Gort is right about all the transistors needing to be in place for accurate measurements. Also, a voltage check at the pins of the individual transistors is in order to effectively determine which ones are faulty. Your TA-100 could be hosed, but until you're sure the supporting cast is good or bad, you won't know. Also, high DC voltages are as a result of supply rail voltage, not the audio signal (AC). The 26.8V that Hopjohn refers to (for meter calibration) is AC voltage, not DC voltage. You can test the transistors out of the boards for a quick and dirty check of obvious errors, but an in situ test of voltages at the E,C & B is more thorough. Don't forget, the board has three diodes and a few electrolytics that if faulty, will cause voltage anomalies.
I'm more curious as to why you have close to rail voltage on pins 4 & 8. Qe1 & 3 are (NPN) matched pairs as are Qe5 & 7 (PNP), so if you replace one of the pair, replace the other as well. When you think you have the supporting cast dialed in, start at the TA-100s and work your way towards the inputs looking for voltage anomalies. If you find one, it doesn't necessarily mean the transistor your checking is bad...it could mean a component in the surrounding circuitry has gone kaput. You just need to narrow it down and fortunately, the other channel works so you a valid comparison of measurements. When you nail it down -- assuming it's not the TA-100 -- you probably won't have exactly 0V DC on pin 5 of the TA, but if the voltage reading is low (mV territory), you can trim it with VRe1, to 0V.