[Opcom]

Most all the good points have been made. A closer look is worthwhile, for me anyway, to try to get into why and how. So here goes.

Comments:
1.) C3 filters noise and hopefully the V4 and V5 cathode current fluctuations out of the heater current.

2.) I do not presume to state the designers' reasons for V1-V2 lower heater voltage, but a low voltage filament operation is found also in high reliability and low noise communication radio and testing circuits. H.H. Scott was not looking for power here. It is not for just any tube but apparently the 12AX7 will work this way.

3.) Normally the two 15K resistors R48 and R49 make up a small part of the heater current for V1 and V2, about 31mA. This may serve to somewhat stabilize the heater power for V1 and V2 as the output tubes V4 and V5 age or other conditions change. The 3K resistor R47 conumes about 6.4mA with the V1 and V2 filaments operating. R48 and R49 dissipate about 5 and 9 W respectively. Was it more economical to buy double the 15K 15W resistors than a set of 15W and 7W ones each?

4.) If V1 or V2 should have filament failure, then R47 the 3K resistor prevents V4 and V5 cathodes from going dangerously positive as mentioned, and also it cuts off V4 and V5 by the resulting positive cathode bias of around 94V. It also passes about 30mA when that fault occurs. At the same time R48 and R49 help hold down the unloaded voltage of the power supply.

5.) The above arrangement, besides supplying heater current to V1 and V2 from V4 and V5 cathodes and from R48 and R49, also puts all the other cathodes 19.5V above ground which is enough to reduce heater-to cathode leakage current in them.

I feel that the discrepancy in the shown heater voltages between V1 and V2 are a typo.

There are inaccuracies due to averaging but I'm satisfied. If there are errors, please point them out so I can understand better.