|Jet in Carina WFC3 IR [Courtesy NASA]
How reliable is geologic dating?
David H. Bailey
Updated 31 March 2019 (c) 2019
In a related article on geologic ages
we presented a chart with the various geologic eras and their ages. In a separate article
we sketched in some technical detail how these dates are calculated using radiometric dating techniques. As we pointed out in these two articles, radiometric dates are based on known rates of radioactivity, a phenomenon that is rooted in fundamental laws of physics and follows simple mathematical formulas. Dating schemes based on rates of radioactivity have been refined and scrutinized for several decades. The latest high-tech equipment permits reliable results to be obtained even with microscopic samples.
Radiometric dating is self-checking, because the data (after certain preliminary calculations are made) are fitted to a straight line (an "isochron") by means of standard linear regression methods of statistics. The slope of the line determines the date, and the closeness of fit is a measure of the statistical reliability of the resulting date. Technical details on how these dates are calculated are given in Radiometric dating. Here is one example of an isochron, based on measurements of basaltic meteorites (in this case the resulting date is 4.4 billion years) [Basaltic1981, pg. 938]:
Reliability of radiometric dating
So, are radiometric methods foolproof? Just how reliable are these dates?
As with any experimental procedure in any field of science, these measurements are subject to certain "glitches" and "anomalies," as noted in the literature. Skeptics of old-earth geology make great hay of these examples. For example, creationist writer Henry Morris [Morris2000, pg. 147] has highlighted the fact that measurements of specimens from a 1801 lava flow near a volcano in Hualalai, Hawaii gave apparent ages (using the Potassium-Argon method) ranging from 160 million to 2.96 billion years, citing a 1968 study [Funkhouser1968]. In the particular case that Morris highlighted, the lava flow was unusual because it included numerous xenoliths (typically consisting of olivine, an iron-magnesium silicate material) that are foreign to the lava, having been carried from deep within the Earth but not completely melted in the lava. Also, as the authors of the 1968 article were careful to explain, xenoliths cannot be dated by the K-Ar method because of excess argon in bubbles trapped inside [Dalrymple2006]. Thus in this case, as in many others that have been raised by skeptics of old-earth geology, the "anomaly" is more imaginary than real. Other objections raised by creationists are addressed in [Dalrymple2006a].
The overall reliability of radiometric dating was addressed in some detail in a recent book by Brent Dalrymple, a premier expert in the field. He wrote [Dalrymple2004, pg. 80-81]:
These methods provide valid age data in most instances, although there is a small percentage of instances in which even these generally reliable methods yield incorrect results. Such failures may be due to laboratory errors (mistakes happen), unrecognized geologic factors (nature sometimes fools us), or misapplication of the techniques (no one is perfect).
We scientists who measure isotope ages do not rely entirely on the error estimates and the self-checking features of age diagnostic diagrams to evaluate the accuracy of radiometric ages. Whenever possible we design an age study to take advantage of other ways of checking the reliability of the age measurements. The simplest means is to repeat the analytical measurements in order to check for laboratory errors. Another method is to make age measurements on several samples from the same rock unit. This technique helps identify post-formation geologic disturbances because different minerals respond differently to heating and chemical changes. The isochron techniques are partly based on this principle.
The use of different dating methods on the same rock is an excellent way to check the accuracy of age results. If two or more radiometric clocks based on different elements and running at different rates give the same age, that's powerful evidence that the ages are probably correct.
Along this line, Roger Wiens, a scientist at the Los Alamos National Laboratory, asks those who are skeptical of radiometric dating to consider the following (quoted in several cases from [Wiens2002]):
- There are well over forty different radiometric dating methods, and scores of other methods such as tree rings and ice cores.
- All of the different dating methods agree--they agree a great majority of the time over millions of years of time. Some [skeptics] make it sound like there is a lot of disagreement, but this is not the case. The disagreement in values needed to support the position of young-earth proponents would require differences in age measured by orders of magnitude (e.g., factors of 10,000, 100,000, a million, or more). The differences actually found in the scientific literature are usually close to the margin of error, usually a few percent, not orders of magnitude!
- Vast amounts of data overwhelmingly favor an old Earth. Several hundred laboratories around the world are active in radiometric dating. Their results consistently agree with an old Earth. Over a thousand papers on radiometric dating were published in scientifically recognized journals in the last year, and hundreds of thousands of dates have been published in the last 50 years. Essentially all of these strongly favor an old Earth.
- Radioactive decay rates have been measured for over sixty years now for many of the decay clocks without any observed changes. And it has been close to a hundred years since the uranium-238 decay rate was first determined.
