jamie@erdw.ethz.ch
                                        fax:1-632-1088
                                        phone:1-632-7804

Dear Perplexed User:

A list of the most recent changes to perplex follows.
If a bug is indicated for a program that you do not
use, then you need not concern yourself with the bug,
i.e., the programs are independent. NOTE: when bugs
are found, the PC (zip) and SUN (tar) binaries and fortran 
sources are updated immediately, Mac versions are only 
updated as indicated.
===========================================================
VERTEX/WERAMI/PSVDRAW/POLYGON BUGS FIXED, NOV 16, 1999

VERTEX  - Vertex's dimensioning has been increased. This 
          may cause memory problems on computers with limited
          RAM. 
WERAMI  - Werami now permits recovery of properties from a 
          phase diagram section along a non-linear profile.
          The profile must be described by a polynomial of 
          the form y = sum(c(i)*x^(i),i=0..n).
PSVDRAW - Psvdraw did not recognize pseudosection assemblages 
          consisting entirely of compounds, resulting in the
          "assemblage not found" errors.
POLYGON - Polygon crashed for certain special cases in which
          the stability field of an assemblage spanned two 
          corners of the diagram.
===========================================================
BUILD/VERTEX/POLYGON BUGS FIXED, OCT 18, 1999

BUILD   - Build was not writing the default plot file name if
          the user entered <cr> in response to the file name 
          prompt. The missing file name then caused Vertex to 
          crash.

VERTEX  - i) For constrained bulk composition calculations with
          fluid components, Vertex was not writing the output
          correctly to the bulk composition plot file used by 
          programs WERAMI and POLYGON, as a result the modes 
          and other numerical information were printed as "NaN"s
          or "Inf"s.
 
          ii) Vertex was not writing the bulk composition plot
          file for assemblages consisting entrirely of true compound
          phases, as a result these assemblages were listed as 
          "not found" by psvdraw, and the corresponding phase fields
          appeared as "holes" in pseudosections.

POLYGON - An incorrect type specification may have been causing
          various cryptic problems on SGI/SUN compilers. Several
          special computational cases have also been accounted for,
          which should reduce the number of pseudosection "holes".

Affected programs: VERTEX, BUILD, POLYGON
Affected file: vertex.f, polygon.f90, build.f, rlib.f, psvdraw.f, plib.f, polyf77.f

IBM and Tar Archives updated, Mac Archives have NOT been updated.
===========================================================
BUILD BUG FIXED, SEPT 14, 1999

A bug, recently introduced into BUILD, that caused the 
program to give incorrect prompts if potential saturated phase components
were chosen as mobile components has been eliminated. Default
file names have also been added to the program.

Affected programs: BUILD
Affected file: build.f

IBM and Tar Archives updated, Mac Archives have NOT been updated.
===========================================================
WERAMI and VERTEX BUGS FIXED, AUG 9, 1999:

* WERAMI - the bug caused werami to crash if a property
  profile was computed parallel to a psedosection axis.

* VERTEX - in fixed bulk composition calculations treating
  fluid components as thermodynamic components, vertex 
  output incorrect properties for the "solids only" component
  of the system phase assemblage. In some cases this bug may
  have caused vertex to generate a corrupt plot file that in
  turn caused psvdraw to crash (divide by zero).

Affected programs: WERAMI, VERTEX.
Affected file: werami.f, vertex.f, rlib.f

IBM and Tar Archives updated, Mac Archives have NOT been updated.
===========================================================
MAJOR REVISION, May 25, 1999. 

* The new version of Perplex faciltates the computation and
  analysis of pseudosections. The new capacity is demonstrated
  in the on-line tutorial at: 
  http://www.erdw.ethz.ch/~jamie/perplex_pseudosection.html

* Input and output files from previous versions of Perplex are
  incompatible with the current version.

* The program documentation (file vdoc.ps) has not been updated.

* For calculations other than pseudosections, Perplex is largely
  as before; however, some bugs have been removed from the code
  and the solution model files have been revised.

All archives have been updated, but the MacIntosh archives
do not yet contain the programs POLYGON and WERAMI or the revised
version frendly.
===========================================================
ARCHIVES RESTRUCTURED AND UPDATED, Jan 7, 1999. Archives 
for macintosh computers are now in zip format. Mac, PC, and
Sun archives updated
===========================================================
SOLUT.DAT UPDATE, Dec 7, 1998.

* The biotite "Bio" solution model in the solution model file
  "solut.dat" was for Fe-Mg-Al mixing on two octahedral sites 
  per formula unit and Fe-Mg mixing on the remaining site. This is 
  inconsistent with Holland and Powell's model (JMG, 1998), which 
  assumes Fe-Mg-Al mixing on one site and Fe-Mg on the two remaining 
  octahedral sites. Holland and Powell's formulation has been
  adopted in the current version of solut.dat, this leads to 
  a slight increase in the biotite stability field for typical
  pelitic bulk compositions. 

