[Enthalpy]

Hello you all!

Looking for molecules often absent from suppliers' catalog, NIST database and similar CRC Handbook of Chemistry and Physics, I generally lack a reasonable estimate of the melting point.

I tried Mpbpvp, included in EpiSuite available there
http://www.epa.gov/oppt/exposure/pubs/episuitedl.htm
and while boiling points are reasonable, melting points are very false, like +50°C predicted for -80°C measured.
As melting points depend a lot on tiny changes in the molecules, maybe the task is too difficult for a software.
I don't even want proteins: just up to 20 atoms of carbon or some of nitrogen, plus hydrogen. But ramified, polycyclic, strained.

So please propose a method! (Free, of course and as usual...)
- A working estimation software
- A huge list of everything...
- A working set of empirical rules? I have one for functionalized alkanes, but it's not intended for strained polycycles for instance.

Thanks!