Mirror Inversion During Cooling?

My 12" mirror (GSO F/4.92) was tested and shown to be slightly undercorrected. This is confirmed by star tests. However, during mirror cooldown star tests suggest that the mirror is over corrected. Is there enough difference in thermal expansion and mirror distortion to cause this change? How much change is possible during cool down?

I think that a lot of the problem is due to lack of precision annealing of the glass before it was ground and figured. This will cause change with temperature and sometimes cause poor performance at or below a certain temperature.
Also the mirror cell adjustment is a big factor as is the support provided by the cell. Pinching will produce astigmatism every time.
Over correction is common on mass produced mirrors or any mirror for that matter. Under correction is less common but does occur.
I remember a conversation with Steve Kennedy where he stated that a mirror showing proper correction on the bench can show over correction in the scope under viewing conditions. He said that he slightly under corrects the large mirrors to compensate for this.