Yes the bulb method is the simplest reliable one, provided you can identify the primaries. Also one should always check the bulb before and after testing.
Otherwise, if you aren't sure which are primaries, you need to inject current limited low voltage AC, and measure voltages, then connect selected windings in series and determine whether the voltages add or subtract to check phasing. Jwillis suggested this method back in reply #25 above, but IMHO 24V is too high a test voltage as many transformers have secondaries that are nominally 12V or even lower, so if you apply 24V to such a secondary, you risk saturation and even insulation breakdown of the primary + the as-yet unidentified primary is a HV hazard. IMHO 6V to 8V RMS is a good range for the test voltage source. Depending on its max output level, your bench signal generator may even be suitable, otherwise get a bell transformer, or an old heavy low voltage unregulated switchable output voltage wallwart you can crack open and mod for AC output and a 15R 5W resistor to put in series to limit the current.
Potted and toroidial transformers with HV secondaries and no data are a PITA - it can be very difficult to determine which is the primary, and if you get it wrong, you may only have basic insulation between your choice of primary and some of the secondaries. Carefully measuring capacitance between winding pairs can help determine their proximity, and thus provide some estimate of the likely insulation between them.