How correct are our dates?

It’s a question of accuracy. With a few errors thrown into the mix. Read on to see what we mean. When we report a date for a human fossil, we are presenting a result that has been calculated using a set of data. We are also making a number of assumptions for each method of dating. The short answer to the question of how correct our dates are is – any reported date is correct if all of our assumptions are correct. The longer answer needs a little explaining. We’re not going to dig too deeply into the statistical analysis that is associated with errors of measurement, but we’ll use an analogy that may help explain things. Imagine that we are standing at the top of a long, dimly-lit indoor archery range, with a bright spotlight behind us. Directly in front of us is a well-lit target. We have a bow and a quiver of arrows, and our job is to hit the target with as much accuracy and precision as possible. You are watching us do this from the stands behind. The first target is located well inside the light. We can all see it clearly. We take our bow, select our arrows and start placing our shots – each shot hitting the target with a resounding thunk! How precise we are is a measure of how closely we can group our shots. Precision refers to how uniform or repeatable our measurements are. If we take ten measurements and the results are very close, then our measurement is precise. The spread is caused by random errors, but we are being precise. In the analogy, we may not necessarily be hitting the bullseye of the target, but our shots are nicely grouped together. How accurate we are is the measure of how close we are to hitting the bullseye. If an arrow hits the bullseye, it is accurate. If it misses the bullseye, it is less accurate. If it misses the target altogether – well it’s not accurate at all. Accuracy refers to how close our result is to the actual date. The offset is caused by some systematic equipment errors here – the condition and use of our bow and arrows. We may also have systematic method errors – the way that we use the science of how the arrow will behave in flight and what corrections we make for the range and conditions to increase accuracy.

Our aim?

What we are aiming to achieve is a tightly-grouped (precise) collection of shots in the centre of the bullseye (accurate). Now this is all well and good when everybody can see our brightly-lit target right there in front of us. This is how we set up our dating methods. We date material that we already know the age of, to see how precise and accurate we are. The result is clear, we can see it and you can see it, right there on the brightly lit target. Now we move the target further down the range – right to the edge of the radius of light from our spotlight. We can just make out the overall shape of the target, but not the details. We take our bow, select our first arrow and begin the process again. We know that our equipment is good, because we set it up earlier for the short-range target. Our method also appeared to be working, as our precision and accuracy were good at short range. But now we need to make some adjustments to accommodate for the longer range. We understand the physics, but we need to factor in a few variables. We fire the shot, it looks pretty good, and we all hear the resounding thunk! as the arrow strikes the target. This is how we calibrate our dating methods, such as adjusting radiocarbon dates using calibration curves. How precise were we? Well, we fire off some more shots, watch the flight of the arrows and listen to the thunk! Precision is easy to measure from the spread of our arrows. How accurate were we? Well, here is the problem: the target no longer has a known bullseye. We assume that the bullseye is within the spread of our arrows. It certainly was when the location of the bullseye was known.

A little further

Now we move the target right to the back of the range, way beyond the limits of the spotlight and into the impenetrable darkness. We take our bow and arrow and fire off our first shot… and… thunk! Ok, we have a result. We try again. The first thing we discover – our arrows scatter widely. The precision is not good. But is the bullseye within the spread of our arrows? If nothing has changed, it should be. But what if there is a sidewind further down the range that we don’t know about? A wind blowing from left to right. All our arrows would be pushed to the right. No matter how often we shoot at the target, none will be close to the bullseye. If we knew about the existence of the sidewind, we can adjust our shots. But what if the sidewind wasn’t constant, and changed in strength from day to day? How did we go? How accurate were we? We feel pretty good about our equipment. For our method, however, we are starting to make some significant assumptions. This is how we use models and theories to map environments and conditions of the past in order to arrive at a result. Is our result correct? Only if all of our assumptions have been correct. A colleague then arrives, steps up with a different bow and she takes a shot. Thunk! But her arrow strikes quite some distance from our shots. We ask her to take another shot and she makes a few adjustments and does. Thunk! Ah, this time her arrow seems to strike more closely to where our shots went.

Many methods

Any analogy can be stretched – and we may be guilty of that here. The point is that we don’t know exactly what the target date is. We can only report our results, based upon our equipment, our methods and the assumptions that we are making. Along with a margin of precision. (We’ll look at some of these assumptions and margins when we consider our three dating methods later this week.) We try to take as many measurements as we can with the same equipment and method. We try to use as many different dating methods as we can on the same target. Then we compare the results. If three completely different methods all return a date range very close to each other, then that would appear to be a confirmation of the date. If three completely different laboratories and scientists using the same method return a date range very close to each other, then that would also appear to confirm the date. All teams would include their margins for error in their reported results. However – reverting back to our analogy – all six results may be experiencing the same unrecognised side wind. We asked Mathieu why it was important to use different dating methods. But (if you will allow us to completely stretch our analogy to breaking point) what if the target was made of some unique and irreplaceable material that was priceless? And each shot was punching a hole in it? We are very limited in the number of tests that we can make on human fossils to confirm our result. This is why we are always working to come up with non-destructive methods of dating human fossils. And one last thing – what if each arrow cost thousands of dollars? You get the idea. As new technologies and discoveries arrive, we will be able to shed more light upon our subject, design better equipment and methods – and continue to test our assumptions further. In the meantime, we do the same as scientists throughout history have always done. The best we can at any given time. Research is continually improving equipment, procedures and the understanding of possible error sources. We describe our methods, explain our assumptions, calculate our errors and report the result.

Your task

Congratulations for working your way through that analogy. So, what do you think? Select the comments link below and let us know if this has surprised you or confirmed what you already knew. Perhaps it is a topic that you have not given much consideration to before now. Take a moment to let us know.

References

Grün, R. (2006). “Direct Dating of Human Fossils”. Yearbook of Physical Anthropology, 49: 2-48. Aitken, M.J. (1990). Science-based dating in Archaeology. Longman Inc., New York.