Could radiometric results be cherry-picked?

Radiometric dating deniers like to portray results far in excess of six thousand years as just “rationalisations” or “just-so stories.”

Take, for example, this article on creation.com, which claims that “long-age geologists” just disregard (or “reinterpret”) any data that doesn’t fit their preconceived notions. The author, Tas Walker, correctly points out that radiometric dating doesn’t always give the results we’d expect, and that when this happens, scientists come up with an alternative, explanation for the results. He gives some examples of such explanations: heating events since the rocks cooled; xenocrysts; contamination; inherited ages; and so on.

But then he says this:

No matter what the radiometric date turned out to be, our geologist would always be able to ‘interpret’ it. He would simply change his assumptions about the history of the rock to explain the result in a plausible way. G. Wasserburg, who received the 1986 Crafoord Prize in Geosciences, said, ‘There are no bad chronometers, only bad interpretations of them!’ In fact, there is a whole range of standard explanations that geologists use to ‘interpret’ radiometric dating results.

This caricature is very misleading. As we have seen, geologists do not just look for a plausible sounding interpretation based on changing assumptions; they have to rigorously demonstrate which interpretations are consistent with the data and which ones are not. In fact, they can often discover a whole lot of other useful information about the rock strata than just their age — information such as their thermal history, for example. They most certainly are not just a case of making up excuses, or of choosing whichever interpretation you like in order to get the results that you want, and it is quite frankly dishonest to portray them as if they were.

What about unpublished results?

In any case, this caricature does not explain why up to 95% of results do not require any such “rationalisation” or “changing assumptions.” To accommodate this high degree of concordance, radiometric deniers instead sometimes claim that discordant results tend to be thrown out, or filed away in a drawer somewhere. But could this be happening often enough to undermine the credibility of radiometric dating?

If it is, the amount of data that they must be withholding is colossal. This article gives some calculations that show that merely to get meaningful isochron plots from random data (remember, isochron dating requires all the points to lie on a straight line) you would have to be throwing away several hundred results for every one that you accept. It gives a couple of examples of single isochrons from the scientific literature that would have cost as much as $1.7 million if they really had been the result of cherry-picking random numbers.

But let’s suppose that all isochron plots are all due to mixing, and do as a result give meaningless “ages.” If this is the case, you would still have to be throwing away perhaps as many as a hundred results for every one that you accept in order to achieve 95% concordance.

Radiometric dating is expensive. Geochron Laboratories quotes $750 for each point on a Rb-Sr isochron graph, and $800 per point for Sm-Nd. An isochron graph requires at least three points, and in some cases can consist of a dozen or more. Since researchers frequently use more than one dating method, the cost of dating a single sample can easily exceed $10,000. It’s fair to say that it costs as much as a small family car.

This means that if scientists really are cherry-picking data in this way, they must be spending a million dollars or more per published result on what amounts to wholescale scientific fraud. As we’ve already seen, the number of radiometric results in the literature runs into the hundreds of thousands, with tens of thousands of new results being added every year. The amount of money being squandered must run deep into the tens if not hundreds of billions of dollars worldwide.

Where, then, are the accountants and auditors complaining about this colossal waste of money? Where are the scientists working in other fields, competing with it for funding, creating a stink because they have lost out on research grants because of it? Where are the documents blowing the gaff on it on Wikileaks? And where are the US Senators — some of whom are YECs themselves — calling for the obvious fix to the problem of requiring all radiometric studies to be pre-registered?

Ultimately, the claim that “long age geologists” don’t accept results unless they match their preconceived notions is nothing more than a conspiracy theory — and a bad one at that. It may sound plausible when you first hear it, but it only takes some simple arithmetic to see it descend into absurdity. It simply doesn’t make sense.

Advertisements

About those dating methods that disagree with each other

Radiometric deniers love to point out examples of cases where different dating methods give different results on the same rock strata. The most comprehensive attempt to catalogue these that I’m aware of is an article by John Woodmorappe in the September 1979 issue of the Creation Research Society Quarterly entitled Radiometric Geochronology Reappraised, which lists about 300 examples that he had culled from the scientific literature.

Now anomalous results do happen from time to time, and lists such as these do look pretty impressive, but they greatly exaggerate their frequency, extent and significance. Here are a few points to bear in mind.

