The extent of duplication within and between barley genebank collections has been determined from a comparative analysis of four European barley collections. These collections were those of the Centre for Genetic Resources, the Netherlands, the Institute of Plant Genetics and Crop Plant Research and the Braunschweig Genetic Resources Centre, both in Germany, and finally that of the John Innes Institute in the United Kingdom. These collections comprise over thirty thousand accessions. A random set of 100 accessions from the CGN barley collections was matched with all four collections to identify probable duplication, resulting in 61% of the accessions being matched. Passport data were used to identify probable duplicates. This was hindered by the low quality of these data; simple perfect matches were rare. The probable duplication, excluding parental duplication, involved 144 other accessions. Using seed traits, field observations and electrophoretic markers, it could be shown that more than three quarters of the matched accessions in the random set, excluding parental duplication, were completely duplicated in the others and one fifth were more or less duplicated in other matched accessions. If individual comparisons between accessions from the random set and the matched accessions were made, about one fifth of the probable duplicates were shown not to be duplicates at all. One half were identical duplicates, and the rest were common or partial duplicates.
