Figure 1a. This figure shows the secondary tilts required to correct for transverse flexures of both the telescope truss and the secondary mirror mounts as the telescope changes altitude. No software corrections for these flexures were installed in the telescope software at the time of these measurements. The circles show the x-tilts of the secondary needed to minimize telescope coma and the squares show the y-tilts required. X-tilts are rotations of the secondary mirror about the Nasmyth axis. Positive x-tilts of the secondary are in the direction of your fingers if your right thumb points along the Nasmyth axis towards the NA2 port.

The small negative slope which seen in the y-tilt data is possibly real. It is reproduced in the data from the next night. A linear fit to these y-tilt data yields the function: y-tilt =-8.71 - 0.056 * altitude. The standard deviation of the y-tilt points after this linear trend is removed is 2.56". Below I will show that after the fits for altitude dependence are removed, the scatter in both x- and y-tilts on this night are identical. In terms of its effects on the imaging quality of the telescope, this possible altitude dependence of the y-tilt is not significant. At 60љ the y-tilt correction is measured to be approximately -12".

The line shown is a least squares fit to the x-tilt data. The functional form of the fit is shown at the top of the inset and was choosen to correspond to that assumed by the 3.5m software. The fit parameters m2 and m3 are in good agreement with those derived from data taken on 14 April 2000. The parameter m1 represents a zero-point offset to the tilt function which is not expected to agree with the data taken on 14 April because the secondary mirror was removed from the telescope between these two measurements. The expected errors to the fit parameters are also shown in the inset.

Figure 1b. This figure shows the the x-tilts required to minimize coma on the 3.5m telescope measured on the night of August 15. Once again, the inset shows the fit parameters for the line plotted. Notice that the data points on this graph show much greater scatter than that seen in Figure 1a. Subtracting the fit from these data and then computing the standard deviation of the residuals yields a scatter of 6.8". A similar calculation with the x-tilt data of August 14 yields a scatter of the residuals of only 2.4", nearly three times smaller than the scatter seen in this graph.

Because of this increased scatter in the x-tilts, the y-tilts have been excluded from this graph for clearity. However, the y-tilt data are very consistent with the y-tilts measured on August 14 (Figure 1a). A linear fit to the y-tilt data on August 15 yields the function: y-tilt =-6.48 - 0.097 * altitude. The standard deviation of the y-tilt residuals after this linear trend is removed is 3.3". A similar calculation yields a scatter of 2.6" in the y-tilt residuals of the August 14 data. At 60љ the y-tilt correction is measured to be -11", which is not significantly different from the -12" measured on August 14.

The slight negative trend in the y-tilt data and the absolute values of the y-tilts reproduce between the two night's worth of data. One very interesting point is that the scatter in the y-tilts on August 15 is nearly identical to the scatter seen in the y-tilt data measured on August 14, despite the fact that the x-tilt data showed an increased scatter by nearly a factor of 3! The "noise" seen in the August 15 measurements is quite specific to the x-tilt data. The source of this noise is not currently understood.

When one takes into consideration the increased noise in the x-tilt measurements, the fits for the data taken on August 14 and for the data taken on August 15 are consistent.