1. Discrepancies are the exception, not the rule.

As we saw last week, the number of radiometric results in the scientific literature runs into the hundreds of thousands, and quite possibly even into the millions. This being the case, three hundred bad dates represents only a tiny fraction of the overall data.

How many results overall are discordant? Without doing an exhaustive literature search, which would take forever, it is hard to tell. However, one thing is clear: when radiometric dating is correctly applied, they are very much in the minority.

This Talk Origins article cites geochronology expert Brent Dalrymple, an expert in radiometric dating with lots of real-world experience, as estimating that only between 5% and 10% of radiometric results are discordant. The article doesn’t explain how he arrived at that figure though.

A more detailed picture comes from this account by geology professor Joe Meert (scroll down to the section “A brief discussion regarding the integrity of radiometric dating”), in which he records a discussion with some YECs on the subject. The figures that they cited were, unsurprisingly, higher, claiming that up to 15-20% or so of dates were discordant. However, he cited fifteen years of his own research, in which he found that fewer than 5% of his results were anomalous, and that each and every one of them had a reasonable explanation.

Even if the figure is as high as 20%, this leaves a serious question for radiometric denialists. Why do so many of the hundreds of thousands of results in the literature show little or no discordance? This would never happen if radiometric dating really were consistently unreliable, and certainly not if it were so off-base that it could not distinguish between thousands and billions.

In any case, there are certain important data sets that show little or no discordance — specifically, meteorites and lunar rocks returned from the Apollo missions. Meteorites in particular are very, very consistent in showing ages between 4.4 billion and 4.6 billion years by up to six different methods. Because they have very simple geological histories and have not been subjected to weathering, erosion, and metamorphic and tectonic processes found on Earth, there is little or no scope for “evolutionists” to fish around for alternative interpretations in order to get the results they want, as creation.com claims to be standard practice. Furthermore, these data sets are relatively small (a few hundred or so samples), so allegations of cherry-picked results are completely unrealistic.

2. Showing that one method fails under specific conditions does not prove that all methods fail everywhere.

There are over forty different isotopes used in radiometric dating. Each of them has a different half-life, applies to a different range of ages, involves different experimental procedures, and works in different situations on some minerals but not others.

Let’s say, for example, that you tried to use carbon-14 dating on a traffic cone, and got an age of 5,000 years. This is quite a plausible result, because traffic cones are made of plastic, which has been heavily processed from oil deposits and will therefore have been heavily contaminated by significant amounts of modern carbon. However, what does this tell us? Simply that carbon-14 dating doesn’t work on traffic cones. It doesn’t tell us whether or not it works on bones, wood, or ancient seeds. And it certainly doesn’t tell us whether or not uranium-lead dating works on zircon crystals in granites.

Before radiometric dating techniques can be used in the field on new types of materials, they are first tested against samples of those materials whose ages are already known. Many of Woodmorappe’s discordances you read about are the results of these preliminary experimental tests. It is dishonest to cite these preliminary tests as evidence that field results using well-established methods are also unreliable.

3. Showing that a method fails when pushed to its limits does not prove that it fails everywhere.

Take, for example, this report of an attempt to date some lava from a lava dome that formed on Mount St Helens in 1986. The potassium-argon “ages” reported ranged from 0.35 million years to 2.8 million years, which the YEC researcher, Steve Austin, described as “preposterous.” In reality, it was his methodology that was preposterous.

The dating was carried out by Geochron Laboratories of Cambridge, Massachusetts. At the time that the research was carried out, their website and their promotional material said this:

Samples less than 5 M.Y. old, or containing less than 0.1%K will incur a 50% surcharge, reflecting the special care and additional analyses required. We cannot analyze samples expected to be younger than 2 M.Y.

Geochron Laboratories did not have the advanced state-of-the-art equipment needed to process samples that young, and consequently, contamination and “memory effects” of lingering argon from previous analyses would have been a very real issue.

The half-life of 40K is 1.25 billion years — roughly a thousand times greater than the ages reported by Austin. This study was the equivalent of using a weighbridge to measure out flour, sugar and eggs when baking a cake for a family of four, then claiming that weighbridges are unreliable when it comes out all mushy.

I’m sorry, but this is called “gaming the system,” pure and simple.

