Going back to John's original questions, as Magnetoman suggests, the relative shapes of the sparking plug electrodes affect the ease with which a spark can be produced. Also, as mentioned in the Chicagoland MG Club article, the relative temperatures of the sparking plug electrodes also affect the ease of spark production. I'd always been led to believe that with traditional sparking plugs, it's the temperature that has the greater influence. Whether that's true or not, I don't know, but it's of no matter because both shape and temperature indicate that, with a traditional sparking plug, the non-earthed centre electrode is preferably negative. Rotating the end housing of a K1F 180 degrees will indeed swap the polarity of the magneto output.
As to John's second question, the end housing should, I suppose, be the way up that Lucas put it originally, and I think I'm correct in saying that they always fitted them with the channel for the cam lubricating felt at the bottom. Whether that results today in the preferred polarity of the HT output then depends, of course, on the history of the magneto. For example, has it had its magnet's polarity swapped? Has its armature been rewound incorrectly? At least with a K1F magneto, the brass end piece of the armature at the contact breaker end only fits the bobbin one way. With many other magnetos, the end piece can be fitted upside down and swap the HT polarity as a result.
To test the polarity of the HT output, the pencil flare method does work, but does need an experienced eye, and as Magnetoman suggests, you'll probably need a helper if the magneto is on the bike. With the meter method, don't use a digital multimeter. (I'd have thought there's a distinct risk of blowing it up, and they're not at all good at telling you anything about brief pulses.) A simple moving-coil ammeter (anything between perhaps 100 microamps and 10 milliamps) with its negative terminal connected to the HT output and positive terminal connected to the magneto earth will twitch up the scale when the points are opened slowly if the HT output is negative, or swing back against the end stop if it's positive. The Snyder polatity (sic) tester should work. If you want to make your own, simply connect the anode of a red LED and the cathode of a green LED to the HT output, and the cathode of the red LED and anode of the green LED to the magneto earth. If the red LED flashes when the points open, you've got a positive HT output, or if green flashes, you've got negative. (When doing this, it's important that both LEDs remain connected together, otherwise you'll likely blow one of them.) Another method is to use a 3-point gap as described on pages 366 and 367 of
http://naca.central.cranfield.ac.uk/reports/1925/naca-report-202.pdf. If the spark tends to bend away from the third teaser electrode, you have a negative HT output. If towards it, then positive. I've even heard it said that if you stuff your finger in the HT output and turn the magneto, you can tell the HT polarity from the shocks you get. I get enough shocks accidentally from magnetos without wanting to try this to see whether it works. Obviously, not to be attempted if you have a weak heart or pacemaker.
If you work out that you've got the wrong polarity with the end housing the right way up, you can either fit the end housing upside down, or if you have access to the required equipment, you can swap the polarity of the magnet.
Ken.