How do I choose a transistor to replace an unknown one?

Situation: I inherited a Soundcraftsmen MA5002 power amp.  One channel (the B channel) has gotten increasingly distorted. 

On the scope, the B channel can be seen to be clipping on the negative peak of the wave. 

The B channel's heat sink, while not overheating, is substantially warmer than the heat sink of the A channel under the same load.  This leads me to suspect one or more of the 4 transistors on that heat sink. 

I got the schemo and service manual and parts list.  This let me narrow down the search to one of three transistors.  Unfortunately the manufacturer used transistors just labeled "#4 power".  I expect that these were simply commercial transistors that they relabeled.  (I looked carefully at the transistor case, to no avail - I expect they used solvents to remove the original numbers.)

So  finally the question: is there any way I can make an educated guess as to what to replace these with?  I expected to see a lot of NPN transistors in TO-3 packages, but so far I haven't found very many. 

Thanks for your help.

Did you find the following links when you were googling this amp  ?  Seems to indicate that they are Hitachi devices.

http://www.diyaudio.com/forums/solid-state/16774-soundcraftsmen-ma5002-power-output-type-help.html
http://www.diyaudio.com/forums/solid-state/16690-soundcraftsmen-amps.html#post194248
http://www.sound-smith.com/fets/

These are excellent replacements for the Hitachi 2SJ-K series mosfets
Exicon ECF-series
http://www.jprelec.co.uk/store.asp/c=826/Power-Mosfets-Audio-Applications
Datasheet.
http://www.amorngroup.com/back_end/amornshop/image_shop/shop_0000001999.pdf
A lot of power amps of this age used the hitachi devices.....
I have a couple of HH X1200 amps that i repaired using the Exicon replacements
with total success....
I would also replace ALL... the smaller electrolytics in the amp as well !

Thanks for all the help so far - looks as if the MJ15015 and recapping (which I've never done before) may be the answer.

OK, more info - the outside of the case has ~12V on it when the unit is running... AND

the schematic shows that they are NPN devices.  Is there any reason to suspect that they would show an NPN when the transistor is actually a FET?  AND

when I take them out and test them with a diode tester on my multimeter, one pin and the case give a ~0.5 reading, that pin and the other pin reads ~0.5, and none of the other pin combinations shows continuity.  Wouldn't a FET or MOSFET show continuity between the source and the drain unless there is voltage applied to the gate? 

Any advice on recapping?  Find and replace all small electrolytics is my guess. 

Thanks again, folks!

Hmm I could be barking up the wrong tree here.........? :
A scan of the schematic would be helpfull to get to the bottom of this !

Rob Flinn's link http://www.diyaudio.com/forums/solid-state/16690-soundcraftsmen-amps.html#post194248

got me the answer I needed: an MJ15024 is the right replacement. 

Thanks for everything! 

FETs are different animals.

Nchannels will show NO continuity(open) between source and drain when the gate is grounded and will show very low ohms when the gate is high.  Likewise, the gate should show NO continuity between itself and either the drain or source at any time.

Pchannels will show short between source and drain when the gate is grounded and will show open when the gate is high.  You should see NO continuity between the gate and either the drain or source at any time.

OK, having read the description of expected tests with FET's I think I get it better. 

Here's a JPG of the relevent part of the schemo.  There are six transistors all labeled "Power #4" in the parts list.  Three of them are in a direct line under the letter A, the other three under the letter B.  Those look like NPN's to me...

Alden