Can we trust radiocarbon dating?

Willard Frank Libby, inventor of the method

Can we trust radiocarbon dating? Young Earth creationists tell us that we can’t. After all, it makes the same range of assumptions as other radiometric dating methods, and then some. Other methods benefit from internal checks or duplications, which in the case of radiocarbon dating are generally absent. There are numerous cases where it appears to give absurdly old ages for young material, while apparent ages of a few tens of thousands of years are regularly reported for material known on other evidence to be millions of years old. So can the Young Earth creationist¹ objections be rebutted, and if so how?

The principle of radiometric dating is simple.² If we know how much of a particular radioactive substance was present when a material formed, how much is still there, and how quickly the substance decays, then using the arithmetic of exponential decay, we can work out how much time has passed. For example, after one half-life we will have half the initial amount of that substance, after two half-lives only a quarter, after three half-lives just an eighth and so on, and there is a simple equation to deal with all amounts in between.

Radioactive decay of 14C (Science Learning Hub)

As for how we know how much of the substance there was to start with, we can in most cases find that out easily enough by adding together the amount still there and the amount of the daughter substance into which it decays. For example, the most long-lived isotope of uranium, uranium-238 decays to give lead-206, so that adding the number of atoms of lead-206 to the number of atoms of uranium-238 still present gives us the number of atoms of uranium-238 that were there originally.

Historically, radiometric dating of any material was said to involve three assumptions. The first of these was that decay rates were constant, so that the half-life of any isotope in the material was same today as it had been since the material was originally formed. The second was that results were not distorted by daughter isotopes initially present. In the example given above, for instance, if the material had contained any lead-206 when it first formed, that would lead us to overestimate the amount of uranium-238 that had decayed, and hence the age of the material. The third was that materials had not migrated. If the lead-206 formed had been able to diffuse away, our estimated age would be too young. Young Earth advocates make great play of criticising these assumptions, quoting papers a hundred years old in order to do so.

George Gamow (R) in 1931 photo of W.H. Bragg’s research group (Bragg seted centre). Public domain

However, for most radiometric dating methods, none of this should concern us, provided the material is competently examined and the results interpreted using up-to-date techniques. George Gamow showed back in 1928 that decay rates can be predicted from the time-dependent Schroedinger equation, and for any radioactive isotope are fixed by the properties of the nucleus, the fundamental constants of nature, and the laws of quantum mechanics. These we know, from analysis of light coming to us from distant galaxies, to have been unchanged for many billions of years, and if they had been different, so would the laws of physics and chemistry, and the material in question would never have been able to form in the first place. The second and third objections are dealt with by a number of techniques. In what are called isochron methods, isotopes not involved in dating (such as lead-204) are used as internal checks. Some minerals, such as zircons, do not accept lead into their crystal lattice when they form, so any present must have been formed in place by radioactive decay. Modern methods involve sampling from extremely small areas, and any effects of material migration would show up as inconsistencies. Moreover, minerals can generally be dated using more than one radiometric technique. Uranium-238/lead-206 dating is automatically duplicated by uranium-235/lead-207 (uranium-235 is the fissionable isotope of uranium, taking part in chain reactions in power plants and in nuclear weapon. It is now present only to the extent of 0.7%, which is perhaps as well). Numerous other methods are available, the most commonly used of these being potassium/argon.

Radiocarbon dating is a little bit different. All other radioactive isotopes were present from the formation of the Earth, and radiometric dating tells us the time since they were embedded in the material under examination. Carbon-14, however, (14C for short), has a half-life of a mere 5730 years, and the only reason that there is any around at all, is that it is being continually regenerated in the upper atmosphere, by the action of cosmic rays on ordinary nitrogen (nitrogen-14). The 14C is then mixed, as CO2, into the atmosphere and hence into the general pool of circulating carbon in the atmosphere and biosphere. Plants exchange carbon with this pool for as long as they are alive. However, as soon as the plant material is removed from the circulating pool by death, or by incorporation into metabolically inactive material such as the heartwood of trees, the amount of 14C begins to fall by radioactive decay, as it changes back into nitrogen. So the radioactive age tells you the length of time since once-living material was withdrawn from the pool. For animal-derived material, the age refers to the plant material from which it was ultimately derived, but this is unlikely to make any serious difference. Care is necessary, however, to ensure that the carbon in the sample really was derived from the general pool. Near limestone caves, for example, recent material will appear ancient if part of the carbon comes from limestone, calcium carbonate, laid down millions of years ago. All of this has been known for decades.

