Exploring the wonders of geology in response to young-Earth claims...

Never been here? Please read my guidelines and background posts before proceeding!

Saturday, April 23, 2011

Radiometric Dating Recap: a response to Mike Riddle

"Δεινον δ'εστι η μη 'μπειρια..." -Αριστοφανης

Does radiometric dating prove the Earth is old? Answers in Genesis author Mike Riddle invited readers to skepticism in an article responding to this very question. Therein, he demonstrated how easily one may cast doubt on conventional interpretations of model ages from radiometric dating techniques—at least for those unfamiliar with the process and typical results. How did he accomplish this goal?

1. Assert that a "straightforward reading" is the only proper approach to Scripture, and designate any deviation from this a compromise on the veracity of God's word.

2. Offer a simplified description of how radiometric dating works in the most ideal case.

3. Remind readers of the assumptions behind "model ages" (without letting them know what a "model age" actually is).

4. Offer anecdotal evidence in which model ages are in conflict.

5. Assure readers that "conventional" explanations for the discordance have been thoroughly ruled out.

6. Conclude that the only reasonable explanation for discordance among unpublished data—which represent a tiny fraction of results from the world's geochronologists—is a past, unquantifiable change in the rate of nuclear decay in radioactive elements.

7. Divert readers from the obvious heat problem (sections 3–5; Isaac, 2007) associated with this physical model of Earth history by deeming it "a new and exciting opportunity for creation research."

I understand that Mr. Riddle's intention is to convey, in popular terms, why he does not accept published ages, and I sympathize with his desire to find concordance between God's word and creation. Having an obligation to the truth, however, I feel it necessary to comment that his approach is misleading, particularly to those inexperienced in geochronology.

1. What is a straightforward reading?

In the past 150 years, scholars have uncovered a wealth of information regarding the cultural and literary world in which the Genesis narrative was drafted. The theological implications of these findings have been debated on all sides, and such continues today. But despite the lack of consensus, it is evident that a "straightforward reading" is more elusive than originally proposed. Moreover, recent perspectives on literary criticism show that the reader's own culture and environment play an equally important role in the interpretation of ancient texts.

While I affirm the perspicuity of Scripture (its theological message can be understood properly across time and culture), I think Mike Riddle concludes prematurely that a "straightforward" reading reveals the Earth is actually thousands of years old. I have no doubt that our post-Enlightenment mentality has crept into this conclusion unnoticed, causing believers and unbelievers alike to read scientific and historical details back into the ancient text. Mr. Riddle's surface-level reading of the Genesis narrative is not sufficient reason to dismiss a priori the conventional interpretations of radiometric dates.

2. Any meaningful critique of radiometric dating must take into account the complexity behind analyses and interpretations, as well as the wide range of geological applications.

Simply put, radiometric dating is not just a method used to date igneous rocks using an hourglass model of parent/daughter isotopes. Nearly as many analyses are now performed on both metamorphic and sedimentary rocks, and the respective methods for all three systems are quite complicated (usually involving 3 or more isotopes). Mr. Riddle overlooks this point, presumably to cast doubt on the age of fossils contained within sedimentary rocks (which, he seems to believe, cannot be dated directly).

But for those interested, metamorphic histories are commonly reconstructed by dating minerals that differ in closure temperature, and/or minerals that form during metamorphism (such as garnet). Multi-domain diffusion models (constructed from 40Ar/39Ar age spectra) are used to interpret thermal histories after crystallization. Combined, these techniques offer a powerful tool to investigate mountain building processes, since they tell us when the rock was at a given temperature and pressure.

Sedimentary rocks commonly contain authigenic minerals (as well as early-stage cements like calcite) that can be dated individually. Such minerals are more susceptible to alteration than in igneous systems, but a bulk of the data are consistent with the conventional geologic timeline. This point is particularly relevant, because Mike Riddle (following the RATE team) interprets long ages of igneous bodies as a product of accelerated nuclear decay during the Flood. If decay rates after the Flood were closer to modern measurements, however, then diagenetic cements should yield very young (indistinguishable from zero) ages. Since they don't, the young-Earth model cannot currently explain the range of available data.

Detrital zircons and micas can also be dated individually to constrain the age of a sedimentary rock. Since these minerals are inherited from igneous rocks that have already crystallized, it is understood that their ages will always be older than the sedimentary rock itself. The youngest ages of detrital zircons/micas, therefore, give the maximum age for deposition. This technique may also reveal the main source of sediments. If a majority of zircons, for example, are about 55 m.y. old, then one could look for the nearest igneous/metamorphic body that dates to 55 Ma to find the primary sediment source.

3. Every radiometric date represents a model age. Nobody claims that model ages actually 'prove' the age of anything. "Model ages" are termed such because they rely on a scientific model. If the physical conditions and various assumptions within the model did not hold for a given sample, then the model age does not equal the true age of the rock. Moreover, all model ages in igneous systems represent cooling ages—not necessarily the age of crystallization. Slowly cooling or reheated rocks yield different ages for different minerals (keep in mind that isochrons constructed from several minerals assume those minerals reached their respective closure temperatures at the same time).

