Once again, as in September's circuit, in this circuit both the Cathode Follower's input overload voltage and linearity are increased by the unbypassed cathode resistor, but at the cost of a worse (i.e. higher) output impedance. 

MOSFET?
The MOSFET is convenient, not necessary. It was chosen because of availability and the relatively high bias voltage it requires, which allows the OP AMP's output to be closer to its power supply midpoint. A transistor or a FET or even a tube could be used. The best choice might be the FET, as these devices are closely akin to tubes in that they are both linear and run in depletion mode: they conduct current even when the gate voltage is negative relative to the source. A suitable candidate might the 2N4391, particularly in the TO-205 metal package. This FET is both quiet and linear, but it probably will require increasing the value of its source resistor to 300-400 ohms.

A pentode might be used, if its grid number 2 were to be attached to the B+ voltage. However, the output would have to muted at startup, while the tubes heat up. A 12BY7 just might work. A triode could be used, if the grid were driven positive relative to the cathode. Ideally, a lot more plate-to-cathode voltage is needed. If the negative rail voltage were increased to -25 VDC, this would allow for a greater selection of pass devices and allow for two identical current draw 12 volt heaters to be strung together, and yet still maintain a suitable power supply for the OP AMP.   

Which tube?
The dual triodes would certainly make for an easier job of building the line stage, as a single envelope could make a stereo line stage. For octal fans, the only choice is the 12SN7 (or two 6J7 or 6BX7, 6BL7's, whose heaters were tied in series). For 9 pin fans, the choices are more varied: 12AU7, 12BH7, 5687, 7119, 5963, or two 12B4's (or two 6DJ8 or 6FQ7's, whose heaters were tied in series).

How much plate voltage and current to use?
Of course the answer depends on which tube is used. Once again, a good starting point would be "more than you would think at first glance." In general, tubes sound best when driven hard, as it usually means moving their operation above their non-linear cutoff region. The harder they are used, however, the shorter their lifetime. For this line stage a good starting point would be 150 plate volts and 10 mA's of current, which would produce 1.5 watts of dissipation per triode.

Which sort of power supply should be used?
The Cathode Follower yields a good PSRR figure so this circuit does not require the supremely low noise DC power supply that a phono stage would. Furthermore, this circuit works in a Class A, single-ended mode of operation, which means that the idle current is high and varies little. Any of the power supply circuits from July's circuit of the month would do the job nicely. One trick would be to make a power supply from a flat-pack transformer. These transformers are available with a 48 VA capability and offer wonderful high frequency isolation from the wall voltage noise, as the windings are not layered upon each other as are those in a conventional bell case transformer. Such a transformer usually come with four windings: two primary and two secondary. The primaries are there to allow use at 117 VAC, when wired in parallel; and  use at 230 VAC, when wired in series. If one of the primaries were hooked up to a bridge rectifier instead, it would yield under load about 155 VDC. The two secondaries could then be wired in series and the connection be used as a center-tap for a CT full-wave bridge power supply for the negative voltage rail, which would allow for less voltage drop across the rectifiers and would use half as much expensive high speed soft recovery diodes as a full-wave bridge rectifier circuit.