At the 1991 ICOG conference in Canberra (the forerunner to the Goldschmidt series of meetings) Claude Allegre began his plenary by stating "I'm sorry you are all wrong - Armstrong was right". Turner went back to his PhD office at the University of Adelaide to read Armstrong's paper from the meeting "The persistent myth of crustal growth". Although his arguments were largely about the balance between crustal erosion versus new additions and freeboard rather than geochemical he was and have always been troubled by the implications for early Earth.
Sitting in the tearoom at the Department in Bristol in 2003, Jon Wade told Turner he'd showed Nb can partition into the core. His interest was the Earth's Ce/Nb ratio but those (like Turner at the time) working on subduction volcanoes had a different question - how did the crust get it's negative Nb anomaly (in particular). This is a hallmark of subduction zone rocks but no-one agreed on how it is/was generated. To the contrary, it was becoming increasingly clear that subduction zone rocks inherit much of their trace element signatures from subducted sediment suggesting that the Nb anomaly might simply be being recycled. Turner realized there might be a connection but didn't explore it further at the time.
Much later (2021) John Creech showed Turner (then both at Macquarie University) the convex-downward patterns for metal loving (highly siderophile) elements for the crust and that estimated for the mantle just after core extraction and they match perfectly but the pattern for the bulk silicate Earth is flat. This is a key piece of evidence for addition of HSE by the so-called "late veneer". The latter was John's interest but Turner realized this might indicate that the crust was extracted very early on thereby inheriting the HSE pattern BEFORE the late veneer. If that was true, then the same core signature would have imparted a negative Nb anomaly.
Suddenly, all three pieces of information came together in my mind and so Turner went to Curtin University to discuss with Tim Johnson who was working on related early Earth processes with an old school friend of Turner’s, Hugh Smithies. However, they also had seemingly conflicting results. Western Australia has 2 major Archaean cratons (Yilgarn and Pilbara) but they though one had formed by subuduction and the other by intra-plate (mantle plume) processes
Johnson contacted Craig O'Neill at Queensland University of Technology because they were independently thinking about the early bombardment history of the Earth and implications for remelting of early crust. As the team discussed our different lines of thinking they realised the profound implications of a protocrust having a subduction-like signature. It would, for example, explain why the numerous studies aiming to use geochemistry to date when plate tectonics (subduction) started on Earth kept arriving at (often vastly) different ages. Perhaps we had been asking the wrong question all along or to quote Niels Bohr: "What idiots we've all been".
The next thing was to contact another long-term colleague, Bernard Wood (University of Oxford), since probably no-one has thought more about issues surrounding core formation in the early Earth. Moreover, he had worked with Jon Wade on the Nb partitioning. He and Turner then worked together on numerical models to predict the composition of the core, protocrust and residual mantle quickly showing how closely the protocrust would resemble todays continental crust. O'Neill calculated the likely volume arriving at a figure also very similar to the current volume of continental crust.
A final part of the puzzle was to consider the implications for isotope systems especially short-lived 142Nd that could only record events in the first few 100 Myr of Earth history. Another long-term colleague, Bernard Bourdon (now at the University of Lyon) and his group, were the first to firmly demonstrate the presence of 142Nd anomalies in ancient crust and Turner remembered thinking when that was published that this must link to the nature of the protocrust. Therefore, he developed a numerical model combining the incompatible trace element signature Wood and Turner had calculated with short-lived isotope signatures it predicted available observations very closely. It also showed that the sort of mixing and recycling O'Neill and Johnson had been working on would mean only some of the protocrust need have 142 anomalies – again consistent with observations. Bourdon also independently verified the partial melting and mass balance calculations Tuner and Wood
There are major implications for subsequent planetary evolution. Many geochemical arguments for when and how plate tectonics began implicitly assume subduction is required to produce the continental trace element signature. These are severely compromised if this signature was already a feature of the Hadean protocrust. To quote Niels Bohr one more time “We are all agreed your theory is crazy. We can’t agree whether it is crazy enough to be likely to be true. I myself feel it is not crazy enough” (Niels Bohr to Wolfgang Pauli concerning Heisenberg’s uncertainty principle, Columbia University, 1958).