Another example here is the RATE project’s claims of discovery of primordial radiocarbon in ancient coals and diamonds. The quantities discovered were very low, close to the limits of the capabilities of many radiocarbon labs, and showed clear patterns that were consistent with contamination, with heavily processed samples showing significantly more 14C than unprocessed ones. Although the RATE team claimed to have taken great care to take contamination into account, radiocarbon experts such as Kirk Bertsche noted that the procedures they followed were incorrect.

4. Discrepancies of a factor of two or three do not prove that all methods must be out by a factor of a million or more.

About three quarters of the discrepancies on Woodmaroppe’s list come within a factor of two of the expected result, about 90% within a factor of three, and more than 97% within a single order of magnitude. Furthermore, about two thirds of the time, the dates reported are too small. This does not help the Literal Six Day Young Earth timescale, which requires the dates reported to be consistently too large by a factor of between a thousand and a million.

A factor of a million is a colossal error. It is like demonstrating that the whole of London would fit into a rucksack, or that you could buy a four bedroom house in Chelsea for five pounds, or that the life and ministry of Christ recorded in the Gospels happened in the space of fifteen minutes just yesterday afternoon.

A minority of discrepancies of a factor of two or three, the majority of them underestimates, falls far, far short of demonstrating that all results are consistently overestimates by a factor of up to a million. Especially when you consider that…

5. The differences are unsurprising, well understood, and useful.

Different results from different radiometric methods are not unexpected, given that many rocks have had a complex history. In fact, the differences usually provide a lot of useful information that extends well beyond the ages of the samples.

One condition which results in different dating methods giving different results is when a rock formation such as granite takes a long time to cool. This is because radiometric dating measures the time since the rock cooled below the “closure temperature.” Closure temperatures can be determined experimentally and are different for different radioisotopes. Consequently the differences in radiometric ages can tell us how long the rock took to cool.

Joe Meert gives one example of a rock formation, the Carion pluton, in central Madagascar, which yielded different dates as follows:

Dating method Closure temperature Age
Zircon–U-Pb (SHRIMP) 850±50°C 532.1±5.2 Ma
Hornblende — 40Ar/39Ar 500±50°C 512.7±1.3 Ma
Biotite — 40Ar/39Ar 350±50°C 478.9±1.0 Ma
K-spar — 40Ar/39Ar 200±25°C 435.0±10 Ma
K-spar — 40Ar/39Ar 100±25°C 410±10 Ma

When plotted on a graph, it can be seen that these give a rather nice cooling curve:

https://i2.wp.com/users.clas.ufl.edu/jmeert/carvoo11.jpg

Far from falsifying radiometric dating, different results from different methods provide additional information about a rock formation’s thermal history as well as its age. This extra information is of particular importance in oil exploration, where geologists need to know the thermal history of the oil deposits as well as their ages in order to determine whether they’re going to yield anything useful.

Woodmorappe claimed, in response to a critique to his list, that cooling explanations are just a “rationalisation,” and that the imbalance between overestimates and underestimates was evidence of cherry-picking by the scientific community, with “nonsensical” results more likely to be withheld from publication. We shall see next week that his allegations of cherry-picking have no merit whatsoever, but for now it should be clear that these explanations are no mere “rationalisation,” but are in fact individually testable.

6. Some claims of discordant dates are blatantly dishonest.

One of the articles that Woodmorappe cited was that of a Rb-Sr isochron that allegedly gave an age of 34 billion years — seven times the age of the earth and two and a half times the age of the universe.

This sounds like a massive poke in the eye for isochron dating until you see the “isochron” in question:

figure6
Rb-Sr “isochron” from the diabase of the Pahrump Group interpreted by Woodmorappe as giving an age of 34 billion years. Source: talkorigins.org

This is not an isochron. Anyone who knows anything about isochron dating will be aware that you need all the points to lie on a straight line in order to get a valid date. When they don’t, the only correct interpretation of the graph is “undateable due to contamination and leakage,” and the 1.09 billion and 34 billion year lines were purely there to illustrate the fact.

For what it’s worth, this is a variant on what scientists call “quote mining” — taking a particular quote completely out of context and misrepresenting it as saying something that it is not. A charitable view here would be merely that Woodmorappe had misunderstood the graph, wasn’t getting his facts straight, or was perhaps being a little bit careless, but scientists who get quote mined do not take such a charitable view. They consider it to be a form of lying.

