Well, thanks for all the input.
Firstly, there's a typo made in haste. The input is wrong. With the resistors mentioned below, the voltage on the input leg of the regulator is 4.3VDC. This *should* be just adequate headroom.
R4 in the full schematic; There is a bit of discussion on the web site I pointed to about it being inadequate to dissipate enough power. The 22 ohm resistor was found lacking by a previous builder, and I have 2 of 27 Ohms & 5 Watts in series. This drops the voltage to 4.3VDC without too much dissipation per resistor. I would have used 2 Watt jobs, but they're hard to source here.
The divider network in the full schematic is not what I'm using. I will be using the values in the little sketch, of 100 and approx 500 Ohms.
the heat sink is a"U" shaped piece of finned aluminium that was bought for the purpose, but probably insufficient. The back of the U shape is about the same area as the back of the chip, and the two sides are also about the same size, less the slots. There 's plenty of room in the cabinet, so a bigger heat sink wouldn't be an obstacle.
Current. No, I haven't measured, but that'll happen after breakfast. It should not be much more than 350 mA when stabilised as there are effectively 6 filaments in parallel, each drawing ~50mA.
The datasheet says the LM317 should be capable more than this. A larger sink may be all I need from here.
Went off half cocked here, I'm afraid. When I measure the actual current drawn I'll be in a better place to assess whats next.
Thanks again everybody,
Clay.