Banded iron formations from the late Archaean, Palaeoproterozoic, and in a few short time intervals linked with Neoproterozoic tillites, have long fascinated geoscientists with their counterintuitive occurrence at times when the oceans contained little if any oxygen. Anoxic water allows iron to exist in its Fe2+ form, thereby able to dissolve readily. The vast thicknesses and masses of BIFs demands an abundance of mobile iron, but being made predominantly of hematite (Fe2O3) their formation requires a balancing superabundance of oxygen. Many geochemists believe photosynthesising blue-green bacteria to have excreted oxygen to oxidise soluble iron to Fe3+ and precipitate it as the oxide in shallow water. Yet plenty of BIFs show such delicate banding that deep water is implicated. All the BIF paradoxes would be resolved if another mechanism had caused the oxidation and precipitation of iron. A new clue to what that may have been is the discovery of iron-oxide stromatolites in the monster BIF deposits around Lake Superior (Planavsky, N. et al. 2009. Iron-oxidizing microbial ecosystems thrived in late Paleoproterozoic redox-stratified oceans. Earth and Planetary Science Letters, v. 286, p. 230-242). Iron isotopes and rare earth elements are good indicators of redox conditions, and those in the BIFs indicate anoxic waters, so free oxygen was not available. The stromatolites, however, strongly suggest biogenic precipitation of iron oxide, which is possible through the action of specialist Fe-oxidising bacteria. Indeed, filamentous microfossils occur in the stromatolites. That opens the possibility of BIFs having formed by direct bacterial precipitation in the oxygen-free world before the Great Oxidation Event around 2.2 Ga, in the absence of cyanobacteria.