Discordant dates do not prove what YECs claim that they prove.

There is a vast difference between showing that something doesn’t always work and showing that it never works, and similarly, there is a vast difference between showing that something can occasionally be out by a factor of two, and showing that it is consistently too large by a factor of up to a million.

Disagreements between different dating methods may give us leave to take individual results with a pinch of salt. This may have some bearing on subjects such as the Shroud of Turin, or the occasional archaeological discovery relating to Bible times. But they present no challenge whatsoever to the overall corpus of data as a whole. In particular, they fall far, far short of falsifying the geological column with its display of a general progression of species through the ages, let alone demonstrating that all radiometric results could plausibly be out by up to six orders of magnitude. Radiometric dating still presents an insurmountable barrier to the young-earth timescale and the discordances highlighted by the young-earth literature barely put a dent in it.

How much evidence is there?

Many debates about creation and evolution fail to appreciate the staggering amount of evidence that there is out there. Joel Duff, a Christian biology professor at the University of Akron in Ohio, has written about this on his blog, The Natural Historian, which examines the evidence from the fossil record in some detail, along with young-earth creationist responses to it. He points out, for example, that:

Estimates such as these are examples of Fermi problems. By coming up with some educated guesses (or known values) for the relevant quantities, we can then come up with an estimate for, say, the amount of kinetic energy in a hurricane. Overestimates and underestimates usually cancel each other out, and the end result will typically be well within an order of magnitude of the true figure.

How much radiometric data is there in the scientific literature?

It’s possible to estimate this value in a similar way.

First, how many geologists are there in the world? According to this blog post, about 500 geology students graduate in the USA every year. In China, on the other hand, there are probably about 40,000-50,000 geology students at any one time. This means that China must be graduating about 10,000-15,000 geologists every year.

I shall assume that the figures for the rest of the world per head of population are similar to those of the USA. Since the USA’s population is about 320 million and the rest of the world’s population is about 6 billion, this would suggest that about 10,000 further geology students graduate every year. Thus, there must be about 20,000-25,000 new geology graduates a year.

Now let’s assume that just one in ten of these graduates end up working as professional geologists, and that their average career length is about 40 years. This means that there would be about 80,000-100,000 geologists worldwide.

So how many radiometric dating results would they be churning out? In the first five years of his career, young-earth creationist geologist Andrew Snelling is listed on Wikipedia as having published an average of one paper per year in the scientific literature, most of them with one or two collaborators. Your average geologist is probably more productive than that, but I shall assume that this figure is fairly typical. I shall also assume that each publication in the literature contains an average of one radiometric result — some will contain more than one, but others will not contain any.

80,000 geologists, working in teams of an average of 2.5, publishing an average of one radiometric result per year, would mean about 32,000 new results per year. This means that in the last thirty to forty years or so, about a million new radiometric results could easily have been published.

Checking this estimate with Google Scholar

After writing the first draft of this post, it occurred to me that there is another way to estimate how much radiometric data there is in the literature: Google Scholar. We can do some simple searches for the most common radiometric dating techniques such as K-Ar, U-Pb, Rb-Sr, and so on.

A search for “K-Ar” (excluding citations and patents) returns about 360,000 results, with about 4,990 in 2016 alone. A search for “U-Pb” returns about 503,000 results, with about 11,200 in 2016. A search for “Rb-Sr” returns about 217,000 results, with about 6,170 in 2016. Even allowing for duplicates and false positives, it seems from these searches that my Fermi estimate came within a factor of two or so of the actual end result. It’s quite clear that we’re talking about hundreds of thousands of results, with tens of thousands of new ones being added every year.

Next time, we’ll ask a related question: how reliable are these hundreds of thousands of results?

This is what trying to fit science to your worldview looks like

Two weeks ago, we saw that the evidence for the age of the earth can not be based on presupposition or worldview because this does not account for evidence from the oil industry, which has to remain strictly independent of such matters for purely financial and economic reasons.

Nevertheless, young-earth creationists still tell me that science must “fit Scripture.” By this, they mean their young-earth interpretation of Scripture, because when it comes to the age of the earth, Scripture leaves a lot open to interpretation. But if you want to interpret science to fit into a six thousand year timescale, we know exactly what that involves, for the YEC organisations have spent vast sums of money attempting to do precisely that.

