Since the first publications reporting the characterization of starch synthase and starch phosphorylase, it was found that they both required exogenous glucan primer to form new a-glucosidic bonds, i.e. they could only transfer glucose from ADPglucose or glucose-1 phosphate, respectively, to a pre-existing glucan primer. Later, however, a number of reports showed that starch synthases from potato tuber and spinach leaf and the phosphorylase of potato tuber can catalyse the synthesis of a-glucan in the absence of added primer, i.e. they have ‘unprimed activity’. Several studies on potato tuber were the basis for the hypothesis that a phosphorylase isoform localized on amyloplasts can synthesize a-l,4-glucan chains on a protein acceptor; the protein-glucan so formed would act as a primer for the starch synthase. If initiation of starch biosynthesis in vivo followed the pathway proposed, unprimed phosphorylase activity would be expected to be present in all plant tissues that accumulate starch. The phosphorylase from spinach chloroplasts was chosen for this study because it is homologous to the potato-tuber amyloplast enzyme and because in the leaf, the chloroplast is the site of starch accumulation. The questions to be answered were the following. Does the chloroplastic phosphorylase have activity in the absence of added primer? What is the nature of the product of the unprimed activity? Is the endogenous primer a protein? Could the unprimed reaction occur in physiological conditions? Indeed, the chloroplast phosphorylase from spinach leaves was found to be capable of synthesizing ct-l,4-glucan in the absence of added glucan primer but, apparently, no protein primer is required. The ‘unprimed’ activity seems to consist of the elongation of pre-existing, endogenous primer associated with the enzyme in a non-covalent fashion. The unprimed activity of the choloroplastic phosphorylase resembles those of the starch synthases and glycogen synthases of several sources as far as the stimulation by citrate, presence of non-covalently attached polyglucose in the enzyme preparation, properties of the product, and in the interactions with branching enzyme. It is concluded that the criteria used in the past for the measurement and characterization of putative proteoglucan are inadequate and that a new approach is required in the investigation of the initiation of starch synthesis.