* Rough models for K-Na-Phengite ("KN-Phen", "MuCel", "PaCel") for
  have been added to "solut.dat". 
===========================================================
SOLUTION MODELS and modifications to the dimensioning and
formatting in BUILD, VERTEX, and PSVDRAW. Dec 1, 1998.

* The solution model file solut.dat has been updated. The
  new file contains a large number of models that are 
  consistent with the new Holland and Powell (JMG, 1998)
  data base. Some solution models formerly included in 
  solut.dat and consistent with earlier formulations of
  holland and powell's data base have been eliminated from
  solut.dat but are preserved in the file old_solut.dat.
  It is wise to read the comments in solut.dat
  before using a particular solution model. 

* Several new linearly dependent endmembers have been
  added to hp98ver.dat (see comments in the file)

* The dimensioning and formatting of PSVDRAW, VERTEX, and
  BUILD has been modified to permit calculations with up
  to 3000 equilibria, 30 solution phases, and 
  5000 pseudocompounds.
===========================================================
ZIP and TAR archives updated. Nov 24, 1998.
===========================================================
VERTEX/PSVDRAW bugs and modifications. Nov 24, 1998.

* Two bugs that caused psvdraw to plot some chemographic
  and mixed variable diagrams incorrectly have been removed 
  from vertex.
  The bugs caused plot file format errors, but not errors
  in the reported phase equilibria (as recorded in the print 
  file). 

* Psvdraw, the default plot format for chemographic diagrams
  has been improved (to test this try running vertex with the
  example input file in5.dat, an AFM diagram calculation). 

* Vertex/Vertexview, the present version of vertex outputs 
  a new plot file format for chemographic and mixed-variable
  diagrams that CANNOT be read by vertexview. Vertexview will
  be updated in the near future. 

* Old chemographic and mixed variable diagram plot files cannot 
  be plotted with the new version of Psvdraw.

* Pseudo-sections, if you compute a pseudo-section, i.e., a 
  phase diagram section with constrained bulk composition as 
  a function of two environmental variables (e.g., P, T, XCO2), then
  Vertex creates two plot files: a normal plot file that contains
  the phase field boundaries; plus a new plot file with the name
  of the normal file prefixed by a "b". This second file can be
  plotted with psvdraw to show the high variance assemblages as a
  function of the environmental variables. By modifying the default
  assemblage labelling in psvdraw (i.e., a drafting option), plots
  from this file can be made to show rock density, volatile contents, 
  or mineral modes.

Affected files: psvdraw.f, vertex.f, rlib.f, frendly.f, build.f,
                in5.dat, solut.dat 
Affected programs: psvdraw, vertex, vertexview
===========================================================
HP98VER.DAT, Nov 12, 1998. Some linearly dependent solution 
end-members added to permit use of solution models in 
solut.dat. Most of the solution models in solut.dat do not
presently include the non-ideal contributions suggested by
Holland and Powell (JMG, 1998). 
===========================================================
ZIP and TAR archives updated. Nov 6, 1998.
===========================================================
VERTEX/FRENDLY/PSVDRAW bugs, Oct 22-Nov 6, 1998.

During at least a portion of the time period Oct 22-Nov 6 incorrectly 
compiled versions of Vertex, Frendly, and Psvdraw were present
in the ZIP and TAR archive files. The compilation error may cause the 
programs to crash with run-time errors or, in the case of vertex,
to cause the program to terminate before completion of a calculation
without internal or system diagnostics.

The versions of VERTEX, FRENDLY, and PSVDRAW available as of Nov 6
eliminate this problem as well as several long-standing bugs.

Affected files: vertex.f, frendly.f, rlib.f, tlib.f, psvdraw.f, pslib.f
Affected programs: vertex, frendly, psvdraw.
===========================================================
ZIP and TAR archives updated. Oct 22, 1998.
===========================================================
PSVDRAW BUG, Oct 22, 1998.

The version of psvdraw created on Oct 18 used an incorrect
degree to radian conversion, as a result labels and ternary
composition diagrams were drawn incorrectly. The error has
been corrected.

Affected files: psvdraw.f, pslib.f
Affected programs: psvdraw.
===========================================================
PSVDRAW POSTSCRIPT in ADOBE and COREL, Oct 18, 1998.

Many users have complained that they sometimes have difficulties 
importing PSVDRAW POSTSCRIPT output into ADOBE or COREL graphics programs.
Typically in COREL the y-axis label is not drawn, while in ADOBE
only a small portion of the plot may be preserved. It appears that
the origin of this problem is that the "bounding box" defined on
the fourth line of the pro.ps file (used by PSVDRAW):

%%BoundingBox: 57 230 495 728 

does not normally include all the graphics plotted by PSVDRAW.
The bounding box is specified in units of points from the lower left corner
of the page, the four numbers are, respectively, the minimum x and y and
maximum x and y coordinates. The bounding box now specified in
"pro.ps" is:

%%BoundingBox: 0 0 572 791

which is the area of overlap between European A4 page format and
US Letter format. This definition should solve the problem for most 
plots generated by psvdraw. ADOBE and COREL users should recopy the
file "pro.ps" and place it in any directory where PSVDRAW is executed. 

