I don't say they didn't find anything useful. But as you pointed out, commercial considerations were probably a large driving factor for their "scientific" conclusions, which uncritical thinkers tend to take as the loudspeaker design gospel because it was done under "careful controlled conditions".
Actually tubes and transistors are both pretty inaccurate as amplification devices. Triode Tubes exhibit a 3/2 order transfer function, Mosfets roughly a quadratic relationship and bipolar transistors are some other higher exponential function. So, as a purely amplification device, triodes are in fact the most linear amplification device ever invented. That is just a scientific fact. How they are used is a different discussion. Using a bipolar transistor like a SET, for example probably won't end up working or sounding very good. FETs do better but then their transfer function is closer to a triodes than a bipolar transistor. As tube amps and transistor amps are designed though, typically a transistor amp has lower objective distortion, although the transistor amps that don't use feedback start having distortion levels approaching non-feedback tube amps. The question then becomes what sounds better to a human listener? Objectively the lower THD and IMD amp should be it and then done. In reality, it is the distortion pattern and level with SPL that matters much more. Geddes demonstrated that THD and IMD are non-predictive of sound quality and some of the lowest levels had the most objectionable sound to the review panel.
You can call it euphonic if you want, and in extreme cases it is clear coloration, but I would say it is not euphony but the absence of AUDIBLE distortion. Distortion of the right character and level is not euphonic it is invisible. The design itself has consequences on the pattern produced and thus it's audibility.