Jan Balewski suggested that I try to determine the BEMC timing curves
from L2 data.  Here is my first attempt from the data Oleg Tsai took a few days
ago:

                                                                                              
TCD Phase Run Number
12 8326025
17 x26
22 x27
27 x28
32 x33
37 x34
42 x35
47 x36
52 x37
57 x38
62 x39
                                                                                              

I computed the Integral(100,2000) for each ADC and added all 160 PMTs
in each crate (none masked).  Although I did not subtract pedestals, I fit
each curve to a gaussian to get the centroid. 

The value of the TCD phase in 2007 was 36 ns.  For 2008 the means for East and
West separately are about 37.3 ns