Radiocarbon dating also depends on estimates of the rate of production at the time that a sample was formed, compared to the current³ rate of production. If, for example, the rate of production had been dramatically lower before a certain date (the data of Noah’s Flood in Young Earth creationist chronology, for example), then radiocarbon dating will bestow spurious antiquity on any material generated before that date. Moreover, since the daughter, nitrogen-14, is ubiquitous, we have no independent way of assessing how much 14C was initially present, and other refinements such as the isochron method and cross-comparisons are not available. So do we have any good way of rebutting the Young Earthers’ challenge?

Spoiler (and I think few readers will be surprised by this), yes. The case is laid out in a recent paper by Gregg Davidson and my friend Ken Wolgemuth in Perspectives in Science and Christian Faith, 70 (2), 75, June 2018, abstract available here (the full text will be freely downloadable from January 2019), of which Ken has kindly sent me a copy. What follows is derived mainly from this paper, with a few observations of my own blended in.

Why this paper and why this journal, when I have no faith, Christian or otherwise? Three reasons.

Firstly, tactics. Young Earth creationism is motivated by fears that one cannot be a good Christian unless one accepts a biblical literalist chronology, according to which the Earth is only some 6,000 years old. By choosing a refutation of Young Earth creationism whose authors are publicly committed to Christianity, I hope to show that these fears are baseless.

Secondly, this is the best exposition I have ever seen of the scientific case for accepting the validity of the method.

The third reason I will leave until later.

The argument brings together evidence from beryllium-10 (also produced by cosmic rays), tree rings, lake varves (annual bands of sediment), volcanology, and archaeology.

The Great Isaiah Scroll, whose wrappings Libby dated in 1955, via Wikipedia

Consider the Young Earth claim that the rate of production of 14C from cosmic rays may have fluctuated. Indeed it must have, since the Earth’s magnetic field has fluctuated, and affects any charged particles approaching the Earth from space. This possibility was realised from the outset, and mentioned by Willard Libby, inventor of the method, in his 1952 book on the subject. Libby argued that the rate of production could not have fluctuated greatly, since the amount now circulating is in equilibrium, with rates of production and decay as far as we can tell equal, and this inventory must have taken a considerable time to accumulate. Moreover, as Libby also mentioned, there was good agreement between the age of the Isaiah Dead Sea Scroll, known from archaeological evidence, and the age that Libby inferred from its 14C content.

There is further confirmation of this, from a totally different direction. Fortunately, 14C is not the only unstable isotope produced by cosmic ray bombardment. It is accompanied by formation of beryllium-10, which has a half-life of over a million years, so its decay can be ignored in our discussion. Beryllium-10 can be measured in ice cores, and ice cores show annual banding, giving us the year by year record that we need. It is indeed true that the production rate varies over time, and that the variation is significant, but we can construct upper and lower expected decay curves for the amount of 14C relative to the present depending on exactly what we assume regarding the production rate. These curves do not differ by very much.

Tree rings in loblolly pine. “Pollinator” via Wikipedia

Next, tree rings. Every year, a growing tree adds one more ring to its heartwood. This ring consists of one layer of lighter wood laid down during the rapid growth season, and one darker band laid down during winter (where growth is limited by sunlight and temperature) or the dry season (where it is limited by the availability of water). Once laid down, this wood takes no further part in exchange with atmospheric 14C.

