Both  -CO-NH- and CO-N< are carboxamides so the PRX would be 2 -1 = 1 since the correction for carboxamide and sulfonamide is -1. The PRX for the -CO-OH is 2. It seems to me that the numbers are reversed in the definition, it should be N/O; X-Y: 1.0, X-A-Y: 2.0 (X, Y: N/O, A: C, S, or P) with a correction (-1) for carboxamide/sulfonamide. 

Mr. Leberecht,

The article is based on the authors' model, which should be regarded as a "semi-quantitative" one, according to the text in the abstract section.

Contrary to the statement above, however, a detail recognition with the main text shows that the correlation model between "molecular structure"-logP is not consistent with the real electronic effects in the molecules and the bond orders.

What implication does the model have, in fact?

The authors have used the rule to 'maximum 4-valence of the 'C' and integrity bond order, based on 4n and 4n+2 rules for aliphatic and aromatic compounds'. The latter rules determine the textbooks rules about the level of saturation in organics. These rules lead to "structures" (in fact, they are 'chemical diagrams', not chemical structures) of molecules employing integrity single, double and/or triple bonds. Such as schematic presentation of a molecular "structure" has only illustrative character about the electronic structure and reflects the molecular skeleton.

The basic point to be noted the sentence above is, that talking about 'molecular structure', we should make major distinction between 'molecular structure', meaning skeleton and 'electronic structure'. For example, pyridine has heterocyclic ...C-C-N-C... skeleton. The order of a chemical bond (n) depends on the 'electronic structure', however. The bond orders are 0 < n < 1, meaning that there is unable to talk about C-C and C=N bonds. Because of, rules '4n' and '4n+2' (so-called Hueckel's rules) are based on the Hueckel's approximation. It does not consider explicitly the interactions between the electrons. But, due to different electronic effects, we have not integrity bond order.

In this context, without experimental or accurate theoretical quantum chemical information about the geometry parameters and electronic distribution there is unable to determine the 'bond order' (n). There are many molecular electronic effects, which can be observed within the framework of one and the same molecular skeleton. In addition to the fact that these electronic effects can be different within the framework of different 3D conformations of one and the same molecular structure.

Information about the bond order can be obtained by experimental high resolution crystallographic and/or computational quantum chemical methods.

The shown parametrization in the paper, cannot express the nature of inter-atomic forces leading to bonds in the molecules. Therefore there is unable to approximate the bonds as integrity numbers like double, triple or higher order bonds.

Furthermore, depending on molecular complexity, there are more then one molecular conformations with close energetics. So that, the summarized in the paper logPs may correspond to more then one molecular conformations/electronic structures. The inaccuracy of reported data as standard deviations is high (Data in Figs. 1 and 2). Or the shown correlation molecular structure-logP may be wrong. Furthermore, this possibility is ruled out, due to lack of detail information about the electronic structure (respectively bond order) of these molecules.

Furthermore, in the present state of the chemistry, to many elements in the periodic table the electronic configurationa are still unknown. To these elements, thereby, there is unable to apply the discussed simple rule (above), when these elements take part in complex biologically active metal-organics, for example.

Simply speaking, that model (in the paper under discussion) is inapplicable to inorganics, metal-organics and/or even simple organics like substituted pyridines or any other simple conjugated organics.

The inaccuracy of the authors' concept about the bond order (meaning the electronic structure) influences the whole parametrization model in a decisive way.

On the other hand, 'logP' accounts for distribution of a solute 's' between two phases. In this case, these are water and an aliphatic alcohol.

But, we have an equation:

psi/psj = exp[-beta(Dmu*sj-Dmu*si)],

where beta = (kT)-1; and

Dmu*s(i or j) - solvation free Gibbs energies of solute s in phases i and j.

The free Gibbs energy is among most sensible quantity, accounting for molecular and electronic structures and intra-molecular forces/factors as well as environmental factors and parameters (ionic strengths, T, P, viscosity, polarity and more).

This means that depending on complexity of your molecule, including electronic effects and corresponding environmental parameters you can obtain precise values of solvation free Gibbs energies in solvents i or j. Thus you can determine logP, in 'quantitative' terms.

There is ample opportunity for computations depending on complexity of an object as molecular and electronic structures as well as complexity of intermolecular interactions. Different theories yielding to solvation free Gibbs energies, account for ionic solutes; flexibility of the molecular skeleton; special electronic effects in inorganics inorganic solutes and metal-organic solutes; and many more. They provide good insight into forces determining different electronic structure, respectively, energetics of corresponding ensemble of interacting solute-solvent species.

You could pay attention to the following major topics in chemistry:

- 'Solvation thermodynamics'

- 'Quantum chemistry'.