But I don't think there is a correct simplistic explanation of how to do this here or elsewhere? So I thought it might help others if I put one up - it would have helped me. Here's the infamous article kindly provided by RG Keen http://www.geofex.com/Article_Folders/ffselect.htm But it's a bit obfuscatory to the average homebrew tinkerer I think. Really I'm just spelling out what KindaFuzzy has already said viewtopic.php?f=10&t=8411&p=89198&hilit ... est#p89198 ...but I don't think that's a bad thing (if the mods do - feel free to flush this post down the FSB pan 
).The easiest way to do this, I believe, is to use a breadboard. You need a fresh 9v battery, a 2.2m resistor, and ideally a a 2.472k resistor... but I wouldn't drive yourself mad aiming for ideals (especially when dealing with germanium transistors). If you can get 2.47k that's good enough I think - I used a 2.2k and a 270r resistor in series to get 2.47k exact. You want to set things up on your breadboard along these lines (n.b. we are dealing with PNP transistors here - alter things to suit NPN if you need):
), here we go.1. Measure leakage: With one end of the 2.2m resistor disconnected and your multimeter set to 20v or 2v DC let the transistor stabilize* before noting down the reading you get. When you feel the transistor has stabilized and you've got a reading you think reasonably accurate compare it with the following guide:
0.137v = 50 uA / 0.05 mA
0.274v = 100 uA / 0.10 mA
0.548v = 200 uA / 0.20 mA
0.822v = 300 uA / 0.30 mA
1.096v = 400 uA / 0.40 mA
1.37v = 500 uA / 0.50 mA
N.B. uA = microamps and mA = milliamps (1000 uA = 1 mA).
E.g. A reading of 0.115v = a leakage of < 0.05mA
According to RG Keen transistors that are <0.30 mA are The Good; those that hit or exceed 0.50 mA are The Bad; those that greatly exceed 0.50 mA would be The Ugly I think
.2. Measure the nominal Hfe: connect the 2.2m resistor and, when you're happy the transistor has stabilized, note down the reading you now get and multiply this by 100. E.g. 1.89v = 189 Hfe.
3. Now the correct Hfe: subtract the reading for leakage from the reading for nominal hfe and multiply the result by 100. E.g. 1.89v - 0.115v = 1.775. 1.775 x 100 = 177.5 Hfe.
And that's it I believe. The example I've used above was a 2n404 - a 70s Germanium I believe when the technology involved in their manufacture was pretty advanced. Earlier transistors (the fabled OC range for instance) are likely to have greater leakage and often lower gain.
*Quite possibly you'll have to wait a while for the transistor to stabilize (especially if you've handled the transistor in question more than a few seconds). Some transistors seem to stabilize quicker than others - some I've had to leave for nearly 5 mins to settle down... meaning if you've got a lot of transistors to test, it'll take you a while. The differences between the early readings you get and the stabilized reading are great - leakage can initially appear to be very high and your transistor useless. 5 minutes later and it's dropped to a nice 0.40 mA or so. I hate to think of all the useful transistors that might have been binned by the hasty
Ac 128,127,172,125 and a AF117n but im not too sure how useful that one will be. Measuring them as i speak. Thank's alot Johnny 
and use a momentary switch (because as noted above, yes, it can take time for the measurement to settle). I ended up using the test described at Small Bear (but I think it's the same thing). Also, to conserve batteries and help provide a more consistent reading(batteries wear out and drop voltage), I made a compact -9VDC supply powered off the AC supply for my Marshall SE100 spk. emulator(something very ordinary and just a couple of parts, diode, regulator, couple of caps, connectors). I built it on a small ready made perf board with a power jack and 9V clip sticking out of it, and for the "housing" I used a piece of clear plastic from some packaging and stapled it together. You have to careful to not step on it, but it works okay. And also, I used cut down 8pin DIP sockets meant for wire wrapping for the Tr sockets which helped since they have long leads. Another thing I didn't try (that I thought might be useful in hindsight) was using SIP sockets(snipped) as "socket savers"(plugs for tubes meant to save wear on a tube tester tube socket) because the contacts can wear out after many repeated insertion/removals. It can take a while to measure a bunch but the GEO info is very solid and the measured/sorted ones I plugged into a FF circuit work great.
) I lifted my voltage/leakage guide from KindaFuzzy's post, and looking again there's a slight error! 274 mV was used instead of 247mV as representing 100 uA of leakage 
. (So great for a Tone Bender build for instance... that would prob sound best in the Winter 