Abstract In the accompanying paper (Morris et al., 1982) we present evidence of Ca 2+-induced association of poly- d-galacturonate sequences from pectin into dimers of 2 1 chain symmetry, with co-operative (“egg-box”) binding of Ca 2+ on specific sites along the interior faces of each chain. We now investigate the role in calcium pectate gel networks of other structural features, in particular methyl esterification and 1,2-linked l-rhamnosyl residues in the polymer backbone. Acid hydrolysis of citrus, apple and sunflower pectins gave polygalacturonate blocks with a relatively narrow molecular weight distribution, and average chainlength of ~25 residues in each case. Since the known relative stabilities of glycosidic linkages would lead to chain cleavage predominantly at l-rhamnose, this result indicates that the length of polygalacturonate sequences between rhamnose interruptions is approximately constant within and between the pectins studied. Calcium pectate gel strength is reduced dramatically by the incorporation of these chain segments when they are de-esterified, but not when they are esterified. This interference with the development of a network structure that resists applied stress, provides further support for our model of junction zone formation from sequences of contiguous deesterified residues, with Ca 2+-mediated chain dimers providing the primary associations that can offer resistance to deformation. Samples with different levels and patterns of esterification were prepared by enzymic (blockwise) and chemical (random) de-esterification of almost fully methyl esterified pectin. In the former series, the extent of Ca 2+ binding (as monitored by circular dichroism) increased almost linearly with the fraction of free carboxyl groups, whereas the latter showed a non-linear relationship of a form consistent with the requirement of this binding for blocks of contiguous non-esterified residues and, in the presence of excess univalent cations, binding was negligible when more than ~40% of the carboxyl groups were esterified. Statistical calculations of sequence length distribution at different degrees of random de-esterification show the best fit with experimental data when binding is assumed to require sequences with seven or more consecutive free carboxyl groups along the participating face of the chain. For 2 1 chain symmetry, this corresponds to a sequence length of 14 residues, in excellent agreement with previous independent studies of Ca 2+ binding to oligogalacturonates. In the absence of competing univalent counterions, circular dichroism changes are similar in form but so large in magnitude that site-binding of Ca 2+ must now go beyond the half-stoichiometry at which it is arrested in their presence. Ca 2+ binding monitored by circular dichroism, and gel strength (yield stress) measured mechanically, both show a similar dependence on the pattern as well as the level of esterification, as expected for network formation by co-operative binding of Ca 2+ within interchain junction zones. To fit this binding data quantitatively, it is necessary to postulate a two-stage process. (1) Initial dimerization, probably corresponding to the “strong associations” indicated by evidence from competitive inhibition (see above), for which a critical minimum sequence of seven residues is again required but esterified residues can now be accommodated within individual sites provided that they are paired with a free carboxylate on the complementary chain. (2) Subsequent Ca 2+-induced aggregation of these preformed dimers, which can occur irrespective of the pattern of esterification on the external faces; the evidence from mechanical measurements shows that these contribute little to gel strength at high stress.