John Fields wrote ---
---
When measuring _total_ harmonic distortion, the contribution of each of
the individual partials is immaterial in that what's being determined is
the contribution to distortion that _all_ of the harmonics due to the
fundamental's presence contribute.


Furthermore, even if the contributions of the individual partials were
to be measured, their voltages would each be measured using a tuned
voltmeter and then the process of determining their contribution
determined mathematically. As a matter of fact, in order to measure the
power directly, the normal load would have to be disconnected and a
bolometer with precisely the same impedance as the load substituted for
the load. Expensive and more than just a _bit_ awkward.

You raise a good point John.
I was going to post something to the same effect
which goes to the heart of the original question by Svante

In the 'measurement' of distortion, putting aside the quantitative accuracy
problem for the moment, it was customary for measurements to be done in two
ways.

- The nulling of the fundamental and then measuring the rest of the garbage.
This also included a noise component which might (not) be significant.
It is, (here I don a flame suit at this moment), the more common way.
Quite often a HP machine (334) is a typical device for this work

- The wave analyser approach whereby distortion products are discretely and
separately measured. This I remember as being the older method. A person then
calculated distortion by a formula which (correct me if wrong) was along the
lines of ...

1 / square root of the sum of the individual voltages (squared).
There was a variant of this formula but others can quibble if need be.

Of course. If you _missed_ some by-products as the result of 'mixing' i.e. by
assuming only harmonic products, then accuracy suffered, but hey, "the world
ain't poifect anyway". You _could_ , with equal validity, have measured the
products as currents (but voltmeters were more common).

This "voltage" approach is probably (well, I would contend that it is) the
single most important influence on the definition and approach to
measuring (and defining) distortion. All of this is predicated on the purity
of the original signal source (sine) which, in itself, is a separate subject
-)

.... Stands back .... prepares grapeshot to repel the oncoming attack ...