Some details of how the plot was made (IMH)

   To get the date from each high-z spectrum I compared it to the time series
   of SN92A. First I normalised all the template spectra to a 2000A
   region at the middle of the high-z spectrum. For each template 
   I then just calculated a chi-sq using the difference between the spectra
   over the region of overlap. I repeated that for templates at different
   dates and found the minimum chi-sq with by fitting a curve to the
   lowest points.
   I did the same for nearby templates as a check (shown as stars on the plot).
   N.B. Some of the hi-z spectra (the ones with boxes round them) 
   have had some galaxy template subtracted. This could affect the 
   correlation since the amount of galaxy to subtract
   was chose arbitarily to make the shape of the spectrum look right!

   Also, the program has problems for dates earlier than about -5 since
   that is the earliest date of template that I have used.