Californian bristlecone pines have been found with more than 5,000 rings (these are sampled by boring, with minimal damage to the tree). By measuring the amount of 14C in each of these, we can construct a direct 5,000 year calibration of 14C amount against time. Actually, we can do quite a lot better than that. Year by year, more or less favourable growing conditions will mean wider or narrower bands, creating a kind of date-specific barcode. By lining up the barcode of wood from a living tree with that from wood preserved, for example, in a peat bog, we can extend the record backwards for a total of 14,000 years. This, of course, is more than twice the age of the Earth as calculated by Archbishop Ussher back in 1650. Ussher was a fair-minded and estimable scholar, who worked just a few decades before the first glimmerings of the modern science of geology, and can hardly be blamed if some of his successors insist on ignoring everything we have learnt since then.

Coring platform at work on Lake Suigetsu (from Lake Suigetsu Varves website)

Next, we have varves, the thin distinct annual bands deposited in placid lake waters. One place with an outstanding varve series is Lake Suigetsu, in Japan. Water flows into this lake by way of Lake Mikata, where coarse material settles out, leaving only the finer material to settle in Lake Suigetsu. Each spring, algae flourish in this lake, giving rise to a pale-coloured layer easily identified in cores taken from the lake bed. In addition, there are layers of volcanic ash (which will turn out later to be of great significance), and enough organic material in the form of leaves and twigs for 14C content to be assessed.

Unfortunately, there was a period during Lake Suigetsu’s history when the water flow was more turbulent and the varves did not form. However, there are something like 50,000 varves below this point, with 14C content decreasing as expected with depth, and a 5,000-varve overlap in 14C content with what is observed in tree rings. So using only the assumptions that the tree rings and the varves both represent years, we can combine the tree ring data, the varve data, and the predicted radioactive decay curves together, to give the Figure shown (Figure 10 in Davidson and Wolgemuth’s paper).

Tree ring and varve count vs 14C (percentage in modern biological carbon). Solid lines represent window for conventional expectations.

What does this mean? The tree ring data mean that the radiocarbon content in wood decreases exactly as predicted if the rate of formation has varied within the limits expected from the beryllium-10 studies, the laws of radioactive decay have held steady, and tree rings really are annual events. The varve data mean that radiocarbon content falls away with depth exactly as predicted if the same holds true for varves. The excellent agreement over a range representing five thousand years between the tree ring data and the varve data add further confirmation that they are measuring events with the same time spacing, and the way that these data fall within the expected limits gives further confirmation of the assumption, queried by Young Earth proponents, that this time spacing is indeed year-to-year.³

The situation is a little bit like the situation with evolution. Evolutionary family trees can be built up with the help of the fossil record, or by the methods of molecular biology such as DNA sequencing, or by reference to deeply built-in anatomical resemblances (homology). For the conclusions to be wrong, not only would each of these three arguments have had to fail, but they would have had to fail in a coordinated manner. Similarly, for radiocarbon dating to be seriously wrong, the arguments from the general nature of radioactive decay, and the data from the tree rings and the varves, would all have to mislead us, and mislead us, quantitatively, in exactly the same way.

Two final scientific arguments, one from geology and the other from archaeology. I mentioned the existence of layers of ash among the Lake Suigetsu varves. Such ash can arise from eruptions in volcanically active regions in Korea or Japan, and we can tell which is which by looking at the chemical composition. There is volcanic ash around the varves that match up to tree ring 10,200, and chemical analysis shows that this came from an eruption in South Korea, an eruption that has been dated to 10,000 +/- 300 years before present. As mentioned above, we also have had, since 1965, radiocarbon measurements for the Dead Sea Scrolls, and these match the radiocarbon measurements for tree ring ~2,230.

One final argument used to cast doubt on radiometric dating. There is a steady stream of reports in the creationist literature of measurements carried out by reputable laboratories on dinosaur bones, coal, or other material known to be many millions of years old, and coming up with ages measured in tens of thousands of years.

