As Ed said, there are many variables and very few absolutes.
There are basically 2 ways to bias a 2A3. Both require the grid to be more negative than the cathode, or to look at it the other way the cathode more positive than the grid.
One way is to hold the grid close to zero via a resistor or choke to ground, then put a fairly large resistor from the cathode to ground so you have a positive voltage on the cathode. This requires a larger B+ since the cathode resistor drop and the tube are in series. It also usually requires a high quality cap to bypass the cathode resistor.
The other common way is to use a small resistor from cathode to ground so it is close to zero and then apply a negative bias to the grid. This requires a lower B+ and eliminates the cathode bypass cap but adds another supply for the grid bias.
Both have their merits and both are commonly used.
Ed's point about floating filaments does not apply to directly heated triodes such as a 2A3 since the cathode and filament are the same element. That technique is sometimes used for input and driver tubes like a 12AX7 or 6SN7 which have an indirectly heated cathode where the cathode and filament are not the same element. If the difference of potential between the two gets too big it can be detrimental so the filament is sometimes biased via voltage divider to a potential that is close to the cathode.
There are basically 2 ways to bias a 2A3. Both require the grid to be more negative than the cathode, or to look at it the other way the cathode more positive than the grid.
One way is to hold the grid close to zero via a resistor or choke to ground, then put a fairly large resistor from the cathode to ground so you have a positive voltage on the cathode. This requires a larger B+ since the cathode resistor drop and the tube are in series. It also usually requires a high quality cap to bypass the cathode resistor.
The other common way is to use a small resistor from cathode to ground so it is close to zero and then apply a negative bias to the grid. This requires a lower B+ and eliminates the cathode bypass cap but adds another supply for the grid bias.
Both have their merits and both are commonly used.
Ed's point about floating filaments does not apply to directly heated triodes such as a 2A3 since the cathode and filament are the same element. That technique is sometimes used for input and driver tubes like a 12AX7 or 6SN7 which have an indirectly heated cathode where the cathode and filament are not the same element. If the difference of potential between the two gets too big it can be detrimental so the filament is sometimes biased via voltage divider to a potential that is close to the cathode.