For unusual plots where text or data is drawn beyond the limits of the 
bounding box ADOBE and COREL users should modify the bounding box accordingly. 

Affected files: pro.ps
===========================================================
PSVDRAW under LINUX, Oct 18, 1998.

GNU Fortran does not support the intrinsic fortran functions
COSD and SIND called from pslib.f and gives an error during
linking. The problem has been resolved by replacing the calls
with COS and SIN. The sources have been updated.

Affected files: psvdraw.f, pslib.f
Affected programs: psvdraw.
===========================================================
VERTEX BUGS, Oct 9, 1998.

Two problems in calculations for one component, or highly
degenerate systems, have been repaired. Tar and Zip archives
have been updated. The problems were:

1) The new version of vertex gave a run-time error for
   any calculations with only one thermodynamic component.

2) In highly degenerate systems, new and previous versions
   of vertex could fail to locate stable univariant 
   equilibria in Schreinemaker's diagrams if a univariant
   equilibrium intersected two edges of the diagram. Although
   errors of omission are still possible (usually indicated
   by warnings from vertex), this particular error is not. 

Affected files: vertex.f
Affected programs: vertex.
===========================================================
EXECUTABLE PROGRAMS IN IBMPC_NO_OPT.ZIP ARCHIVE, Sept 25, 1998.

Despite the name of this archive, the executable programs were
compiled with what Microsoft Fortran calls full optimization.
This optimization appears to cause an error in a routine (chkphi
in rlib.f) used by the new versions of build, frendly and vertex 
compiled after Sept 10, 1998. 

The executable programs have been recompiled with the 
optimization options /Op and /Ob2 and tested without error.
The IBMPC_NO_OPT.ZIP archive has been updated.
===========================================================
HP98VER.DAT COMMENT, Sept 10, 1998 - Several users have reported
problems with the HP98VER.DAT data base. These are the stability of
molten silica at low temperature, and the unrealistically large
stability field of laumontite is. These problems, which R. Powell 
is aware of, appear to be intrinsic to the data (Holland and 
Powell, JMG, 1998). A temporary solution to these problems 
is to exclude the phases in question from computations. 
===========================================================
PERPLEX '98 REVISION, Sept 10, 1998:

Tar and Zip archives updated. The Mac version of PERPLEX has 
not been updated.

CHANGES IN THE NEW VERSION OF PERPLEX INCLUDE:

1) The affected programs are:  VERTEX  
                               BUILD  
                               FRENDLY
                               PSVDRAW

   The affected files are:     vertex.f
                               build.f
                               frendly.f 
                               psvdraw.f 
                               actcor.f
                               rlib.f
                               flib.f
                               tlib.f

   The affected documents are: vdoc.ps
                               README.PERPLEX
                               makefile

   The file vlib.f is obsolete. Be sure to eliminate
   this file reference from compilation scripts. 

2) The earlier version of FRENDLY did not allow users to make
   linear combinations of phases described by nonlinear 
   volumetric equations of state (e.g., the Murghnahan EoS).
   This restriction is generally valid, but in the data bases
   published by Holland and Powell (JMG, 1998; hp96ver.dat
   and hp98ver.dat), the pressure derivative of the bulk 
   modulus is assumed to be equal and constant for all phases. With this
   assumption, it is permittable to make linear combinations.
   Accordingly, FRENDLY has been modified to allow this. 

3) The structure of all input and output files is unchanged.
   However, ealier versions of BUILD wrote comments after 
   the component names in the "computational option" file. 
   These comments will cause the new version of VERTEX to 
   crash. Either re-make the computational option files with
   the new BUILD program, or remove the comments by editing.

4) A simpler algorithm for the computation of phases as a 
   function of composition. The algorithm is slightly less 
   efficient than the previous algorithm, but has the advantage
   that it is easily generalized to any number of thermodynamic
   components. the maximum number of thermodynamic components 
   is specified by the parameter K5, currently set to 12.

5) The simplified algorithm eliminates about 3000 lines of 
   code, requires an order of magnitude less memory, and allows
   for a much simpler parameter structure. For most purposes
   the critical parameters are (current values in parenthesis):

   k1  (2000 ) maximum number of phases (or pseudocompounds)
   k2  (3000 ) maximum number of high variance (p>c) equilibria
   k10 (200  ) maximum number of stoichiometric compounds
   j9  (40000) maximum number of low variance (p<c+1) equilibria

   for most purposes the current value of j9 is excessive, to 
   save memory reduce this parameter and parameter k2.

6) The new algorithm is particularly useful for calculations 
   for a fixed or constrained bulk composition (computational
   option 2 in program BUILD). The new version of VERTEX makes
   the earlier PERPLEX program TERTEX obsolete. Several bugs
   that caused problems for fixed bulk composition calculations
   have been eliminated. 
   
