Although the first United Kingdom oil refineries came into production as early as the I920S, the industry has only become established on a major scale since I945. Closely associated with this spectacular post-war increase in refinery capacity has been the establishment and rapid growth of an entirely new manufacturing technology which has significantly affected the structure and organization of the chemical industry. The position of the oil refineries as suppliers of feedstocks to the petro-chemical sector of the chemical industry has involved the development of numerous product transters between the two industries. The relationship between these physical linkages and the spatial association of the two industries is examined by reference to the development of three major oil refinery/petro-chemical complexes in the United Kingdom. The case-studies are examined in relation to the concept of a model complex in which an input-output balance between successive stages of production is maintained. It is shown that, in practice, oil refinery/petro-chemical complexes rarely approach the highly integrated input-output system of the model. Technical factors associated with economies of scale and the introduction of long-distance pipelines are partly responsible for this situation. In addition, the different approaches to corporate growth in the petro-chemical industry adopted by the three major oil companies involved in each of the casestudy complexes have superimposed a behavioural variable upon their technical structure. The discrepancies between the model complex and reality are significant not only in terms of the relationship between linkage and agglomeration in manufacturing industry, but also as far as the use of oil refinery/petro-chemical complexes as 'growth centres' in regional planning is concerned. THE distribution of economic activity over space is clustered rather than dispersed. There is much evidence to support this generalization and various models have postulated the concept of 'cumulative causation' to indicate the inter-acting forces encouraging the geographical polarization of economic growth.1 This tendency is especially apparent in the manufacturing sector and, despite some indications of a recent decline in the spatial concentration of industry within advanced economies,2 the need for a better understanding of agglomeration remains an important problem in industrial geography. There have been many theoretical attempts3 to elaborate upon Weber's4 rather superficial analysis of agglomeration, but the basic difficulty of combining the many factors implicit in the term 'agglomeration economies' into a comprehensive theory of industrial location has not been surmounted. Thus, Wood probably reflects the consensus view when he suggests that the analysis of agglomeration in manufacturing is '... one of the most unsatisfactory aspects of traditional location theory'.5 The deficiencies of location theory in the analysis of agglomeration in industrial location ensure that empirical studies of the problem lack a unifying conceptual framework. Most casestudies of industrial linkage and agglomeration have been confined to small-scale and highly specialized activities, often of 19th-century origin. Wise,6 for example, has studied the gun and jewellery 'quarters' in Birmingham and Martin7 has examined the distinctive concentrations of precision engineering and furniture manufacture in parts of Inner London. Keeble's8 work represented a significant advance in the sense that he concentrated upon modern 'growth' industries such as engineering and his conclusion that linkage on national and regional scales is becoming more important than local input-output relationships confirmed similar observations by Goodwin9 relating to the motor vehicle industry and is also supported by the findings of