Historical Geology/Absolute dating: an overview
In this article, we shall take a look back at the methods of absolute dating, and see how we know that they can be relied on.
- 1 Basis of the methods
- 2 Sea-floor spreading
- 3 Marine sediment
- 4 Radiometric dating and paleomagnetism
- 5 Comparison with historical dates
- 6 Radiocarbon dating, varves, and dendrochronology
- 7 Radiometric dating, sclerochronology and rhythmites
- 8 Agreement with relative dating
- 9 Internal consistency of radiometric dates
- 10 Mutual consistency of radiometric dates
- 11 Summary
Basis of the methods
One argument in favor of the absolute dating methods presented in the preceding articles is that they should work in principle. If they don't, then it's not just a question of geologists being wrong about geology, but of physicists being wrong about physics and chemists being wrong about chemistry; if the geologists are wrong, entire laws of nature will have to be rewritten. Science, since it concerns just one universe with one set of laws, constitutes a seamless whole; we cannot unpick the single thread of absolute dating without the whole thing beginning to unravel.
Still, it has happened in the past that scientists have thought they'd got hold of a law of nature and then found out it was false. There is no particular reason to suspect that this will turn out to be the case when it comes to the laws underlying absolute dating; nonetheless, an argument from principle alone can never be entirely convincing. Let us therefore turn to the evidence.
You will recall from our discussion of sea floor spreading that the sea floor spreads out from mid-ocean rifts, and so ought to be younger nearer the rifts and progressively older further away from them.
What is more, we can measure the rate of spreading directly by GPS, SLR, and VLBI. This means that if we didn't have any other way of doing absolute dating, we would as a first approximation take the age of basalt on a spreading sea floor to be the distance from the rift divided by the rate of spreading.
Now if we estimate the age of the sea floor like that, then we get a good agreement with the dates produced by radiometric methods. It is hard to think that this is a coincidence; it is also hard to think of any mechanism that could produce this agreement other than that the rocks are as old as radiometric methods tell us.
We began our discussion of absolute dating by saying that sedimentation rates could not be relied on for absolute dating. If there is one possible exception to this, it would be the deposition of marine sediment, since it is not subject to erosion, and since we would expect the rates of deposition of the various sediments to be, if not actually constant, then not subject to such a degree of variation as (for example) glacial till. Based on the known rates of deposition, we may therefore at least say that the depths of marine sediment found on the sea floor are consistent with the ages of the igneous rocks beneath them as produced by radiometric dating.
Radiometric dating and paleomagnetism
The polarity of the Earth's magnetic field is a global phenomenon: at any given time it will either be normal everywhere or reversed everywhere. So if our methods of radiometric dating are correct, then we would predict that rocks dated to the same age would have the same polarity, which they do.
If this does not completely prove that radiometric dating is correct, it does at least show that (barring a wildly improbable coincidence) there is at least a one-to-one relationship between the dates produced by radiometric methods and the true dates, and so it must be taken as an argument in favor of these methods.
Comparison with historical dates
It is possible to test radiocarbon dating by using it to put a date on historical artifacts of known date, and to show that it is usually very accurate.
It has also been possible to test Ar-Ar dating against the historical record, since it is sufficiently sensitive to date rocks formed since the inception of the historical record. For example, Ar-Ar dating has been used to give an accurate date for the eruption of Vesuvius in 79 A.D, as recorded by Roman historians at the time. (See Lanphere et al., 40Ar/39Ar ages of the AD 79 eruption of Vesuvius, Italy, Bulletin of Volcanology, 69, 259–263.)
Radiocarbon dating, varves, and dendrochronology
Because varves contain organic material, it is possible to compare the dates from varves with the dates produced by radiocarbon dating, and see that they are in good agreement. We also see close agreement between dendrochronology and uncalibrated radiocarbon dates. (I specify uncalibrated dates because as radiocarbon dating is calibrated against dendrochronology, the agreement of calibrated radiocarbon dates with dendrochronology is inevitable.)
Now, each of these three methods relies on a different underlying physical process: radioactive decay, outwash from glaciers, and the growth of trees. We can hardly suppose that there is some single mechanism which would interfere with all three of these very different processes in such a way as to leave the dates derived from them still concordant.
But it is equally far-fetched to imagine that three different mechanisms interfered with the three processes in such a way as to leave the dates concordant; that would require either a preposterous coincidence, or for natural processes to be actually conspiring to deceive us: an idea which is, if anything, even more preposterous.
Now, preposterous things do happen occasionally. But in this case there is a perfectly reasonable and straightforward explanation for why the dates are concordant, namely that they are correct.
Radiometric dating, sclerochronology and rhythmites
Are we to believe that one single mechanism interfered with the decay of radioactive isotopes, the secretion of calcium carbonate by molluscs, and the action of the tide? Absurd. But are we instead to believe that three separate mechanisms interfered with these processes in such a way as to leave all the dates concordant? That would be equally absurd. The straightforward explanation for the concordance of the dates is that they are in fact correct.
