The value of the enthalpy of ordering is a point frequent point of confusion in formatting 
THERMOCALC models for Perple_X. In papers by Holland & Powell the enthalpy of ordering for
a reaction of the form 

1/2 (A + B) = C               [1]

is

DH = H_C - 1/2(H_A + H_B)     [2]

where H_i is the enthalpy of endmember i. Because Holland & Powell invariably choose 
the endmembers A, B, and C in such a way that the configurational entropy change of 
the reaction is zero, the Gibbs energy of ordering for [1] is identical to the enthalpy 
of ordering. Thus [2] can be written equivalently 

DH = DG = G_C - 1/2(G_A + G_B)     [3]

or

G_C = 1/2(G_A + G_B) + DH          [4]

where G_i is the effective Gibbs energy of endmember i. The need for distinguishing the 
"effective" Gibbs energy is that Holland & Powell often introduce "DQF corrections" to 
Gibbs energies indicated in their thermodynamic data bases such that

G_i = G0_i + DQF_i                 [5]

This is all fine and dandy, but the "make definitions" in THERMOCALC data files used to 
define the Gibbs energy of the ordered species C are written in terms of the endmember 
Gibbs energies without DQF corrections, i.e., 

G_C = 1/2(G0_A + G0_B) + DQF_C     [6]

Because equations [4] and [6] look similar, Perple_X users frequently mistake the entry 
for DQF_C in THERMOCALC solution model files with the enthalpy of ordering term DH 
required in Perple_X for the definition of an ordered species. Substituting [5] into [4] 
gives

G_C =  1/2(G0_A + G0_B) + 1/2(DQF_A + DQF_B) + DH   [7]

and comparison of [6] and [7] gives

DH = DQF_C - 1/2(DQF_A + DQF_B)                     [8]

the value to be entered in Perple_X.