There is a simple and well-known reason for this, namely bacterial contamination. Bacteria are everywhere, even in the solutions used to remove them when samples are cleaned for examination. If as little as one part in a thousand of the carbon drawn into the analysing system comes from this contamination, then however old the sample, and however small its true 14C content, it cannot possibly show a 14C content less than 0.001 of that for recent material, and therefore it cannot possibly show an apparent age greater than around ten half-lives, or 57,000 years. 1% contamination would imply a maximum apparent age of 38,000 years.⁵ Laboratory tests confirm that when cleaning procedures are applied to ancient samples of graphite, derived from coal⁶ that has been heated to very high temperatures during its geological lifetime, it suddenly acquires such a spurious illusion of relative youth.

Now for my final reason for citing this particular paper, namely its peroration, addressed specifically by science-accepting Christians to Christians who in the name of their faith deny the scientific evidence. The authors point to the wonders of the world, and the even more wonderful fact that we are capable of understanding it through scientific investigation. Yet Young Earth proponents would undermine this understanding. Commenting as I must from the outside, I cannot do better than quote their actual words, showing all of us how we can address those we disagree with in their own terms:

“Tree-ring growth, atmospheric carbon-14 production, radiometric decay rates, sediment couplets, and ash chemistry are all independent phenomena… The data, in total, fit amazingly well with conventional geologic understanding, requiring no disruption of natural laws or unfathomably improbable alignment of unrelated processes.”

Young Earth creationism, in contrast,

“requires supernatural manipulation of nature with no apparent purpose other than to mislead. Many in the world marvel at the handiwork of God while denying that Creator. In response, young-earth advocates demand that to acknowledge the Creator, we must deny his workmanship. Can there be a more ineffective witness?”

One does not need to be a believer to respond to these sentiments.

h/t Lorence Collins. Thanks to Gregg Davidson and Ken Wolgemuth for discussions and access to materials.

1] As opposed to Old Earth creationists, who accept the science of geology while rejecting that of biology.

2] My favourite exposition of radiometric dating methods comes from Roger Wiens and is posted on the ASA site. Wiens is a distinguished scientist working at Los Alamos, designer of equipment deployed on the Mars Rover. The ASA, which is also publisher of the paper I am discussing here, is committed to a biblical Christianity in which God has created a world that is intelligible.

3] Strictly speaking, what is used is the rate of production in 1950, which is the “present” referred to when geologists speak of dates “before present”. That is because of the formation of large amounts of 14C in the atmosphere in the 1960s, as a result of nuclear weapons testing, as I discuss here.

4] The authors discuss in greater detail than would interest most readers here the alternatives suggested by Young Eaerthers, who have seriously suggested that there were major fluctuations in the rate of 14C production, or that under the disturbed conditions of the Flood catastrophe varves and tree rings could have occurred with greatly enhanced frequency, but the paper shows that in order to match the data, disturbances in the formation of tree rings, varves, and 14C would have had to be extremely tightly coordinated, although no one has ever suggested the mechanism by which this would happen, and that in addition it would have been necessary to generate thousands of varves and tree rings within a single year.

5] This follows from the decay equation, which shows that the fraction remaining after n half-lives is 1/2ⁿ.  For example, 0.01 = 1/2^6.64, implying an age of 6.64 half-lives, while 0.001 = 1/2^9.96 . The contamination error will be small for young samples, but impose a ceiling on the apparent age of all samples however ancient they may actually be.

6] Coal generally contains noticeable amounts of organic molecules that include nitrogen, which acts as a further background source of 14C; this has been known since the 1930s. However, the more extreme the conditions that the coal has met since its initial formation, and the more thoroughly this has moved it in the direction of becoming pure graphite, the less 14C contamination there is from this source.

Tree ring and varve count vs 14C, figure from paper cited, courtesy the authors. Willard Frank Libby image copyright Nobel Foundation. Fair use justification – historic personage, no pubic domain available, copyright holders hold higher resolution image.