Hacker News Reader

Comment by rl3 | Hacker News Reader

>... they’ll be able to piece together more precise time stamps for history’s uncertainties: the construction of Egypt’s Great Pyramid of Giza, the collapse of the Mayan civilization, and maybe even the arrival of the Vikings in the Americas.

That should require some really old trees, no? Wikipedia puts the The Great Pyramid of Giza for example at 2580–2560 BC. That's about 4600 years ago.

https://en.m.wikipedia.org/wiki/List_of_oldest_trees

According to that, there's only three known trees on the planet that were alive at the time.

Of course, maybe I'm misunderstanding; perhaps it's the case that by accounting for the past few thousand years of solar events, carbon dating will become far more accurate even for periods in which no tree alive at the time is available today.

View on HN

While the oldest living trees are there, its not living trees that are being dated. If you find a timber used in the construction of the pyramid, you can then match its tree rings up with another set of tree rings that overlap, and so on back to the present age. In theory.

See https://www.crowcanyon.org/index.php/dendrochronology

The challenge is there are gaps and spots where it doesn't quite match up. A bad few years and it throws the tree rings off. However, if you can pinpoint some reference event that spans the globe, you can line the samples back up again. This is known as cross dating ( http://www.ltrr.arizona.edu/lorim/basic.html ) - it doesn't need to have a living tree.

Radiocarbon dating will never be precise (being able to pin down a specific year) because of the probabilistic nature of decay and the errors on that are on the sample. Carbon dating only gives a one sigma. Something that is 3500 BP (before present) can have a +/- 60 year error on it. Give https://en.wikipedia.org/wiki/Calculation_of_radiocarbon_dat... a read.

That's incredibly interesting, thanks.

>It doesn't need to have a living tree.

Faulty assumption on my part. Obviously one can date dead wood just the same, I just assumed that non-living samples from ancient eras would be harder to come by or of lower quality having been deconstructed. While it seems that there's some truth to that, it couldn't be more wrong from a composite point of view.

That's very cool, could be a interesting problem for using machine learning. Train it with all the pictures of tree rings with known age, and then see if it can reliably guess the age of something it hasn't seen.

Its much more efficiently solved with the naive 'represent the distance between each adjacent ring as a number' and slide the array of numbers representing the sample along the known line and for each possible offset, calculate the error. Then you're left with some shortlist of candidate dates and their probabilities.

It reminds me of the algorithms for locating reads on the genome.

That's so darn cool. Dendrochronology seems like a science right out of a novel somewhere that someone said, wait, that might actually work in reality!

You don't need a single tree across the entire timespan you want to look at. You can assemble a series of overlapping samples from a number of trees that lived at different times during the timespan of interest.

From Wikipedia [1]:

> A fully anchored and cross-matched chronology for oak and pine in central Europe extends back 12,460 years, and an oak chronology goes back 7,429 years in Ireland and 6,939 years in England. Comparison of radiocarbon and dendrochronological ages supports the consistency of these two independent dendrochronological sequences. Another fully anchored chronology that extends back 8500 years exists for the bristlecone pine in the Southwest US (White Mountains of California).

[1] https://en.wikipedia.org/wiki/Dendrochronology#Growth_rings (the final paragraph of this section)

>You don't need a single tree across the entire timespan you want to look at.

Yeah, I was thinking more in terms of scarcity being an issue for samples that existed during particularly ancient periods in question. However, it seems the samples need not be living nor complete, and we've quite the extensive archive of dead wood dating further back than I could have imagined.

does this mean they have samples of tree for every year between now and those nearly 12,460 years into the past?

Reply on news.ycombinator.com