues.
Coax has both RMS and Peak ratings. While it's possible to exceed ratings and
have it work, it's not recommended.
You don't disclose, but one might infer that you're in an industrial environment
with considerable risk of DEATH!!
While it might work, several classes of people salivate when you
exceed specs, including
Safety inspectors
Personal injury lawyers
Your boss's boss
Your heirs...
Some other things to think about:
Scope probes have peak and rms voltage ratings that are VERY
frequency dependent. Your short pulse may exceed the probe
spec by a large margin, check it out.
Certainly depends on other unspecified constraints, but you're
more likely gonna have to put a voltage divider right at the cap divider
and run terminated 50-75 ohm coax to the scope. Make the division
ratio as big as the scope sensitivity will allow to reduce loading
on the cap divider. Once you get the DC part sorted out, you still have to
worry about frequency compensation/transient response.
You may have to correct the transient response of one or both dividers.
The proper way to compensate is to put a resistor over the high voltage
side of the cap divider. But this may not be possible because of the voltages.
There's relatively simple calculus to determine what to do. Problem is that
there are parasitic elements that you don't know. And that's what
you're trying to compensate.
But even after all this, your scope trace may not accurately show the pulse.
There's a lot of ART involved in measuring transients accurately.
You need a way to confirm that the measurement is accurate.
Something as simple as transient ground potential differences between the scope
and the measurement apparatus. If I'm interpreting you correctly, you've
got 400KV stuff going on.
The customers for your products expect you to have the required
experience to get the right answer.
This does not appear to be a "which cable" issue...