7) For fixed or constrained bulk composition calculations VERTEX
   now outputs the density of the assemblage at the conditions of
   interest. In addition, if a fluid is present in the assemblage
   vertex computes the density and volatile composition of both
   the full assemblage and of the solid phase assemblage.

8) In contrast to the earlier version of VERTEX, the new version
   of VERTEX does not require that the phases of a system span 
   all possible compositions that can be defined by a positive
   linear combination of the components of the system. This
   change partially obviates the need for "fictive composants".
   It also has the consequence that a user may specify a bulk
   composition that cannot be spanned by the phases specified for
   a calculation. 

9) The programs have been tested on Sun (F77), SGI (F90), Mac (MPW)
   and PC (MDS) computers. The new version of VERTEX did not run 
   correctly when compiled on an SGI Origin 2000 with F77 due to 
   what appears to be incorrect memory management, VERTEX runs without
   errors on this machine if compiled with F90. 
===========================================================
UPDATE,  May 21, 1998. Tar and Zip archives updated. Mac
archives have not been updated.
===========================================================
CORK Bug, May 19, 1998. CORK BUG.

The CORK mixed volatile equation of state (Holland and Powell 
1991) was programmed incorrectly. The error effected 
water fugacities at T < 695 K. Additionally the PERPLEX 
version of CORK for CO2 used the approximate 
fugacities given from Eq 8 of the Holland and Powell paper.
To make PERPLEX consistent with the Thermocalc program and
data base, the current version of PERPLEX now uses the full 
MRK equations.

Affected files: flib.f
Affected programs: vertex, frendly, rk, cohsrk, species, etc.
===========================================================
UPDATE, April 9, 1998. Tar and Zip archives updated. Mac
archives have not been updated.
===========================================================
hp98ver.dat file created, April 9, 1998.

This file is the "research" version of the holland and powell 
data base available as of this date. The "tutorial" version
corresponds (at least approximately) to the hp96ver.dat file.
See comments within the data files and at Tim Hollands home
page for additional information and caveats concerning the
use of these data. Phase abbreviations are deciphered in the 
header text of the hp98ver data file. 

Affected files: hp98ver.dat
==================================================
Program documentation error, April 9, 1998.
Equations 2.4 and 2.5 are not the expressions used internally
by the PERPLEX programs. The equations are now reported correctly
in the program documentation.

Affected file: vdoc.ps
===========================================================
VERTEX/FRENDLY BUG, April 9, 1998.
A parameter for the expansivity function used for calculations
with the Murgnahan EoS was in error by a factor of 2 (parameter
b5 in the documentation Eq 2.5). Calculations made using
the hp96ver.dat data file were affected by this error, which
causes discrepancies on the order of 10 K in the calculated 
location of univariant reactions for pressures less than 10 kb.
The error has been corrected.

Affected files: rlib.f (routines conver and unver)
Affected programs: vertex.f, frendly.f
===========================================================
VERTEX BUG, Feb 10, 1998. The option to destabilize solution
endmembers in the new version of vertex (i.e., July 2, 1997)
does not work. If you attempt to use this option by modifying
the IEND flags of a solution in the solution model data file, 
the entire solution will be destabilized.
===========================================================
PSVDRAW BUG, Jan 27, 1998. The version of PSVDRAW created
on Dec 8, 1997 (see below) contains a bug that causes 
the axes for composition diagrams to be drawn in correctly.
The bug has been corrected.

Affected files: psvdraw.f
Affected programs: psvdraw

Tar and Zip files updated. 
===========================================================
UPDATE, Dec 8, 1997. Tar and Zip archives updated. Mac
archives have not been updated.
===========================================================
ACTCOR Bug, Dec 8, 1997. Actcor crashs if it encounters
comments in a thermodynamic data file (such as are present 
in hp96ver.dat). The bug has been corrected and the program
has been revised. The new version of the source must be 
linked to BOTH rlib.o and tlib.o (object libraries).

Affected files: actcor.f 
Affected programs: actcor
===========================================================
PSVDRAW/COREL/ADOBE Bug, Dec 8, 1997. Postscript generated by
PSVDRAW can cause a floating point error when imported into
Adobe or Corel. The error arises from the text graphics 
transformation in PSVDRAW that may produces small numbers
that Adobe and Corel cannot handle. With Corel the error
is recognized but causes the program to crash, with Adobe
the error is unrecognized but the editor produces an incomplete
picture. 