Consider the following analogy: a clockmaker sells us an electric clock, a pendulum clock, and a spring-driven clock, and guarantees that they are shockproof. Skeptical of the clockmaker's claim, we subject the clocks to shock: we shake them, drop them, hit them with hammers and shoot them out of a cannon. Throughout this process, they all go on showing exactly the same time. Is it plausible that we have damaged their very different internal mechanisms in such a way that they are all running fast or slow but still in perfect synchrony? Or is it more likely that they are synchronized because nothing that's happened to them has affected their working?
Agreement with relative dating
Relative dating by definition does not produce actual dates, but it does allow us to put an order on the rocks, and so if absolute dating is to be trusted, it should agree with this order, telling us, for example, that Ordovician rocks are older than Triassic rocks; and it does.
It is hard to see this as a coincidence; it is equally hard to think of some alternate explanation of why we can correlate isotope ratios or sclerochronological data with the relative order of rocks as deduced from stratigraphic methods — other than the straightforward explanation that absolute dating is producing the right dates.
Internal consistency of radiometric dates
In our discussion of radiometric dating, we have seen that many, indeed most, radiometric methods are self-checking.
So in the U-Pb method, we check that the two uranium isotopes produce concordant dates. In the Ar-Ar method, we check that step heating yields the same date at every step. In Rb-Sr, Sm-Nd, Lu-Hf, Re-Os, La-Be, La-Ce and K-Ca dating, we check that the points we plot on the isochron diagram lie on a straight line.
It is, as we have explained, possible for the occasional incorrect date to slip through this filter, since it is possible for some of these confounding factors to accidentally change the isotope ratios in such a way as to produce something that looks like a good date: apparently concordant dates for Ar-Ar or U-Pb, or a false isochron for the various isochron methods.
It would indeed be remarkable if this never happened, since one-in-a-thousand chances do in fact occur one time in a thousand. But by the same token, the other 999 times they don't, and so although any particular date produced by these methods might be called into question, it must be the case that the vast majority of dates that pass through these filters must be good; for we can hardly suppose that the confounding factors are actively conspiring to deceive us, and so these long-shot events must be as rare as statistical considerations would lead us to expect.
Mutual consistency of radiometric dates
You might perhaps suggest that if some unknown factor, contrary to our present understanding of physics existed that sped up or slowed down radioactive decay in the past, then we would expect the radiometric dates to be concordant whether they were right or wrong.
This is, as I say, contrary to our present understanding of physics, and so is mere unfounded speculation. What is more, the reader should recollect that "radioactive decay" is not the name of one process; it is the name of any process that rearranges the nucleus. So to leave dates produced by different radiometric methods still concordant, nature would somehow have to conspire to fool us by changing the rates of alpha decay, of beta decay, and of electron capture, in such a way that the different dating methods based on these different modes of decay come up with the same dates.
Another point to bear in mind is that a change in the rate of radioactive decay, even if it was carefully coordinated in this way, would still not change every radiometric date in the same direction: if, for example, radioactive decay sped up at some time in the past then this would make U-Pb or Ar-Ar dates older than they should be, but it would make the dates produced by cosmogenic surface dating younger than they should be.
It is possible to doubt any particular date obtained by absolute dating methods. But it would be bizarre to doubt the general picture they paint. For what we see is a massive agreement between the different radiometric methods, varves, dendrochronology, sclerochronology, rhythmites, paleomagnetic data, deposition rates, sea floor spreading, and relative dating methods.
For the dates obtained by absolute dating to be wrong in general and yet wrong in such a way as to be in agreement with one another and with other observations, we would have to suppose either that we are looking at an inconceivably massive coincidence, or that the whole Earth is a fraud designed to deceive us.
Ideas to the latter effect have actually been proposed from time to time; most notably by the nineteenth century religious zealot Philip Gosse, whose eccentric work Omphalos proposed that the Earth was a mere few thousand years old, but that God had created it to look much older. To this the Reverend Charles Kingsley memorably answered: "I cannot believe that God has written on the rocks one enormous and superfluous lie for all mankind". That of course would be a theological rather than a geological question, and so is outside the scope of this textbook. What can be said is that geology is a science, and that in science it is necessary to proceed on the basis that the universe is not a lie; because if we believed that, we could believe that anything at all was the case and disregard all evidence to the contrary. The scientific method compels us, then, to disregard the possibility of divine malice; and mere natural processes, being mindless, cannot be actually malevolent.
What, then, of coincidence? Well, there are limits to the degree of coincidence we can believe in, otherwise again we could believe nearly anything. The scientific method requires us to discard such remote possibilities unless there is at least a hint of a shred of evidence for them.
We are left with the conclusion that the great majority of the dates produced by absolute dating methods must be reasonably accurate.