This "parent rock" was very probably basalt of the sort produced on the ocean floor, they say.

The team reports its work in this week's Science Magazine.

With most rock on Earth being broken down by weathering or taken back into the planet's interior just a few hundred million years after production, there is very little, truly ancient material to look at.

Geological genealogy offers a very useful alternative in the circumstances, says Dr Jonathan O'Neil from the University of Ottawa, Canada.

"To put everything into perspective, the Earth is about 4.6 billion years old, and I would say that rocks that are 3.8 billion years and older - we can count them on the fingers of our hands.

"We have very limited places to learn something about the first billion years of Earth history. That's our challenge," he told the Science in Action programme on the BBC World Service.

Dr O'Neil and colleagues examined samples of rock from the "Canadian Shield", a swathe of territory moving east of Hudson Bay in Canada.

Image copyrightUNIVERSITY OF OTTAWA

The specific pieces they took into the lab were relatively young - "only" 2.7bn years old.

But by looking at an atomic tracer in the minerals, the team could determine the parentage.

This tracer takes the form of a radioactive decay chain centred on versions, or isotopes, of the chemical elements samarium and neodymium.

The isotope of samarium with an atomic mass of 146 is unstable and decays very rapidly into the isotope of neodymium that has a mass of 142.

In fact the decay is so rapid that any samarium-146 that was present on Earth at its formation 4.6 billion years ago is all but gone within 500 million years. But even though "the clock" stops running, it leaves the mark of its existence in the concentration and distribution of the neodymium.

When the scientists take into account the known age of their granite samples, the time it would take for these granites to form from a recycled parent rock, and the neodymium isotope record present in the samples - they conclude the pre-cursor material must be older than 4.2 billion years.

"We can also say something about [the parent rock's] nature," said Dr O'Neil.

"Were they granites or basalts? Were they continental crust or were they oceanic crust? We think they were oceanic crust. So, the primitive earliest crust on Earth - at least the evidence we have in our samples - was oceanic crust at approximately 4.3 billion years old."

Image copyrightUNIVERSITY OF OTTAWA

Does any of this original material still exist? Dr O'Neil points to a nearby sliver of rock on the shore of the Hudson Bay called the Nuvvuagittuq Supracrustal Belt (NSB). This is presumed to be oceanic volcanic rock and has been dated to be between 3.7 and 4.3 billion years old. It is an age bracket that also relies on a neodymium analysis.

"The younger rocks that we have analysed are around [the NSB], so it makes sense that this is the piece of crust that has been recycled to form our continents," said Dr O'Neil.

"It fits the bill. It has exactly the right composition and the right mineralogy to be it."

The Nuvvuagittuq Supracrustal Belt was reported earlier this month by a team led from University College London (UCL), UK, to contain the fossilised remains of tiny lifeforms.

However, it has to be said that there is significant debate over the exact age of the NSB, which may not quite fit with this genealogy story.

"The neodymium isotope compositions in the volcanic rocks of the NSB do indeed yield an age of 4.28 billion years, but this has been interpreted both as the true age of this rock (the age of when the lava froze into rock) and alternatively as an older signature inherited from the early mantle of the Earth in these younger rocks, that have a minimum age of 3.77 billion years," explained UCL's Dr Dominic Papineau.

"This debate can possibly be settled using new approaches and techniques, as well as using careful microscopy-based observations of rocks from key geological relationships in the field," he told BBC News.