Using co-authored scientific papers as indicators of research collaboration the pattern of research linkages in the UK during the 1980s is analysed. All sectors collaborate with each other at a rate proportional to publishing size to a first approximation. However, there are deviations from this pattern. Analyzing these deviations, two groups of sectors are found that collaborate with each other more than expected: the GIPU group composed of government, industry, polytechnics and universities, and the HSNR group composed of hospitals, special health authorities (SHAs), non-profit organisations and research councils. This suggests that a biomedical innovation system co-exists with the more commonly recognized, industry-oriented system. Better knowledge of the biomedical system would augment our understanding of innovation processes and facilitate deeper insight into the contributions of a previously neglected component of the UK science base.