I have compared the fluxes for Argon, Neon, Krypton and continuum lamps with the predictions (provided by Jim and Erin). Those predictions are attached to this ticket. The analysis has been only done in red.

I have looked at the fluxes for 3 brightest spots for Argon and Krypton, and 5 for Neon. I have also looked at 3 different position for continuum lamp. All of the analysis has been done on raw images, on the fiber at x=2545. This is one of the brighter fibers that I have see by visual inspections of the images.

Argon.png shows the predicted and measured flux for the spots. 3 measured spots are denoted by orange points, and the wavelength has been indicated by the gray line. For lines at 763.50532027, 811.5521071 , 912.32685895 nm, the measured ratio between the predicted and measured values is 20.71333146, 23.75641397, 18.92842034, for mean ratio of 21.

For Neon the ratio has been measured for spots at  640.22319559, 650.61847332, 667.85471749, 692.9226948 , 703.23178381 nm. The observed ratio is 9.85922264, 8.53241257, 4.13189759, 6.77602784, 4.98336609, for the mean ratio of 6.9.

For Krypton the ratio has been measured for spots at 760.13753536, 811.27260789, 877.66812554 nm. The observed ratio is 11.74454834, 16.88226871, 14.40388648, for the mean rati of 14.3.

For continuum the ratio has been measured at 697, 795 and 920 nm. The observed ratio is 55.6, 71.98, 73.34, for the mean ratio of 66.97.

This ratio is roughly 3 times higher than previously claimed 20, as Erin explain that the value provided in continuum_red.csv and shown in continuum_counts.pdf is ``this is integrated over the wavelength range of a single PFS pixel in the dispersion direction, then divided by 3.22 to try to account for the spread in the spatial direction''. As such these values, which are in the range of between 6000 to 8000 counts for large portions of red dectector as per one physical pixel, and not integrated over the spatial direction, as I assumed previously. 

For HgCd the ratio has been measured for spots at 508.59404793, 435.80522598, 546.10778179, 576.99145691, 579.03170651 nm. The observed ratio is 10.72966319, 3.4212724 , 6.19183228, 5.02193196, 4.82289145, for the mean ratio of 6.

The ring lamp is around 30 to 60 times brighter, depending on which part of the spectrum one is observing. See continuum and ring lamp simple spectrum.png. I also want to note that the nonuniform flux behavior between neighboring fibers, where flux is increasing from one side to the other within a block is very similar in the exposures taken with ring and/or continuum lamp. 

Update September 23:Update September 23:
1. Figure rotation_0.png shows: Left panel: The flux in the exposure with rotator at 0. The scale is from max=128k counts to 64k counts, i.e., yellow are the brightest fibers and bluest fibers are at the half flux. The gray dots show the deduced center of the illumination (x=-10,y=-4), which has been deduced using data shown in the center panel. Center panel: the flux in the fibers as a function of the distance from the center. Center has been fitted in order to minimize the scatter around the 2nd degree fit. Right panel: The residual after fit to the distance from the center. I do not see any strong residual dependence.
2. dif_from_mean_rotator.png shows the difference from the mean value at each rotator position. ``Mean value'' has been constructed as an average of all 4 flux measurements. We can observe the rotation. The scale is from -6k to +6k. The max value in the images is as above, roughly from 128k to 64k.
3. comparison_rot0_and_ring.png shows: Left panel:  The flux in the exposure with rotator at 0. Same panel as in Figure rotation_0.png. Center panel: The flux in the exposure taken with ring lamp, prepared in the same way. The center, gray dot, is still the same center as in the left panel. By accident, the fluxes in both exposures are very similar. Right panel: The difference between two illuminations. 

Further discussion with Jim and the whole group will continue. Comparison with the fluxes from the sky observation to be placed in a new ticket.