- A recent survey of the rubidium-strontium method found only about 30 cases, out of tens of thousands of published results, where a date determined using the proper procedures was subsequently found to be in error.
- Both long-range and short-range dating methods have been successfully verified by dating lavas of historically known ages over a range of several thousand years.
- The mathematics for determining the ages from the observations is relatively simple.
Rates of radioactivity
One question that sometimes arises here is how can scientists assume that rates of radioactivity have been constant over the great time spans involved. Creationist Henry Morris, for example, criticizes this type of "uniformitarian" assumption [Morris2000, pg. 91]. But numerous experiments have been conducted to detect any change in radioactivity as a result of chemical activity, exceedingly high heat, pressure, or magnetic field. None of these experiments has detected any significant deviation for any isotope used in geologic dating [Dalrymple1991, pg. 86-89; Dalrymple2004, pg. 58-60].
Scientists have also performed very exacting experiments to detect any change in the constants or laws of physics over time, but various lines of evidence indicate that these laws have been in force, essentially the same as we observe them today, over the multi-billion-year age of the universe. Note, for instance, that light coming to Earth from distant stars (which in some cases emanated billions of years ago) reflects the same patterns of atomic spectra, based in the laws of quantum mechanics, that we see today. What's more, in observed supernova events that we observe in telescopes today, most of which occurred many millions of years ago, the patterns of light and radiation are completely consistent with the half-lives of radioactive isotopes that we measure today [Isaak2007, pg. 200]. As another item of evidence, researchers studying a natural nuclear reactor in Africa have concluded that a certain key physical constant ("alpha") has not changed measurably in hundreds of millions of years [Barrow2007, pg. 124-128]. Finally, researchers have just completed a study of the proton-electron mass ratio (approximately 1836.1526), and found that it has not varied more than 0.0005 percent over the history of the universe ranging back to 12.4 billion years ago [Srinivasan2016].
Thus scientists are on very solid ground in asserting that rates of radioactivity have been constant over geologic time. The issue of the "uniformitarian" assumption is discussed in significantly greater detail at
Responses to specific creationist claims
Wiens' online article, mentioned above, is an excellent resource for countering claims of creationists on the reliability of geologic dating. In an appendix to this article, Wiens addresses and responds to a number of specific creationist criticisms. Here is a condensed summary of these items, quoted from Wiens' article [Wiens2002]:
- Claim: Radiometric dating is based on index fossils whose dates were assigned long before radioactivity was discovered.
Response: This is not at all true, though it is implied by some young-earth literature. Radiometric dating is based on the half-lives of the radioactive isotopes. These half-lives have been measured over the last 40-90 years. They are not calibrated by fossils.
- Claim: No one has measured the decay rates directly; we only know them from inference.
Response: Decay rates have been directly measured over the last 40-100 years. In some cases a batch of the pure parent material is weighed and then set aside for a long time and then the resulting daughter material is weighed. In many cases it is easier to detect radioactive decays by the energy burst that each decay gives off. For this a batch of the pure parent material is carefully weighed and then put in front of a Geiger counter or gamma-ray detector. These instruments count the number of decays over a long time.
- Claim: If the half-lives are billions of years, it is impossible to determine them from measuring over just a few years or decades.
Response: The example given in the section [in Wiens' article] titled, "The Radiometric Clocks" shows that an accurate determination of the half-life is easily achieved by direct counting of decays over a decade or shorter. This is because: (a) all decay curves have exactly the same shape (Fig. 1 [in Wiens' article]), differing only in the half-life, and (b) trillions of decays can be counted in one year even using only a fraction of a gram of material with a half-life of a billion years. Additionally, lavas of historically known ages have been correctly dated even using methods with long half-lives.
- Claim: The decay rates are poorly known, so the dates are inaccurate.
Response: Most of the decay rates used for dating rocks are known to within two percent. Uncertainties are only slightly higher for rhenium (5%), lutetium (3%), and beryllium (3%), discussed in connection with Table 1 [in Wiens' article]. Such small uncertainties are no reason to dismiss radiometric dating. Whether a rock is 100 million years or 102 million years old does not make a great deal of difference.
- Claim: To date a rock one must know the original amount of the parent element. But there is no way to measure how much parent element was originally there.
Response: It is very easy to calculate the original parent abundance, but that information is not needed to date the rock. All of the dating schemes work from knowing the present abundances of the parent and daughter isotopes. The original abundance N0, of the parent is simply N0 = N ekt, where N is the present abundance, t is time, and k is a constant related to the half life.
- Claim: There is little or no way to tell how much of the decay product, that is, the daughter isotope, was originally in the rock, leading to anomalously old ages.