Affected files: pslib.f 
Affected programs: psvdraw, pspts
===========================================================
Program documentation typo, Nov 17, 1997. Equation 2.4 should
read V(T) = V0 + int(alpha,T) NOT V(T) = V0 * int(alpha,T) 
as it currently reads in file vdoc.ps.
===========================================================
PowerPC PERPLEX, Nov 5, 1997. The powerPC version of vertex
cannot read old versions of input files created by build
(some cryptic file formatting problem); however, If the 
input files are regenerated with build there is no problem
reading them with the powerPC vertex.
===========================================================
HP96VER.DAT, Nov 4, 1997. The fictive composant "fe2o3" has
the wrong composition (i.e, FeSi(2)O(7)), therefore vertex
will not accept it as a composant for fe2o3 (this will not
cause problems unless hematite (hem) is excluded from 
a calculation with the component fe2o3). The datafile has
been corrected. Archived versions have not been updated. 
===========================================================
UPDATE, Oct 28, 1997. PowerPC versions of vertex, build,
frendly, and psvdraw have been placed in an SEA archive. 
===========================================================
UPDATE, Oct 22, 1997. MAC, TAR and PC archives updated. 

MAC/SGI PERPLEX BUG, sloppy use of function variables caused 
vertex to crash on MAC and SGI machines, this does not appear 
to have effected performance on PC, DEC, and SUN computers. 

Affected files: rlib.f, vlib.f, vertex.f, frendly.f 
Affected programs: vertex, frendly
===========================================================
UPDATE, Oct 21, 1997. Tar and Zip archives updated. Mac
archives have not been updated (see Aug 26 notice below).
===========================================================
VERTEX BUG, Oct 21, 1997. A bug that sometimes caused 
vertex to "hang" during calculations has been removed. 
The affected file is vlib.f (routines wway and assip),
the affected program is vertex.
===========================================================
SUP93VER.DAT, Oct 19, 1997. The sup93ver.dat data file 
was converted incorrectly from the pre-July-97 format.

The compositional data for phases with components other than
Na2O-MGO-AL2O3-MGO-K2O-CAO-TIO2-MNO-FEO-O-H2O-CO2 is 
missing. The corrected data file is in:

http://www.erdw.ethz.ch/~jamie/perplex/datafiles

Tar, Mac, and PC archives have not been updated.
===========================================================
UPDATE, Sept 25, 1997. All binaries, archives and datafiles
are up-to-date, EXCEPT the MacIntosh archives (see Aug 26
notice below).
===========================================================
VERTEX BUG, Sept 25, 1997. Vertex fails on caclulations in 
which one or more solution phase endmembers lie entirely 
within the component saturation space (not very likely). 
The error has been corrected, the effected file is vlib.f
(subroutine uproj), the effected program is vertex. 
===========================================================
G97VER.DAT, Sept 22, 1997. The correct enthalpy of formation
from the elements for anthophyllite in Gottschalks data base 
is -12069.033 kJ/mol. The data file has not been corrected.
===========================================================
FRENDLY BUG, SEPT 5, 1997. If FRENDLY is used to write 
thermodynamic data to the thermodynamic data file it may
invert the order of the lambda transition and disorder
flags on the name card of the output data. If the relevant
phase or entity has a lambda transition the data becomes
unusable and may cause other programs to crash.

The only affected (effected?) file is frendly.f, the only affected
program is frendly.

IBM and Sun versions and archives of perplex have been updated.

Mac archives have not been updated pending resolution of the 
problem with running the applications on PowerPC's.
===========================================================
VERTEX BUG, AUGUST 26, 1997. Vertex will not calculate
phase diagrams correctly if CO2 is specified as a thermodynamic
component (rather than a fluid phase component) and H2O is
not. This error has been corrected in all files and archives
except the Mac SEA archives. The affected source file is 
rlib.f (routine gcpd), the affected binary is vertex.
===========================================================
VERTEX BUG, AUGUST 26, 1997. The application "vertex" in 
the mac 68000 sea does not run on Power PC's and may not
run on other Mac's as well. I have not identified the 
cause of the problem. If you have this problem the only
solution for the time being is to use the earlier version 
of perplex (in old_perplex) or use a different machine. 
===========================================================
G97VER.DAT, JULY 7, 1997.

The initial version of g97ver.dat contained two errors,
the errors have been corrected in the version presently
in:  

      ~jamie/perplex/datafiles (www)
      /usr/pub/perplex/datafiles (ftp)
      and the ibmpc zip archives.

The errors have not been corrected in the unix and mac
archive files (users of these archives should copy the
file from ~jamie/perplex/datafiles. The errors are/were:

1) The correct entry for antigorite is:

ant      1 0 0 0 antigorite       mg(48)si(34)o(147)h(62)    
  .00 48.00   .00 34.00   .00   .00   .00   .00   .00   .00 31.00   .00
-.6615342E+08  3596.317      174.9130      9011.990      .0000000    
-.1459624E+09  .0000000     -63981.00      .0000000      .1907022E+11
 .0000000      .0000000      .3200000E-04  .0000000      .0000000    
 .0000000      .0000000     -.2060000E-05

2) The correct entry for the fictive composant "mgo" is:

mgo      1 0 0 0                                                          
  .00  1.00   .00   .00   .00   .00   .00   .00   .00   .00   .00   .00
 .1000000D+03  .0000000D+00  .1000000D-04  .0000000D+00  .0000000D+00
 .0000000D+00  .0000000D+00  .0000000D+00  .0000000D+00  .0000000D+00
 .0000000D+00  .0000000D+00  .0000000D+00  .0000000D+00  .0000000D+00
 .0000000D+00  .0000000D+00  .0000000D+00

3) Gottschalk has not assigned an enthalpy for anthophyllite, the
   entry should be commented out (it will never be stable anyway).
===========================================================
PERPLEX, JULY 2, 1997. 

