the structure of ~u, which is related to that of Ga I in the atomic coordination, will be reported in the context of these phases elsewhere. In the systems mentioned, metastable ~, phases can be retained in quantities up to 100 %, at the appropriate compositions, which permits a better structure analysis of ~ than in the case of ~, (Ga-AI), where a maximum of 25 % ~u was retained by quenching. The rather large metastable solubility of AI in Ga I (up to 8 at. % AI) is again due to the small size difference and identical valence electron concentration of Ga and A1, and is the highest observed solid solubility of any element in Ga I. The various models of graphitization which have been proposed during the past two decades, the assumptions on which they were based, and the related ordering parameters are considered. It is shown that there are apparent inconsistencies in the literature when attempts are made to relate the various ordering parameters calculated for identical materials. The postulated relationships between these ordering parameters and the experimentally determined average interlayer spacing are examined with the use of new data together with those available in the literature. A modification of the Franklin equation has been obtained which more nearly satisfies these experimental data, and it is concluded that none of the classical models is satisfactory over the whole range order in carbon materials. The recent ideas of Maire & M6ring are judged to give a better interpretation of the process of graphitization. Crystallite dimension data have been analysed as general functions of stacking order and the resulting relationships compared with those published in the literature. It is concluded that only general relationships can be indicated, while the more specific relationships must be qualified by comprehensive statements of the correction procedures used. Differences in the behaviour of cokes and artefacts have been found: in order to attain a given degree of stacking order, the constituent crystallites of the artefacts have to be larger than those of the cokes. It is shown that the shape of the crystallites varies with heattreatment temperature and hence with stacking order. Initially the crystallites are plate-like, becoming more cylindrical and columnar with heat-treatment and finally reassuming their initial shape at high temperatures.