I guess now would be a good time to dispel this old husband's tale.
When the union between heatsink and transistor is undisturbed, dried out gr ease that is like cement is perfectly fine. I have conducted a few studies on it albeit subjective. That is that the temperature difference between a transistor and the heatsink to which it is mounted is not significant, AS L ONG AS IT IS NOT DISTURBED.
That assumes it was installed properly in the first place.
First of all you never spread the compound, you put a dab or a bead where i t will be squished out as it is tightened. this eliminates air pockets. If you have a big air pocket between there you would be better off with no com pound at all.
Granted, the old dried out compound does not have as good thermal conductiv ity as nice new wet goop, but it is seated over the years and works quite w ell. If you doubt what I say take a meat thermometer and check the case of a TO-3 and its heatsink on one that has been there for decades and ne that you remount with new goop now. In fact to make sure there is no error, clea n those surfaces with "things". Scrub pads, alcohol, maybe acetone, and fin ish with a coffee fitter to get the surface nice and smooth. Apply a dab or bead right where the ounting will exert the mos even force, on a TO-3 for example, that will be about in the center and it is normal for some of the goop to squeeze through the holes for the pins. then get that thermometer o ut after using the unit to get the heat up, like on a stereo amp, run both channels at like 1/3rd power for an hour, that'll warm er up. As soon as yo u detect heat, measure. It has JUST been produced so it is still conducting to the heatsink. That would be the most accurate comparative test, but not for absolutes. It is not a number except for "how much" one is better than the other. Your actual calories of heat/watts dissipation and all that shi t matters not. Iti s simply better or worse. Try it.