Hifi, why bother?

[Mike Watterson]

Although they all have just about identical characteristics, I reckon there are 3 different electrode geometries.

I would think so. Shrinking was done of the first Sylvania 1.4V valves from loctal/Octal button base to RCA B7G button base. Some nearly identical characteristics. Octal valves actually switched from pinch to button bases. Typically the tube sits deeper into a collet rather than on a higher Bakelite surround, though the wires are still soldered into the pins.

The DL92 was mentioned earlier (1948?). Either a copy or likely licenced 3S4 (November 1940) designed by RCA based on the Sylvania Octal 3Q5 (= DL33). That's why the DL92 is a Beam Tetrode.
https://www.radiomuseum.org/tubes/tube_3q5.html

The 1C5 is the 1.4V only 100mA version, the DL2 is the P8 Edge connect version. The 1S4 = DL91 is the B7G version (1940 and 1948).

Probably all these were Beam Tetrodes.

Philips did rimlock 8 pin base valves before the 1.4V B7G. Telefunken had slightly lower filament voltage and parallel only Y8 base (for NiCd) from during WWII up to early 1950s when the last makers switched to B7G.
Early 1.4V valves had nickel filaments due to accuracy issues with tungsten for series operation. The later Philips 25mA series have tungsten filaments, which might be spring loaded. The Russian rod tubes have spring loaded filaments. The DK40 was roughly 1LC6 performance so used on some UK SW sets as the DK91 was poor. US makers used the US version of the DK91 only till the 1L6 was available. The DK92 is similar performance for SW. The DK96 is poor at higher SW, so the DK92 was used virtually till battery valve radios ceased.