NEW VERSION OF PERPLEX, ALL PROGRAMS THAT READ OR 
WRITE TO A THERMODYNAMIC DATA FILE ARE AFFECTED.

THE NEW THERMODYNAMIC DATA FILES ARE NOT COMPATIBLE
WITH EARLIER VERSIONS OF PERPLEX.

IF YOU WISH TO CONTINUE USING THE PREVIOUS VERSION OF 
PERPLEX, THE RELEVANT FILES (I.E. FILES THAT HAVE 
BEEN CHANGED IN THE NEW VERSION) ARE STORED IN THE
DIRECTORY OLD_PERPLEX.

NONE OF THE DOCUMENTATION OR EXAMPLES ARE AFFECTED WITH 
THE EXCEPTION OF THE "PROGRAM DOCUMENTATION" (FILE VDOC.PS);
AND NONE OF THE GRAPHICS FILES ARE AFFECTED. 

The new version of PERPLEX supports the MURGNAHAN
equation of state (as used for the more recent versions
of Holland and Powell's data) and the equation of state used
by Gottschalk (EJM, 1997). This change required a number 
of modifications in the way in which PERPLEX programs
managed thermodynamic data. Additionally, because 
there are numerous parameterizations of the MURGNAHAN
eos, the number of parameters read for each phase in
the thermodynamic data file has been increased from 
16 to 18. Refer to the revised program documentation 
file (VDOC.PS) for a description of these parameters.

Several bugs have been removed from Vertex and Frendly,
but it is possible that new bugs have been added, so 
if you decide to use the new version be cautious.

The affected programs are:

           ACTCOR
           BUILD
           CTRANSF
           FRENDLY
           VERTEX 
           TERTEX

Affected files are:

actcor.f     rlib.f         b92ver.dat
build.f      vlib.f         bb84ver.dat
ctransf.f    tlib.f         sup93ver.dat
vertex.f     hp90ver.dat    ba96ver.dat
frendly.f    hp93ver.dat    g97ver.dat
abartex.f    hp94ver.dat    vdoc.ps
tertex.f     hp96ver.dat

TERTEX has not been revised. I will update this source and
other less frequently used programs as time allows. 

Three new data files are included with perplex:

G97VER.DAT - Matthias Gottschalk's (EJM '97) data base.

HP96VER.DAT - Holland and Powell's data base based on
the MURGHNAN equation of state.

BA96VER.DAT - Berman and Aranovich's TWEEQU data base.

WARNINGS!!!

I have made no effort to maintain consistency between
the phase names used in the recent data files and those used
earlier. This means that, for example, although appropriate 
solution models may be present in solut.dat, perplex programs 
may not recognize them as such. To correct such problems modify 
either the endmember names in the solution model file solut.dat
or in the thermodynamic data files. 

Do not include beta quartz in calculations with HP96VER.DAT. 
In the new HP data bases with HP'S landau
formulation the phase q has the properties of the stable
quartz polymorph at all P-T conditions. If you include bq together
with q in a calculation you will obtain a bq field with 
boundaries determined by numerical slop. 

In the old version of PERPLEX it was assumed that the 
reference state properties of phases that involved 
"landau" disordering incorporated the effect of the 
landau transition. It is now assumed that the landau
effect is NOT incorporated in the reference data. This
means that the new version of PERPLEX may compute 
slightly different properties for phases with landau 
ordering. Furthermore, if you use FRENDLY to compute the
reference state properties of a phase with landau ordering
you will obtain the reference state property value plus
the landau effect (formerly you obtained the reference
data exactly). 
===========================================================
VERTEX May 6, 1997. Updated vertex.exe and added the Berman
and Aranovich '96 data base (ba96ver.dat) to the IBM PC archive:

ibmpc_no_opt.zip

Be cautious using the Berman and Aranovich data base. 
===========================================================
VERTEX, FRENDLY, BUILD, ETC. Feb 27, 1997. Mac
applications updated in the following archives, the 
updated versions are worth copying:

