Replication from the origin of Escherichia coli has traditionally been visualized as two replisomes moving away from each other, each containing a leading and a lagging strand polymerase. Fluorescence microscopy studies of tagged polymerases or forks have, however, indicated that the polymerases may be confined to a single location (or a few locations in cells with overlapping replication cycles). Here, we have analysed the exact replication patterns of cells growing with four different growth and replication rates, and compared these with the distributions of SeqA foci. The SeqA foci represent replication forks because the SeqA protein binds to the newly formed hemimethylated DNA immediately following the forks. The results show that pairs of forks originating from the same origin stay coupled for most of the cell cycle and thus support the replication factory model. They also suggest that the factories consisting of four polymerases are, at the time immediately after initiation, organized into higher order structures consisting of eight or 12 polymerases. The organization into replication factories was lost when replication forks experienced a limitation in the supply of nucleotides or when the thymidylate synthetase gene was mutated. These results support the idea that the nucleotide synthesis apparatus co-localizes with the replisomes forming a 'hyperstructure' and further suggest that the integrity of the replication factories and hyperstructures is dependent on nucleotide metabolism.