Dear Professor Kurucz:

  I am confused about the calculation of WTMOLE in ATLAS9 and ATLAS12.
I was looking at the code and found for Atlas9:


      WTMOLE=0.
      DO 1514 IZ=1,99
 1514 WTMOLE=WTMOLE+XABUND(IZ)*ATMASS(IZ)

Similar coding is in Atlas12 on your website.  This equation would properly
compute the mean molecular weight if the sum of XABUND(IZ) is equal to 1,
and if molecules are not present in the atmosphere (but CO is an important 
molecule).  OK CO formation would change WTMOLE by 0.02% for solar composition.
The code error, even for 10 times solar composition, only amounts to 0.5% change
in WTMOLE.  A more powerful change would come from formation of the H2 molecule
in cool dwarfs.  I estimated the change for a 3500K [M/H]=0.00 star with logg 4.5:
the 17th level down from the top (of your 64 layer CD model) had a significant H2
fraction, which changed the WTMOLE from 1.27 to 1.61 (unless I made a mistake).

However, in the subroutine BLOCKE Atlas9 sets H and He number fractions:

C                   1H    2HE
      DATA ABUND/ 0.911,0.089,

followed by the log10 abundances of Z>2 elements.  

Since N(H)+N(He)=1.0 then the sum in the code, above, must be 
greater than 1.0 (I suppose about 1.02 for solar composition), and
the current code over-estimates WTMOLE.  

Appropriate replacement code might be:

      WTMOLE=0.
      WTSUM = 0.
      DO 1514 IZ=1,99
      WTMOLE=WTMOLE+XABUND(IZ)*ATMASS(IZ)
 1514 WTSUM = WTSUM + XABUND(IZ)
      WTMOLE = WTMOLE / WTSUM

Perhaps the exact WTMOLE value is not important; if so, then why bother
calculating the contribution from each element?  You could just stick to 
H, He and the solar composition.

It could be that I have made some dumb mistake, which is not unusual; if
so, then please could you point out my error.

Best regards,

ANdy McWilliam