PeRpLeX.68000.sit.Hqx
PeRpLeX.68030.sit.Hqx

PowerPC versions have not been updated.
===========================================================
VERTEX BUG. Jan 9, 1997. Vertex may crash giving an 
"undefined error" message in calculations in which
one or more compounds have >5 thermodynamic components
(not very likely). To correct the error recopy file 
vertex.f and recompile, or in your copy of vertex.f, 
replace the following line in subroutine simpl6 (line 3495):

                  call dgdegn (zpot,id6(j,l))

with:
                  call dgdegn (zpot,id6(l,j))
===========================================================
VERTEX/TERTEX BUG. Dec 6, 1996. VERTEX/TERTEX Bug for
fixed bulk composition calculations, this bug resulted
in incomplete diagrams under certain conditions (degeneracy).
The only effected file is vlib.f (subroutines bulkun and sfol2).
===========================================================
Unix tar_gzip archives updated. Nov 26, 1996.
===========================================================
IBM Zip archives updated. Nov 26, 1996.
===========================================================
VERTEX/TERTEX BUG. Nov 26, 1996. VERTEX/TERTEX Bug,
the correction for these programs made on Sept 12
introduced new problems that causes the programs to
fail to compute complete diagrams. I believe the
code should now work properly, but I haven't tested
it on a wide range of problems. The only effected 
file is vlib.f and the only effected routine is wway.

If you do not want to recopy PERPLEX you can (usually)
avoid the bug by reducing the default variable increments
in the computational option file created by build.

Mac/IBM/tar archive files have not been updated.
===========================================================
PC-Windows 95/NT PERPLEX programs are now available
(from me) for windows 95/NT systems. Nov 18, 1996.
In directory /home/jamie/www/perplex/ibmpc.
===========================================================
TWQ DATA BASE ADDED. Nov 18, 1996. A new TWQ data base
revision by Berman and Aranovich has been formatted for
PERPLEX in file ba96ver.dat.

Mac/IBM/tar archive files have not been updated.
===========================================================
TWQ DATA BASE ERROR. Nov 18, 1996. The species H2O in file
b92ver.dat was for the liquid reference state, not, as
expected by the PERPLEX programs, for an ideal gas reference
state. The error has probably been in the file for at least
one year, and possibly since late 1994.

Mac/IBM/tar archive files have not been updated.
===========================================================
VERTEX, FRENDLY, BUILD, COHSRK. Sept 20, 1996. Mac
applications updated in the following archives:

PeRpLeX.68000.sit.Hqx
PeRpLeX.68030.sit.Hqx

PowerPC versions have not been updated.
===========================================================
BUILD, FRENDLY, VERTEX and any programs that use, or identify,
fluid equations of state. Sept 18, 1996. Some high-pressure
fluid equations of state have been added.
affected files: flib.f 
IBM/Mac versions have not been updated.
===========================================================
BUILD UPDATED. Sept 17, 1996. Bulk composition options
improved. affected files: build.f
IBM/Mac versions have not been updated.
===========================================================
VERTEX/TERTEX BUG. Sept 17, 1996. The "correction"
for the Mar 24, 1996 bug listed below, introduces a new
problem, namely, that the chemical potential of a saturated
phase component may have a finite value when the amount
of the component is zero (e.g., the chemical potential of
CO2 in an H2O-CO2 fluid is not zero when X(CO2) = 0, thus
vertex may allow carbonates to be stable with a pure water
fluid). You can avoid this bug by setting the limits for
X(CO2) to values slightly different than 0 and 1, e.g.,
0.000001 and 0.999999. Alternatively recopy the following
files and recompile tertex and vertex: 
vertex.f, tertex.f, vlib.f 
IBM/Mac versions have not been updated.
===========================================================
VERTEX, TERTEX, Sept 12, 1996. VERTEX/TERTEX frequently
issue warnings (**warning ver045**) that more than one
equilibrium occurs withing the minimum search increment,
this problem may result in incomplete phase diagrams, 
particularly in mixed-variable type phase diagram calculations.
In old versions of these programs the problem can be 
corrected by reducing the default variable increments 
(assigned in the computational option file, normally by BUILD). 
The problem has now also been corrected in the fortran code.
The effected code is that of subroutine SEARCH in the file
vlib.f. IBM/Mac versions have not been updated.
===========================================================
SPECIES. July 5, 1996. Species has been made a little
more user freindly, the only affected source file is 
species.f. Mac versions have not been updated.
===========================================================
CORK BUG. June 28, 1996. Perplex programs running with
Holland & Powell's CORK equation of state (CMP 109:265)
crash at high-T. This occurs because at high-T CORK can
yield three roots (strange but true), and Perplex 
automatically chooses the lowest density, which in this
case turns out to be negative. 

