The discrepancies between molecular modelling and experiment conceming the crystal structures, densities and relative stabilities of long n-alkanes is shown to be caused by thermal motion (chapter 2). Mimicking dynamic thermal motion by increasing the static Van der Waals radii of carbon and hydrogen solves the problems for all even n-alkane and most triacylglycerol crystal-structures. We call this adapted potential the Thermal Motion Force Field (TMFF).

Thermal motion adequately solves the problems connected with the molecular modelling of the crystal structures of n-alkanes and adding charges does not change the results ( chapter 3). Apparently, charges do not play a role in modelling the structures or the relative energies of the crystal structures of n-alkanes. Several methods for calculating 'the' point charges of n-alkanes are compared; only one yields physically realistic charges, at the expense of 20 additional charge centres for a molecule as simple as n-C20H42.

Having dealt with the chains of n-alkanes, what is left are the end groups. They are the groups that meet in between two layers of parallel alkyl-chains and as such determine the third dimension of every n-alkane or triacylglycerol crystal structure. lnteractions between two methyl groups are compared to interactions between either two chlorine atoms or two bromine atoms (chapter 4). The experimental data are scarce, the quantum-mechanical model oversimplified and the features sought for subtle. Definite conclusions can therefore not be drawn, but it seems justified to say that attempts to obtain crystals of TAG molecules in which all methyl groups have been replaced by either a chlorine or a bromine atom should yield isomorphous crystal structures with improved crystallographic properties.

n-Alkanes and triacylglycerols have so much in common, that the major part of n-alkane crystal structures is present in triacylglycerol crystal structures without modification. Filling out the remaining gaps with portions from other TAG crystal structures folIowed by energy minimisation lead to a consistent model for the beta'-2 crystal structure of 10.12.10 (chapter 5). A homologously isomorphous crystal structure of an isomorphous analogue was known for the beta'-2 crystal structure of 16.18.16. Complemented with the molecular conformation from another crystal structure, this provided a good starting model for Rietveld refinement ( chapter 6). Therefore, this thesis is the first to actually feature two polymorphs, rather than separate crystal structures, for a triacylglycerol family. Molecular modelling seems to agree on the experimentally observed relative stability of the two polymorphs. In the light of chapter 2, however, it is surprising that it should be the longest member of this series which is repeatedly reported to crystallise in beta polymorph, whereas this should be more common for the shorter members.

CSD searches show that the alkyl chains in the crystal structures of almost all long chain compounds are parallel. This renders the crystal structures of each series homologously isomorphous. Therefore, polymorph predictions of long-chain compounds can be carried out using a short member (chapter 7). This speeds up the computations and improves the sampling of the polymorphic phase-space. Furthermore, it is shown that the results obtained for the triclinic and monoclinic polymorphs of even n-alkanes are consistent with the remaining three polymorphs. The conformation of the ester fragment in TAGs is shown to be independent of the crystal structure, and a possible application of this knowledge to arrive at a better description of the electrostatics of a TAG molecule is described.