*Footnote to my post above-
The clock-rate difference "drift factor" is pretty much constant in that each clock may exhibit significant difference over time in comparison with another clock, yet will still be very consistent with itself unless something is very wrong. Here's one interesting example of that- Say one records two sources using two separate non-clock-linked recorders only to find there is significant drift between sources when transferred to a computer for playback. If one instead plays back the resulting files on the same non-clock-linked recorders which originally recorded the files, the sources will effectively retain synchronization for the the entire recording without otherwise having to correct for the drift. That's because each recorder then operates as it's own complete self-contained sample and reconstruction device, both sampling and reconstructing using it's own clock, which remains consistent with itself. Each recorder is it's own analog>digital>analog world in isolation, and effectively cancels out its own clock rate error, as long as that error remains constant.
The trick in actually doing that is accommodating for the inevitable time-offset between sources at the start of playback. I used to do this before I owned multichannel recorders, using two original Edirol R-09s , accommodating time-offset by using a series of rapid jabs to the play/pause/play button of whichever machine was slightly ahead of the other. It helped greatly if the two sources were playing back through separate speakers and I could place my head midway between them to carefully and listen for full elimination of delay and phase effects while jabbing. It took concentration and looked pretty funny, but it worked for multichannel playback, and helped me develop the ability to identify very slight degrees of misalignment by ear. I also tried it a few times with three separate recordings, which was incredibly difficult!
Non-constant drifting of the clock rate can of course happen as well, it's just usually more subtle. At the level of digital samples that manifests as "jitter" distortion. At an order of magnitude larger time-scale, analogous time fluctuations manifest as tape-speed "flutter" distortion, and at a still larger timescale the "wow" distortion of a record-player turntable. All these are "clock" speed-rate distortions of various types in different mediums.