The RATE project.

RATE stands for Radioisotopes and the Age of The Earth. It was a research project run by the Institute for Creation Research and the Creation Research Society from 1997 to 2005, which sought to answer the question: if the earth is six thousand years old, why does radiometric dating consistently give results that are much older? It was funded by $250,000 from the Institute for Creation Research, and more than $1 million in donations.

The RATE team admitted that there is overwhelming evidence that hundreds of millions of years’ worth of nuclear decay has taken place since Creation. This was certainly ground-breaking for young-earth claims, because they had always insisted up until that point that “evolutionists” were throwing away any results that didn’t conform to their “uniformitarian presuppositions” — in effect, claiming that radiometric dating was no more than a fraudulent exercise in cherry-picking random numbers.

Rather than admitting that this meant the earth was old, however, they decided that nuclear decay rates must have been up to a billion times higher in the past — notably, during the first two days of Creation Week, and during Noah’s Flood.

22,000°C? Really?

The RATE team set about looking for evidence to support their hypothesis, and after eight years they came up with four studies that they claimed did so. I will examine the details of these claims in future blog posts, but for now all you need to know is that no independent peer reviewer has ever found their claims satisfactory, no other researchers have ever replicated their findings, and the subject matter experts who reviewed their work, many of whom were evangelical Christians themselves, documented numerous flaws of a purely technical nature, many of them serious, and all of them, when combined together, more than sufficient to falsify their entire conclusions.

But it was the RATE team themselves who noted the biggest problem of the lot. Radioactive decay generates a lot of heat. So much heat, in fact, that if you squeezed all the observed decay into just a fraction of six thousand years, you would raise the temperature of the earth to 22,000°C.

Just let that sink in for a minute.

Twenty. Two. Thousand. Degrees. Centigrade.

They admitted that neither conduction, nor convection, nor radiation, could have removed this much heat fast enough, and furthermore that any cooling process would have had to cool granites much faster than water to avoid the oceans from freezing over. They speculated about one or two possible solutions to the problem, but these also involved proposing equally absurd new laws of physics which are not supported by any evidence whatsoever either. In the end, they said that the removal of heat must have been supernatural, but admitted that the problem was nevertheless unsolved.

Despite this extraordinary contradiction, the RATE project has been marketed as a success, with books, DVDs, and conferences proclaiming accelerated nuclear decay as a rock-solid fact, supported by “overwhelming” evidence. The heat problem is casually dismissed as merely a minor issue — if in fact it ever gets mentioned at all. Accelerated nuclear decay is now touted as the cornerstone of the YEC objections to radiometric dating.

The crown jewels of young-earth research?

If you’re a young-earth creationist, struggling to figure out why even many born-again, Spirit-filled Christians find the LSDYEC position an embarrassment despite having a high regard for the Bible, this is why. The sheer absurdity of the RATE team’s willingness to downplay the 22,000°C heat problem is mind-boggling. Reading the conclusion of the RATE report, it is almost impossible to believe that it is genuine creationist apologetics, and not some sort of wild atheist parody like the Flying Spaghetti Monster or Last Thursdayism. Even one young-earth creationist friend I spoke to about it thought it was some kind of hoax to “discredit creationism.” 22,000°C is four times as hot as the surface of the sun, and hot enough to vaporise the entire planet.

But this is no parody, nor is it the idea of some fringe armchair physicist, nor can it be excused as shoestring research that was only flaky through being under-funded. As far as young-earth research goes, this is about as mainstream as it gets. The RATE project was the most expensive, extensive and thorough research project ever carried out by YEC scientists. They threw eight years, six of their best physicists and geologists, and more than a million dollars at the problem. It has been endorsed by all the major YEC organisations. It represents the very crown jewels of young-earth science.

This is what happens when you try to make scientific evidence fit conclusions that it does not support. The consensus of professional geochronologists is that the earth is 4.54±0.05 billion years old, and any challenge to that figure needs to be based on sound methodology, honest reporting of results, and careful scrutiny through a process of independent peer review. It simply isn’t honest to make up science fiction to try and fit the conclusions you want to get. It does not honour God and it does not uphold the Bible. On the contrary, it makes a mockery of it.