Response: A good part of [Wiens' article] is devoted to explaining how one can tell how much of a given element or isotope was originally present. Usually it involves using more than one sample from a given rock. It is done by comparing the ratios of parent and daughter isotopes relative to a stable isotope for samples with different relative amounts of the parent isotope. For example, in the rubidium-strontium method one compares rubidium-87/strontium-86 to strontium-87/strontium-86 for different minerals. From this one can determine how much of the daughter isotope would be present if there had been no parent isotope. This is the same as the initial amount (it would not change if there were no parent isotope to decay). Figures 4 and 5 [in Wiens' article], and the accompanying explanation, tell how this is done most of the time. While this is not absolutely 100% foolproof, comparison of several dating methods will always show whether the given date is reliable.
- Claim: There are only a few different dating methods.
Response: [Wiens' article] has listed and discussed a number of different radiometric dating methods and has also briefly described a number of non-radiometric dating methods. There are actually many more methods out there. Well over forty different radiometric dating methods are in use, and a number of non-radiogenic methods not even mentioned here.
- Claim: A young-Earth research group reported that they sent a rock erupted in 1980 from Mount Saint Helens volcano to a dating lab and got back a potassium-argon age of several million years. This shows we should not trust radiometric dating.
Response: There are indeed ways to "trick" radiometric dating if a single dating method is improperly used on a sample. Anyone can move the hands on a clock and get the wrong time. Likewise, people actively looking for incorrect radiometric dates can in fact get them. Geologists have known for over forty years that the potassium-argon method cannot be used on rocks only twenty to thirty years old. Publicizing this incorrect age as a completely new finding was inappropriate. The reasons are discussed in the Potassium-Argon Dating section [of Wiens' article]. Be assured that multiple dating methods used together on igneous rocks are almost always correct unless the sample is too difficult to date due to factors such as metamorphism or a large fraction of xenoliths.
- Claim: Different dating techniques usually give conflicting results.
Response: This is not true at all. The fact that dating techniques most often agree with each other is why scientists tend to trust them in the first place. Nearly every college and university library in the country has periodicals such as Science, Nature, and specific geology journals that give the results of dating studies. The public is usually welcome to (and should!) browse in these libraries. So the results are not hidden; people can go look at the results for themselves. Over a thousand research papers are published a year on radiometric dating, essentially all in agreement. Besides the scientific periodicals that carry up-to-date research reports, [there are] textbooks, non-classroom books, and web resources.
As noted above, creationists make great hay out of "anomalies" in radiometric dating. It is true that some "anomalies" have been observed, although keep in mind that these have been identified by professional scientists in published literature, not by creationists or others outside of peer-reviewed scientific literature.
First of all, many of these claimed "anomalies" are completely irrelevant to the central issue of whether the Earth is many millions of years old. This is certainly true when errors are in the range of a few percent in specimens many millions of years old. This is also true of anomalies noted in carbon-14 dates. Carbon-14 dating cannot be used to date anything older than about 50,000 years, since the carbon-14 half life is only 5730 years. For additional discussion, see
In any event, it is important to keep these anomalies in perspective. For example, out of literally tens of thousands of dates measured using the rubidium-strontium dating scheme (see description of the Rb-Sr scheme in
Ages), only about 30 cases have been noted where the individual data values initially appeared to lie nearly on a straight line (as is required), but the result was later found to be significantly in error. And each of these 30 cases is fairly well understood -- none of these is truly "mysterious" [Wien2002].
Anomalies and other objections that have been raised by creationists are dealt with in detail in Roger Wiens' article
[Wien2002], Mark Isaak's book
[Isaak2007, pg. 143-157] and a 2006 online article by Brent Dalrymple
[Dalrymple2006]. A detailed response to other claims of scientific evidence for a young Earth is given by Matthew Tiscareno [Tiscareno2009].
Radioactive isotopes and the age of the Earth
Until recently, only a large scientific laboratories could afford mass spectrometers, which are the principal tool used to measure dates of rock samples. But recently the prices of these devices have dropped to levels that even amateur meteorite hunters and others can afford. Used mass spectrometers are currently available at eBay.com for as little several thousand dollars. Some have said that the last of the flat-earth believers did not give up until they could hold GPS receivers in their hand that give their latitude-longitude position. Will skeptics of old-earth geology wait until mass spectrometers are in every home before finally conceding that the Earth is older than 6000 years?
