Arrays of neighbouring histone H1 molecules are present in rat liver chromatin, and can be cross-linked to each other by bisimidoesters. The product, poly(H1), can be extracted from cross-linked chromatin in high yield with 5% perchloric acid and thus distinguished from cross-linked oligomers of the core histones. On analysis in dodecyl sulphate/polyacrylamide gels poly(H1) gives a striking pattern of alternating strong and weak bands which have been shown, by the use of cleavable cross-linked reagents and two-dimensional gel electrophoresis, to arise respectively from H1 homopolymers and H1 polymers linked to the nucleosome core histones. H1-H1 proximities as measured by cross-linking exist both at low ionic strength where the nucleosome filament is extended, and at higher ionic strengths at which it is folded into a 30-nm-diameter fibre, probably in the form of a solenoid, although some additional H1-H1 contacts seem to occur in the folded form. A similar pattern of H1-H1 cross-linking is observed for nucleosome oligomers too short to form a complete turn of a solenoid. The lack of any strong dependence of the H1 cross-linking pattern on ionic strength, and the results for short oligomers, suggest that in a solenoid the major H1-H1 interactions are lateral (i.e. they occur between neighbouring nucleosomes along a turn) although additional vertical contacts (between successive turns) are not excluded. For a short nucleosome oligomer containing n nucleosomes, the number of H1 molecules that may be cross-linked into an array at low ionic strength (approximately 15 mM) is n, consistent with the presence of one H1 per nucleosome. H1-H1 cross-linking occurs only within the chromatin framework; when cross-linking is carried out in 0.5 M NaCl so that H1 is dissociated from chromatin, no poly(H1) is formed. When H1 which has been dissociated in 0.5 M NaCl is allowed to reassociate by gradual lowering of the ionic strength, the poly(H1) is identical with that in native chromatin, suggesting faithful rebinding of H1 molecules. H1-H1 proximities do not seem to be mediated by the presence of bound high-mobility-group non-histone proteins because the poly(H1) pattern persists when cross-linking is carried out at 365 mM ionic strength, when these high-mobility-group proteins are dissociated from chromatin. The H1 cross-linking pattern described here could prove to be a useful assay for the native arrangement of H1 molecules in reconstituted chromatin and for changes in H1-H1 contacts that might result from cell-cycle modifications (e.g phosphorylation) of H1. It may not prove a good assay for higher-order structure because of the relative insensitivity to ionic strength of the poly(H1) pattern as studied here. A detailed analysis will be necessary to determine whether there are any subtle differences in the interactions between H1 molecules in the folded and extended nucleosome filament.