An extract from Thomas' e-mail ...

For each WG+ribbon 5 plots (less if data are not taken)
are shown on 2 pages:
-page 1: 
 --plot 1: average p.e. as a function of fiber number, LED on the first
           neighbour is flashed (e.g. LED H3A for waveguide H2A). 
           A linear fit to the av. p.e. is overlaid whereas the first
           and last 50 fibers are excluded in the fit (to be independent
           of edge effects).
           The edge of the LED panel is at high fiber numbers (here:
           decrease of signal likely)
 --plot 2: the same for the second neighbour
           Ribbon is in middle of LED panel.
 --plot 3: the same for the third neighbour
           The edge of the LED panel is at low fiber numbers (here:
           decrease of signal likely)

-page 2:
 --plot 4: In order to suppress effects due to the inhomogeneous light
           output from the LED panel, the average difference from the
           linear fit to the p.e. yield is shown as a function of fiber
           number. Here the data from any 50 fibers which are next to
           the edge of the LED panel which is flashed are excluded from
           the average. Ideally this distribution should therefore
           corresponds to the "real" transmission of the waveguide fibers.
 --plot 5: as 4 but blown up y-scale

---
.ps files are under d0mino:~nunne/cft_onl/ledfits/hbook/wg_qc