Long-term stability of optical surfaces

It is not uncommon in high-precision testing of optical surfaces to find that repeated measurements produce inconsistent results. The most obvious cause is where the test system is not in thermal equilibrium, or where, for example, in the case of massive telescope mirror substrates, the support structure loading has changed. But even where the ambient temperature is precisely controlled and the test setup, procedure, and overall environment are apparently unchanged, it is a matter of experience that unexplained discrepancies can sometimes remain. Dew reported the results of long-term tests of a 30.5cm-diam optical glass flat with a thickness of 5.1 cm. The glass type was Chance hard crown ~HC 524589!, classified as ‘‘fine annealed,’’ with no birefringence evident in a strain viewer when viewed normal to the two flat surfaces. During a period of 8 yr it was found that the optically flat surface changed systematically with time to give a cumulative error of 0.4l (l5546.1 nm), the rate of change decreasing toward the end of the period. Other flats measured by this author also revealed changes with time. No apparent correlation was found between the level of observed strain bire-