Finally, model ages do not prove the antiquity of rocks, because a history with uniform natural laws is already assumed within the model. So yes, model ages are contingent on the uniformity of nature and assumptions about the rock's physical history.

On the other hand...

4–5. The overwhelmingly consistent results from radiometric dating do highly corroborate the interpreted history of geological features (i.e. demonstrate that the interpreted history did in fact take place, or else the data were specifically designed to give this illusion). If this were not the case, the RATE team (referenced by Mike Riddle) would not have resorted to accelerated nuclear decay as a means to explain long ages. Instead, they would continue to cite discordant age data and geochronologists would be out of business. The fact that thousands of researchers spend millions of dollars each year to date rocks should provide sufficient reason to believe that a vast majority of radiometric dates are concordant.

Another way is to search through scientific literature oneself, or speak directly to a lab manager. I've done both, and I am happy to tell you that radiometric dating works. Within the article, however, Mike Riddle provides a number of tables with results from the RATE team. It's apparent that model ages are not always concordant (agree with each other), so what are we to make of these results?

Young Volcanic Rocks
Mr. Riddle begins by citing cases where young volcanic rocks do not yield "zero" ages. I've addressed this topic at length before, so I will only mention that the apparent problem has been known for more than 40 years. Brent Dalrymple, who invented the K-Ar dating method, discovered early on that several historical lava flows (not most, not even a majority, but some) contained sufficient radiogenic argon to give dates that were too old (~30,000–500,000 years). He predicted that the lava flows contained material inherited from older rocks (such as in fluid inclusions or microscopic xenoliths), but was unable to test this prediction. As technology improved, his hypothesis was confirmed, and young volcanics are now dated by more sophisticated methods (40Ar/39Ar, electron-spin resonance, thermoluminescence) that need not assume the rocks were originally argon-free.

But these findings have not stopped ICR researchers like Steve Austin and Andrew Snelling from spending thousands of dollars on rigged, radiometric dating games. For example, samples from Mt. St. Helens were shipped to a laboratory that openly stated their technology could not detect argon levels in rocks less than 2 million-year-old. Moreover, the only minerals that yielded ages distinguishable from zero were ferromagnesian silicates, which likely crystallized before the eruption (more info here). When historical lava flows are dated using K-Ar isochrons, 40Ar/39Ar, and other methods, the ages are indistinguishable from zero.

Isochron Dating
In the first table (Beartooth Mountains), 4 K-Ar ages are listed along 4 isochron ages. Mr. Riddle concludes that "the results show a significant scatter in the ages for the various minerals and also between the isotope methods." To say that scatter is significant, however, requires some knowledge of the statistical variance for each data set, and uncertainties are not provided here. In fact, the interpreted age from each isochron is within statistical uncertainty of the published age (2,790 Ma), so there is no demonstrable discordance.

Nonetheless, Mr. Riddle exposes his unfamiliarity with such data by noting that "in some cases, the whole rock age is greater than the age of the minerals, and for others, the reverse occurs." What does this mean? In the former case (Rb-Sr isochron), the difference is found in the population (number of samples) for each isochron age. One is built from 5 points; the other from 30. The reverse is true for K-Ar dates, because some minerals do not retain argon as well as others. It is not unexpected that the "Quartz-plagioclase" mineral yielded a younger date than either biotite or hornblende, given the lower retentivity of argon and higher susceptibility to alteration or thermal disturbance.

Recently, I took a closer look at the RATE team's treatment of isochron plots in the case of Precambrian sills from the Grand Canyon. In short, K-Ar data (including those from Austin and Snelling) are consistent with the accepted Rb-Sr isotope age of 1103 Ma. The variation in K-Ar dates listed in Mike Riddle's table (Bass Rapids Sill Sample Results) reflects the complex thermal history of the rocks, which have also been highly altered. In other words, the physical assumptions of each model age were not met. Alternative approaches demonstrate that these assumptions (no loss of daughter element; no gain of parent element) were falsified for the K-Ar method, and give a more clear and consistent estimate of the initial cooling age (of course, Mr. Riddle does not cite all the age data available).

Two other isochron methods (Pb-Pb and Sm-Nd) apparently gave slightly older dates for the Grand Canyon sills (~1.3 Ga), but there are a few suspicious features about the data from Snelling and Austin (2003; published here). First, this age is identical to that of the basement rock through which the intrusive magma flowed. Mr. Riddle cites isotopic mixing as a possible explanation for the discordance but claims that it was ruled by the authors. Such is not the case, however, and previous workers have interpreted the geochemistry of the sills to reflect incorporation of the country rock. In the case of Pb-Pb isochrons, a false isochron may be constructed from mineral samples that formed at very different times, but were part of an isotopically homogenous reservoir at one time. The false isochron age indicates the last point at which the samples were in isotopic equilibrium (in this case, about 1.3 Ga—the age of the country rock).

