[David Simpson]

Those are really good results. Right enough, the first 6AU5 tabulation is pure DC, the second is my MK3(AC), and the empty column is for your results. I always include my MK3, partly because I can regularly check its spec, but more so to see that any wee "kinks" are duplicated. These offsets from a pure Xsquared Gm graph happen many times with many valves. As I've said before, a maths teacher would want a student to draw an averaged-out pure curve, but a test equipment technician wants to see kinks, warts & all. For DC testing, I'm constantly monitoring Vh & keep tweaking to exactly 6.3V, or whatever. Along with any tweaking of Va & Vs to exact book values. AVO VCM's, most now of pensionable age, can exhibit lowering AC Vh's across H+ & H- with valves that draw 1 to 2 to 3A of Ih. Yep, AVO Manual's calibration spec. might show a generous % of allowable Va, Vs & Vh, and indeed for example, show higher switch selection voltages off-load. But then, for many folk, the very nature of the RMS - ish values given by AVO, can be jolly confusing.
I was extremely lucky with my DC tester construction in having an old AVO MK1/2 as a donor & thus was able to use their valve holder panel complete with its 9 way thumbwheel switch. However, long 9 way/9 contact wafer switches,(Yaxley type), can be sourced & made up. RS used to supply construction kits. A lot better than fannying about with swapping banana plugs, leads, & sockets.

Regards, David