In any event, there is a simple way to see that the Earth must be at least 1.36 billion years old, which does not require any mass spectrometers, isochron graphs, calculus or statistical software (provided one accepts a few very-well-established measured rates of radioactivity). Consider the list of all known radioactive isotopes with half-lives of at least one million years but less than one quadrillion years, and which are not themselves produced by any natural process such as radioactive decay or cosmic ray bombardment [Nuclides2012]:
|Isotope ||Half-life (years) ||Found in nature? ||In-115 ||4.41 x 1014 ||yes
||Gd-152 ||1.08 x 1014 ||yes
||Ba-130 ||7.00 x 1013 ||yes
||Pt-190 ||6.50 x 1011 ||yes
||Sm-147 ||1.06 x 1011 ||yes
||La-138 ||1.02 x 1011 ||yes
||Rb-87 ||4.97 x 1010 ||yes
||Re-187 ||4.12 x 1010 ||yes
|Lu-176 ||3.76 x 1010 ||yes ||Th-232 ||1.40 x 1010 ||yes
|U-238 ||4.47 x 109 ||yes ||K-40 ||1.25 x 109 ||yes
|U-235 ||7.04 x 108 ||yes ||Pu-244 ||8.00 x 107 ||yes
|Sm-146 ||6.80 x 107 ||yes ||Nb-92 ||3.47 x 107 ||no
|Pb-205 ||1.73 x 107 ||no ||Cm-247 ||1.56 x 107 ||no
|Hf-182 ||8.90 x 106 ||no ||Pd-107 ||6.50 x 106 ||no
|Tc-98 ||4.20 x 106 ||no ||Bi-210 ||3.04 x 106 ||no
|Dy-154 ||3.00 x 106 ||no ||Fe-60 ||2.62 x 106 ||no
|Tc-97 ||2.60 x 106 ||no ||Cs-135 ||2.30 x 106 ||no
|Gd-150 ||1.79 x 106 ||no ||Zr-93 ||1.53 x 106 ||no |
(In the above chart, years are displayed in scientific notation: i.e., 1 x 106 = 1 million; 1 x 109 = 1 billion, etc.)
All of the above isotopes are readily produced in nuclear reactors, so there is every reason to believe that they were formed along with stable isotopes, in roughly the same abundance as nearby stable isotopes of similar atomic weight, when the material forming our solar system was produced in an ancient stellar explosion. A quick calculation shows that after an elapsed period of 20 times the half-life of a given isotope, the fraction 1/220 = 1/1048576 (i.e., roughly one part in one million) of the original isotope will remain, which is a small but nonetheless detectable amount. Similarly, after 30 half-lives, roughly one part in one billion will remain, and after 40 half-lives, roughly one part in one trillion will remain, which is near the current limit of detectability.
Now note that an absolutely clear-cut fact is revealed in the above table: every isotope in the list with a half life less than 68 million years is absent in nature, evidently because all traces of these isotopes have decayed away, yet every isotope with a half life greater than 68 million years is present at some detectable level. This is incontestable evidence that the material from which our Earth and solar system was formed is at least 20 Sm-146 half-lives, i.e., 20 x 68 million (= 1.36 billion) years old, and, more likely, is at least 40 x 68 million (= 2.72 billion) years old. For details, see [Dalrymple2004, pg. 202-204; Miller1999, pg. 69-72].
Radiometric dating, like any other experimental discipline, is subject to a variety of errors, ranging from human errors to rare anomalies resulting from highly unusual natural circumstances. But while errors and anomalies can occur, the burden of proof is not on scientists to fully explain each and every error. Instead, the burden of proof is on skeptics of old-earth geology to explain why tens of thousands of other carefully measured ages are all internally and externally consistent.
The creationist-intelligent design approach of highlighting a few known faults and errors in past measurements is a classic instance of the "forest fallacy" -- picking a fault or two in the bark of a single tree, then trying to claim that the entire forest doesn't exist. But the forest does exist -- literally tens of thousands of carefully peer-reviewed radiometric measurements have been made (and thousands more are published each year), using equipment and techniques that have been improved and refined over at least 50 years. This huge corpus of very well-established results cannot be so easily dismissed.
Indeed, there is no known physical phenomenon that can yield consistent results in many thousands of measurements, year after year, except one: that these specimens really are as old as the data shows them to be. As biologist Kenneth Miller has observed, "The consistency of [radiometric] data ... is nothing short of stunning." [Miller1999, pg. 76].
For additional information on radiometric dating, including detailed responses to specific issues that have raised by creationists, see: [Dalrymple1991; Dalrymple2004; Dalrymple2006; Dalrymple2006a; Isaak2007, pg. 143-157; Miller1999, pg. 66-80; Stassen1998; Stassen2005; Wiens2002].
See also the articles on this website on the ages of the geologic periods
and the "uniformitarian" assumption