Secondly, the results of each isochron are statistically imprecise. In fact, the 1249 million-year Pb-Pb isochron age is actually within statistical uncertainty (±140 Ma) of the conventional age of the formation! The Sm-Nd age is beyond 2σ from the conventional, but the uncertainty (±170 Ma) is still unreasonably high. Such imprecision in the calculated isochron age is likely due to the fact that the samples were not truly cogenetic. In other words, they don't form a true isochron (this is most obvious in the Sm-Nd isochron). The major element geochemistry is variable enough between samples to suggest that significant fractional crystallization occurred. Hydrothermal circulation undoubtedly played a role in the rock's history as well. Finally, both isochrons are heavily weighted by felsic, granophyre samples from the top of the sill (which are moderately altered), so the discordance of the isochron ages is circumstantial at best.

6. Forced concordance on the RATE team's data sets causes discordance in the majority of geochronological data. Even giving Austin and Snelling the benefit of the doubt (i.e. ignoring uncertainties and problems arising from alteration), calling for accelerated nuclear decay in Earth history does nothing to solve the discordance. How so?

The reason is that a vast majority of isochron ages are, in fact, concordant. If we suppose that the decay rate of Sm was accelerated more than that of Pb, which was accelerated more than that of Rb, which was accelerated more than that of K, we can manipulate the Grand Canyon data so that they yield the same model age. But this mathematical "fix" would cause nearly every other published age to become suddenly discordant. Thus the RATE team's outrageous proposal would reduce nearly all geochronological data to absurdity for the sake of a handful of samples, but neither Mike Riddle nor the RATE team have been explicitly clear on this point.

7. We can be fairly confident that nuclear decay rates never changed. The first reason is that any major increase in radioactive decay would have left noticeable marks on the planet. Nuclear decay produces heat (the basic premise behind nuclear power plants). Faster decay would produce proportionally more heat. In the most conservative case for the RATE team, 1.1 billion years worth of decay occurred in the Grand Canyon sills during or since the Flood (let's say 5,000 years). That's a 220,000-fold increase in decay rates, on average, over the past 5,000 years, which would have produced enough heat to destroy all life on Earth (as well as the hydrosphere).

Dr. Larry Vardiman at ICR has considered this problem publicly, and rejects the conclusions of the RATE team. I understand that Mike Riddle, along with the RATE team, is confident that a solution will be found (i.e. how to dissipate enough heat to vaporize the planet, and then devise a sound reason behind the arbitrary premise). In the meantime, however, the problem should be stated more explicitly, especially to his lay readership.

Finally, Mike Riddle cites the helium diffusion study of Dr. Russell Humphreys, which I reviewed here, as the clear scientific reason for believing in accelerated nuclear decay. A close look at Dr. Humphreys' tactics shows that he not only employed bad scientific practice, but espoused unwarranted confidence in the results. Dr. Humphreys and others have never repeated the results of this decade-old experiment, but will often remind us that real science demands replication of results.

In the meantime, hundreds of other researchers have taken advantage of helium diffusion in zircon as a means to date exhumation (uplift) events in sedimentary rocks and igneous plutons. Yet not a single one yields a 6,000-year age. Ages based on helium diffusion in zircon are commonly consistent with the conventional geologic timescale, and falsify Dr. Humphreys' hypothesis thoroughly.


Does radiometric dating prove the Earth is old? Well, no, in a strict philosophical sense. But it does highly corroborate the conventional understanding of Earth history and the geologic timescale. Moreover, it falsifies the young-Earth interpretation of Earth history on every point. Despite their vested efforts over several decades, members of the RATE team have not been able to explain the range of geochronological data in a young-Earth paradigm. Nor have they been able to discredit the published results of geologists. Mr. Riddle's closing comment that "radiometric dating methods are highly unreliable" is not convincing to those familiar with the process and results, because it simply does not correspond to reality.

"Inexperience is a dreadful thing..." -Aristophanes


  1. Good post. Regarding Ar-Ar dates, we generally spoke of closing temperatures: For Hornblende, it is about 500°C, and for biotite, 300°C. Thus if the rock gets re-heated beyond 300, but not reaching 500 (ie low grade metamorphism), you'll get disparate ages between the minerals.

    But what these RATE folks ignore is that in a vast number of cases, there is reasonable correlation between radiometric dating by a wide array of methods. For insyance, I've seen studies of intrusions that were dated by different people from different laboratories by different methods, all coming within 2σ error - this using methods like U-Pb SHRIMP dating of perovskite, Rb-Sr dating of phenocrystic phlogopite, as well as Ar-Ar step heatingof groundmass phlogopite (the phenocrystic and groundmass phlogopite have different histories). Such correlation, which is by no means isolated, undermines the doubt that RATE and others attempt to sow.

  2. Yes, I've noticed that closure temperature is rarely taken into consideration, although the RATE team spends most of its efforts discussing discordant K-Ar and Ar-Ar dates.

    Thanks for your insight. It's nice to have a voice of experience here to confirm the concordance of these dates.