so a normal oscilloscope can't do it?
if you know what you are seeing and limitation of the entry level scope... analyzing in FFT domain may give you some clue. for example, using
http://www.soasystem.com/download/visadso/ a file of pure sine signal in csv format, generated from a simple computer software, not from imperfect real world, is provided in sample directory. you may import the file in (a) raw data and then (b) byte data, with each type of data you may plot the FFT and look at the THD of both. with theoritical perfect sine captured as real data, the THD is 0%. but when downsampled to 8 bit data, the THD is 0.3205%, this is the best a dso can give you, assuming your generator is a perfect sine generator with 0% THD, and the dso has a perfect sampler system, for example sampling period dT(n) = dT(n+x) with x is positive integer, no variable offset in the ADC etc to not screw up the Nyquist sampling requirement... when you have a little bit idea about this, ie your scope limitation, and how it looks like for a perfect sine's FFT downsampled to 8 bit on the screen, then you may take necessary action to compensate or subtraction, when you do a real measurement with your scope and the VisaDSO or any other SW capable of doing large data set of FFT like matlab etc. some other people swore by the averaging function of the scope, but i'm not one of them at this particular application. i may use it, but in conjuction with non-averaged signal, just to check the validity or to get an idea how far i may be off. but then, this is a poorman method, the proper way is to use the dedicated instrumentations.