Figure 10. The circles show the thermal defocus corrections that were observed on August 15, 2000. These data were derived by combining the defocus measurements seen in Figure 6 with the thermal measurements of the truss seen in Figure 7. Knowledge of the telescope collimation coefficients which were active on the night of August 15 was used to translate the thermal measurements in Figure 7 into the actual piston measurements which the telescope applied on the night of August 15. The telescope piston measurements were added to the defocus measurements shown in Figure 6 to give the defocus measurements which would have been observed if the telescope collimation had been turned off. I then subtracted from these data the theoretical altitude sag [271 * sin(altitude)] which was calculated in Appendix A. These are the data shown in circles. The circles therefore represent the thermal defocus corrections that one would have seen on August 15 if the telescope had behaved as theoretically expected as a function of altitude.

The thermal behavior of the truss is known and I assume here that I also know the thermal coefficients which describe the secondary cage sag as a function of the truss and rod temperatures. The major unknown is then the thermal behavior of the rods on the night of August 15. I then assumed that the rod temperature was identical to that of the truss in every respect accept for the thermal time constant. I then fit the thermal defocus data given by the circles above using only the thermal time constant and an arbitrary "zero-point" constant as fit parameters. The fit obtained is given above in the solid line. The functional form of the total fit (including the assumed altitude dependence!) is shown in the figure inset. The fit parameter m1 represents fit solution for the thermal time constant of the rods in units of inverse seconds. The parameter m2 is the somewhat arbitrary zero-point constant of the fit.

The fit for the thermal time constant of the rods is reasonably consistent with the measured data shown in Figure 8. M1 corresponds to a thermal time constant of 0.37 hours (22 minutes) for the rods.