15.37.10  GAUSSIAN    

Does a fit to:
     Y=a0*EXP(-0.5*((X-a1)/a2)**2)[+a3+a4*X+...]      for     NONLINEAR=ON
(Default) or
     LN(Y)=LN(a0)+-0.5*((X-a1)/a2)**2 for NONLINEAR=OFF
     for data points with Y,DY not 0.0.

Since  only  data  with  both  Y,DY  nonzero  are  fit,  there may be some
inaccuracy for histogram data with zero in a large  number  of  bins,  and
small  numbers in a few bins.  Such data should be smoothed or rebinned to
remove the bins with DY=0.

For  linear  fits  if  you  specify the a0 or a1 coefficient you must also
specify a3.  If you wish to fit data  which  has  negative  or  zero  data
points, you must either SMOOTH the data to get rid of them, or use the BIN
command to rebin the data into large bins.  If all y  data  is  less  than
zero, you may make it positive with the command:
     TD:Y=Y * -1

When  a NONLINEAR fit is used, and NTERMS=n is greater than 3 a polynomial
background of n-3 terms is added  to  the  gaussian.   If  NTERMS=4,5,6  a
constant, straignht line, or parabola are used for the background.  Larger
values are possible, but not recommended. 
For  LINEAR fits the OFFSET is used as the initial guess for the center of
the gaussian.  When done, it is approximately equal to a1.

WARNING If you get the warning message that negative data points have beem
omitted in a Gaussian fit, the fit may be incorrect.