One take-away from all this discussion is that it's very tricky to make good measurements at higher frequencies, especially with scope probes. It's hard enough even comparing probes attached to fairly well controlled signals and structures. Then think about hooking up to an actual circuit or DUT - everything will be different.
I think Chuck Harris said it best near the beginning of the thread, about not expecting good results using a 10X scope probe above 50 MHz (and the ground must be good too). Remember that traditional passive scope probes evolved from the tube days, for big, slow voltage signals. These are for general purposes, from DC to line to B+ and then lower circuit voltages. They're great at low frequencies up to mid-RF, and durable against say 400V LF just fine. You can try to measure almost anything with direct scope inputs or via probes, but you have to be aware of the limitations. You can readily get a qualitative picture of what's going on, which is usually sufficient to say if something seems to be working or not.
For more precision at higher frequencies, a true 50 (or 75 or whatever) ohm transmission line conveyance is best, where everything can be better characterized (but still never perfect). Between there and a 1 meg//some pFs input, you can make decent compromises and scale factors work out. For higher frequencies, lower R is of course better, so I like the medium 100kR, 10kR, and 1kR ranges, for instance. These can be tricky to use inside active circuits though, where the DC biasing and such can be affected by the probe resistance, so AC-coupling is often needed too. If the frequencies of interest are high up enough, the coupling caps don't need to be very large. An option for getting both flat low-R RF, and DC-coupling is to have a DC-offsetting function included, like in power rail probes for instance. These need to have very low source R, so are not suitable for poking around in circuits, but the same principles can be applied to add utility to medium-R probing systems.
Don't expect lower-R probes to be good for general purpose - go with the classic high-Z 100X/10X/1X ones then, which are comparatively indestructible. They really are different functions.
So anyway, I think you can make decent comparisons between various probes, but only in relative terms, not absolute. When they're actually applied, keep the circuit/DUT characteristics and probe limitations in mind, while you do mental math and adaptation to get a better picture.
Ed
