Hoi,

I tested  the result of the PRL  analysis wrt to variations  of the SF
parameters. I use the numbers from CLEO's CBX 

  http://www.slac.stanford.edu/~ursl/talks/061603/cbs0159.ps.gz

in table 1: 

-------------------------------------------------------------------
Run              deltamb       deltaa     Result (stat error only)
-------------------------------------------------------------------
sf-e-00.dat      -0.065        -0.32      0.0231331 +- 0.00280705
sf-e-01.dat	 -0.301	       +0.67	  0.0380874 +- 0.00443508
sf-e-02.dat	 +0.005	       -2.92	  0.0181131 +- 0.00222707
sf-e-03.dat	 +0.055	       -2.23	  0.0159838 +- 0.00193727
sf-e-04.dat	 +0.045	       +1.26	  0.0207438 +- 0.00254212
sf-e-05.dat	 -0.015	       +0.14	  0.0222166 +- 0.00270575
sf-e-06.dat	 -0.015	       -2.30	  0.0158894 +- 0.00192506
sf-e-07.dat	 -0.065	       +0.39	  0.0246229 +- 0.00298269
sf-e-08.dat	 -0.065	       -1.14	  0.020265  +- 0.00246847
sf-e-09.dat	 -0.110	       +0.61	  0.0269626 +- 0.00324463
sf-e-10.dat	 -0.110	       -0.76	  0.0229559 +- 0.00277994
sf-e-11.dat	 -0.220	       +0.74	  0.0334033 +- 0.00394548
sf-e-12.dat	 -0.220	       +0.16	  0.0310471 +- 0.00368471
sf-e-13.dat	 -0.270	       +0.73	  0.0364401 +- 0.00426178
sf-e-14.dat	 -0.270	       +0.51	  0.0354744 +- 0.00415781
sf-e-15.dat	 -0.320	       +0.69	  0.0393289 +- 0.00456418
sf-e-xx.dat	 +0.0  	       +0.0 	  0.0213044 +- 0.00259884
------------------------------------------------------------------

where the  rows correspond to  the rows in  CLEO's CBX. I run  the fit
with the non-resonant model only.

A display of this is given in 

  http://www.slac.stanford.edu/~ursl/talks/061603/brbr.eps.gz
  http://www.slac.stanford.edu/~ursl/talks/061603/lbarl1.eps.gz

where the first  plot shows the distribution of BRBR  (top) and at the
bottom "default  - BRBR" (where  default is sf-e-xx,  with nonresonant
only). The  second plot shows  lambda1 vs. LambdaBar, in  black filled
circles  the CLEO points  (their central  value is  shown in  the open
circle).  In  red the  central values  we use and  the error  we quote
(\Delta\chi^2=1 ellipse).

The  differences to the  "default" (non-resonant  only) BRBR  seem too
large to represent the real error (maybe this is wishful thinking).

If I took  the numbers in CLEO's CBX at face  value, I'd conclude that
the shape function  parameters are known with much  less accuracy than
the HQET parameters.   I think we need to  understand very clearly how
much connection there is between these two interpretations, as CLEO is
probably going to complain, whereas  Zoltan and Thomas had a different
opinion.



Cheers,
--U.