Yes that is true, and my comment wasn’t meant to imply that it was merely shooting in the dark, but rather that we don’t necessarily have a lot of people nowadays who actually understand how to create new tube types.
One of the factors to consider was that back in the 50’s, the height of tube production, RCA was grinding up and throwing away 50,000 tubes a day because they didn’t meet spec. This was an era when tubes were created because a circuit demanded it, rather than the other way around.
So the question of how the ‘best sounding’ tubes were made is not actually how engineers at the time would have thought about it - it’s largely a facet of our current capabilities which are generally limited to using old stock or replicating existing tube designs. Back in those days, engineers were more familiar with how grid spacing, cathode grounding and relative vacuum or EF potting could effect the impedance, transfer curves and thermal conductivity of a tube. The handbooks of course will tell you ‘make the grid spacers like X to achieve Y coefficient’ but the execution of said matter is then up to the mechanical engineers who are faced with building the darn thing.
As I noted before, it’s rather ironic that new production 300Bs command such high prices, because they’re one of the easiest tube types to build. They’re priced the way they are because of course any tube is costly to build nowadays and 300Bs aren’t in particularly high demand outside the small hi-fi community, but an average 1950’s valve schematic engineer likely would scoff at the notion that there was anything special about a 300B. By then, pentodes, beam tetrodes and even exotic transmitter tubes had already been invented. Companies like Brook who stuck resolutely to triodes were considered old-fashioned by their peers and eventually went out of business.
Transformers are largely the same way - the only way to build or replicate an amazing piece of Peerless, Freed or Langevin (W.E.) iron is to take apart and find out that they’re incredibly complex… trial and error was used to reach design parameters which were determined by folks doing the math. That brings me to the next point which is that the while scopes and some fairly sophisticated measurement tools existed at the time, measuring gear was not nearly as precise a science back then as now. Engineers still had to use their ears, and we encounter some fairly strange designs from say Rein Narma, the chief engineer of Fairchild and Gotham. If you’ve ever heard an original 660 limiter, or Gotham Scully/Westrex system, it’s jaw dropping how good they sound.
Open them up however, and it’s a fairly complex nested feedback design with over 39dB of loop gain, a theoretically ‘wrong’ transformer primary of 1600 ohms, and a host of other strange issues (instability with 811a tubes causing them to self immolate occasionally) So how does the darn thing sound so good?
The evidence on the other side is the old classic Lansing/W.E. amplifier which is simply an input transformer, a pair of 2a3s and an output transformer. As dead simple as one can get - and it also sounds amazing, but in a different way.
(Edit: I totally butchered this)
I can’t recall the name, but it’s an often used quote from, I believe, an EMI engineer which goes something like this:
“If it sounds good and measures poorly, or if it sounds bad and measures excellently, you’re measuring the wrong thing.”
Somewhat ironic given the current day propensity towards measurement. I think the takeaway is that folks back then really had math and design chops, and occasionally made magic based on the pen and paper limitations of the day.
As an extra little tidbit, industrial amplifiers from back in that era tend to sound far superior to commercially available hi-fi. We think of low power, zero feedback and somewhat warm/rolled off sounding amps. But plug in a McCintosh MI200 (if you can find one) and you might be shocked. 400W RMS of DHT power with wider bandwidth than some modern solid state amplifiers.