To correct this bug replace the following lines in file
flib.f (subroutine crkh2o), this correction is sort of
ad hoc, since I don't know if the false roots are always
negative, but it's a good bet, otherwise CORK has a serious
problem:

         if (p.lt.psat) then
            vol = xmax
         else 
            vol = xmin
         end if

      end if 

      cc = a/b/rt/t12


with: 

         if (p.lt.psat) then
            vol = xmax
         else if (t.lt.700.) then
            vol = xmin
         else 
            do 1 i = 1, 3
               if (x(i).gt.0.d0) then
                  vol = x(i)
                  goto 2
               end if
1           continue
         end if

      end if 

2     cc = a/b/rt/t12
===========================================================
Mac PowerPC Applications. June 26, 1996. Thanks to
Lukas Baumgartner, there are now PowerPC applications
available in the macapps directory.
===========================================================
ANONYMOUS FTP. June 21, 1996. We (at last) have 
anonymous ftp, to copy perplex using this utility
log on to eurasia.ethz.ch and change directory to 
/pub/perplex.
===========================================================
DATAFILE BUG. May 29, 1996. Some data files contain
fictive entries (composants, e.g., k2o, al2o3). To
prevent these fictive composants from destabilizing
real phase relations I have generally assigned them
large positive standard state Gibbs enrgies (1d12). 
However these values are so large that round-off may 
cause a computational error that vertex reports as 
error 917. To correct this error reduce the standard 
state Gibbs energies of fictive composants (values 6 
orders of magnitude above those of real phases should 
be a safe bet).
===========================================================
VERTEX/TERTEX/SWASH BUG. May 22, 1996. These programs
may not compute the properties of solutions involving
endmembers which have lambda-type transitions or
temperature dependent disorder as treated by Holland
& Powell. I have yet to verify this, but if this is
true, the effect is to destabilize intermediate compositions
of the solution in question. A simple, temporary, solution
to this problem is to avoid the use of such solutions, e.g., 
use a high feldspar model (Kf(h)) instead of one that involves
the endmembers with variable order (Kf), the difference is
in any case minor.
===========================================================
LIAM BORN. Mar 29, 1996.
===========================================================
TERTEX BUG. MAR 28, 1996. When the user opts to specify
the initial assemblage, if this assemblage is metastable
or compositionally degenerate TERTEX does not inform the
user, but rather prints a message that no stable equilibria
have been identified. To correct this error copy files 
tertex.f and tlib.f and recompile tertex. (The new version
of TERTEX also prints better progress diagnostics).
===========================================================
VERTEX/TERTEX/FRENDLY BUG. MAR 24, 1996. This bug occurs
when a saturated phase is specified and one of the possible
components of the saturated phase is specified as a mobile
component, e.g., when a water fluid is specified as being
saturated and the chemical potential of CO2 is taken as
mobile. The bug may cause the chemical potential of
the mobile component to be set to an arbitrarily low,
constant, value and leads to an obviously incorrect diagram.

To correct this bug replace the following line in files
tertex.f and vertex.f (subroutine ufluid):

       uf(i) = -9.9d09

with:

       uf(i) = 0.d0
===========================================================
POSTSCRIPT CONVERTER PROGRAM FOR UNIX. MAR 19, 1996. I had
sworn to stop reporting these things, but here is one last
converter, AIMAKER. AIMAKER converts generic postscript, 
such as that generated by PSVDRAW, to something which can
be edited ADOBE or CorelDraw editors. AIMAKER is nice 
because, it is small, self-contained, general, and free.
The installation script and instructions are in file
aimaker.shar in my graphics directory.

Another free graphics editing tool for UNIX users is
tgif, in combination with the conversion script pstoedit. 

For other graphics possibilities read files
PERPLEX.UPDATE.OLD and README.PERPLEX.
===========================================================
SWASH BUG. MAR 19, 1996. This bug may cause SWASH to output
parameters for a "Lambda" transition, when none is appropriate.
The resulting thermodynamic data file is then inconsistent
and cannot be read by PERPLEX programs. To correct 
this bug, after the lines in file swash.f (subroutine
soldme):

      if (ilam.ne.0) then
c                               determine number of transitions from

to:

      if (ltyp(iphct).ne.0) then
c                               determine number of transitions from
===========================================================
COHSRK BUG. MAR 13, 1996. For calculations with the 
H2O-CO2 equations of state, COHSRK prints the volume
computed with the MRK equation of state. The correct
volume can be computed with FRENDLY or RK. To correct 
this bug, after the lines in file cohsrk.f:

c                                  output results:
      igo = 1

add the lines:

      xfh = fh2o
      xfc = fco2

and replace the lines:

c                                 as written many of the
c                                 binary H2O-CO2 routines do
c                                 not calculate volume. The
c                                 volume of the fluid from these
c                                 routines can be calculated by
c                                 evaluating (dg/dp) by finite 
c                                 differences, this is done in
c                                 the PERPLEX program RK, but I 
c                                 was to lazy to do it here.
         if (ifug.lt.2) write (*,1300) vol

with:

c                                  increment pressure for
c                                  finite difference estimate of
c                                  volume:
         p = p + 1.
         call cfluid (fo2, fs2)
         write (*,1300) 83.14*t*((1.-xo)*(fh2o-xfh)+xo*(fco2-xfc))
===========================================================

SEE PERPLEX.UPDATE.OLD FOR AN UPDATE FILE 
CONTAINING NOTICES FROM 8/94-4/95.