All this is done on hven (just like xspec; think of ciao as a big set of related tasks that operate on and transform files, like xspec - but while xspec starts up and then you're in a relatively enclosed environment, ciao tasks run directly from the Linux command line. This first command is executed only once per window; you can think of it as activating a particular version of the CIAO software: [cohen@hven zpup]$ ciao4.5 -o On the ciao thread for dmgroup, it looks like the grouping only works on pha1 files, so first you have to split the pha2 file: [cohen@hven zpup]$ dmtype2split "acis640_pha2.fits[tg_part=2,tg_m=-1]" "zpup_mm.fits[SPECTRUM]" [cohen@hven zpup]$ dmtype2split "acis640_pha2.fits[tg_part=2,tg_m=+1]" "zpup_mp.fits[SPECTRUM]" Now you'll be working with the new pha files (zpup_mm.fits, zpup_mp.fits") and you can try a grouping command that puts (at least) 20 counts in each bin: [cohen@hven zpup]$ dmgroup zpup_mm.fits zpup_mm_numcts20.pha grouptype=NUM_CTS xcolumn=channel ycolumn=counts binspec="" grouptypeval=20 [cohen@hven zpup]$ dmgroup zpup_mp.fits zpup_mp_numcts20.pha grouptype=NUM_CTS xcolumn=channel ycolumn=counts binspec="" grouptypeval=20 note that all these parameters can be viewed by typing: plist dmgroup (see the thread) *edit load_meg_group.xcm XSPEC12>plot XSPEC12>iplot PLT> wdata zpup_meg_group20cts.dat PLT> exit *now you've got one file with two spectra again; you can make two copies, one for each dataset, deleting all extraneous lines And you'll have your grouped spectrum file(s) Another kind of grouping you can try is to manually combine a particular consecutive run of bins together, with a command like: [cohen@hven kelley]$ dmgroup 9sgr_summed_mm_numcts20.pha 9sgr_summed_mm_numcts20_SiXIII.pha grouptype=BIN xcolumn=channel ycolumn=counts binspec="7029:7076:#1" grouptypeval="" (Note that this command was issued in a different directory and for a different dataset, but it shows you all the settings you should need. You can open up a text file of your spectrum in an editor that will show you what line number a given line is, and estimate the channel numbers ("channels" are just the original (ungrouped) bins), and adjust the command accordingly. Explore the thread linked from the webpage to see what else you can do.]