For use with a typical 100MHz scope? with a 1Meg input?and (say) a parallel input capacitance of about 18pF, the ideal 50R terminator would have a 50R resistor and a negative capacitor in parallel with it with a value of -18pF.
As negative caps aren't realisable with passive parts, the next best thing to do is to add some inductance to the 50R resistance to try and mimic a -18pF capacitor over as wide a bandwidth as possible. I suspect that some terminators designed for scopes will do this to some degree.
Compared to terminating with a perfect 50R resistor, this should give a lower VSWR over a fairly wide bandwidth when the (slightly inductive) terminator is added to the scope input as the inductance will offset most of the 18pF capacitance of the scope over quite a wide bandwidth.
