There is an issue with the K-band domeflats. They are almost certainly saturated, and we should use shorter integration times in the second run. We have not yet inspected twilight flats, or variation in the dome-flats over the course of the first run.
In the second run we should again take "1000 points of light" test images, on M67 or another cluster. A denser pattern than done before would make the analysis easier. We did 8x8; 10x10 or 12x12 would give a denser and higher S/N map of the sensitivity.