Sol -- If I may, the replacement resistor is indeed the correct replacement resistor, but I think because of the way the resistor is failing (in spectacular ways), its failure is likely not due to the actual circuit it serves.
That resistor is one of two resistors that make up the total of R13 on the amplifier schematic -- R13A and R13B -- both 5.1K resistors, which are connected in series to form the 10.22K resistor shown as R13A+B. The particular resistor that has failed is the resistor that is on the ground end of the total R13 resistor.
But notice, its mate -- the other 5.1K resistor it is connected in series with -- is located in a complementary way near the edge of the board, and it is perfectly fine. But the resistor in question has failed twice now. Therefore, it is quite unlikely that there is a fault in the R13 circuit (the cathode resistor for the pre-driver, which is the stage immediately preceding the cathode follower driver stage for the output stage. If there were a fault in that circuit, then the R13 mate would also show some level of distress as well, which it does not.
More likely then, it is quite possible that the fault is actually due to the layout of the physical board itself. That is, look at the two large resistors that are mounted perpendicular to the resistor in question -- those are two 100K resistors connected in parallel to form the 50K plate resistor R17A/B for the "bottom" pre-driver stage. The two complementary 100K resistors nearer the edge of the board form the 50K plate resistor R16A/B for the "top" pre-driver stage. However!.......
Note that the 320 vdc source that supplies R17A/B runs directly under the resistor that has failed -- but this is not the case for the supply trace that supplies R16A/B. The 320 vdc trace supplying those resistors takes a physically different path than directly under the R13 mate resistor. Therefore:
The section of R13A+B that has NOT failed (1) Operates at a higher potential above ground, since that portion of the total R13 resistor is connected to the cathode of the predriver, NOT ground, and (2), Does NOT have a 320 vdc trace running directly under it. In fact, the trace running under it operates at only about 57 vdc.
But the section of R13 that HAS failed operates at a lower potential relative to ground, since one end of it is in fact grounded, and (2) DOES have a 320 volt trace running directly under it. Therefore, the difference in potential between the traces running under the two R13 resistors, and the R13 resistors themselves, is very different. I strongly suspect that this -- based on the way that resistor is failing -- is the real root of the problem.
Therefore, I suggest that the failed resistor be removed, and the board and components under, near, and around it be carefully cleaned, so that no carbon traces remain from the previous blow up event(s). Dirt itself can contribute to arc over problems under the right circumstances. So bottom line, get the area where that portion of R13 resides good and clean. Check the bottom of the board in that area as well. Then, when you installed the new resistor, mount it notably ABOVE the board itself so that the body of the new resistor and the 320 vdc trace below it are not in close proximity. I think there is a very good chance that this will resolve your issue.
I hope this helps!
Dave
PS -- As to your glowing output tubes, it sounds very much like the screen grids are glowing. Therefore, I would check that the complete connections between pin 3 of the two output tube sockets in question, and where they connected to their respective OPT primary lead are all good, with no heat stress cracks that might cause the plate connections to partially disconnect from their OPT lead. If that happens, then the screen grids would have to carry all (or much more) of the total quiescent current in each tube, which would cause those grids to light up for sure!
