All anoxygenic bacteria currently known have photosynthetic reaction centres of only one type, either Type I or II, each working with a dedicated light-harvesting antenna. In contrast, all oxygenic photosynthetic systems – of plants, algae and cyanobacteria – have both Type I and Type II reaction centres. Oxygen evolution from oxidation of water requires a Type II reaction centre that is connected, in series, with a Type I reaction centre. Around 2.4 billion years ago, the evolutionary origin of this series connection enabled photosynthetic water oxidation, which began to transform our planet irrevocably. Here I consider the question of how separate Type I and Type II reaction centres diverged from a common ancestor. How they later became linked together, to become interdependent, is also considered, and an answer proposed. The “redox switch hypothesis” for the first oxygenic cyanobacterium envisages an evolutionary precursor in which Type I and Type II reaction centre genes were present in the genome of a single anoxygenic bacterial lineage, but never expressed at the same time, their gene products serving to provide different, separate reaction centres for light energy conversion under different growth conditions. I suggest that a mutation disrupting redox control allowed the two reaction centres to co-exist. This arrangement was selected against prior to the acquisition of a catalyst of water oxidation, but had a selective advantage thereafter. Predictions of this hypothesis include a modern, anoxygenic descendent of the proto-cyanobacterium whose disabled redox switch triggered the Great Oxidation Event, transforming both biology and geology in the creation of the our aerobic planetary habitat.