I measured the output of the clock. At that point in time I did not have the tech to measure the output of the dac with adequate resolution. We were able to perceive the difference doing blind tests. We built a and marketed a transport which dominated the high end rooms at CES for ~ five years. This was ~ 1995ish. A lifetime ago.
The defense guys who need very precise clocks knew about this problem a long time ago and had some very interesting solutions.