Firstly: please try not to bring creation/evolution into this. I will award the correct answer to a response which provides links to evidence and sound explanations.

Motivation for the question to follow:

Some of the common mistakes we make in evaluating claims are resisting contrary evidence, looking for confirming evidence, and preferring available evidence. To counteract these tendencies, we need to take deliberate steps to examine critically even our most cherished claims, search for disconfirming evidence as well as confirming, and look beyond evidence that is merely the most striking or memorable. (see link)


I heard about a group of people (yes they happen to be creationists with an agenda, but this should be irrelevant to the question I am posing! ) who obtained some samples of rocks from a lava flow from Mt Ngauruhoe in New Zealand. They claim that the rocks they obtained were from a lava flow which came out of the volcano in 1945. They sent these rocks to 2 labs and had them dated by potassium-argon dating to be between 270, 000 and 1 million years old. (see relevant bits of the link - and please ignore all agenda-based stuff in there! )

My question:

Since the real age of the rocks was around 50 years, does this demonstrate that K-Ar dating is inaccurate? I can think of several possibilities in response to this question:

  1. K-Ar will never give a lower number than something like 200, 000 years
  2. They lying about the results
  3. They sent in the wrong rocks

2 & 3 seem easily falsifiable - anyone else could simply repeat the procedure and see if their results were the same. I haven't heard of this being done, however if you have some evidence to this effect please share it.

The answer to 1 may be what I'm after. I have very little knowledge in the field of radioactive dating, and I'm not even sure if 1 is a true statement. However if it is, then wouldnt this invalidate any results made using K-Ar dating?

Please respond with and flaws in my reasoning or any additional reasons why the experiment was flawed.

edit 1

above i said

They sent these rocks to 2 labs and had them dated by potassium-argon dating to be between 270, 000 and 1 million years old.

which i wrote based on a quick glance of table 1 in the link. however i should have read the article more carefully and written this:

They sent these rocks to 2 labs and had them dated by potassium-argon dating to be either less than 270, 000 years or up to 3.7 million years old.

(the upper limit comes from the 3.5+0.2 figure in table 1). apologies.

  • 1
    – going
    Commented Apr 14, 2011 at 6:09
  • 2
    I'm not an expert, but I remember that all those dating methods have only a certain range where they're accurate. If you're outside the range you won't get reliable results, but you can just use different isotopes that give good results in the desired range.
    – Mad Scientist
    Commented Apr 14, 2011 at 6:19
  • @Fabian: that was kinda what i was getting at with #1. but im no expert either. what would be the range of accuracy for k-ar? Commented Apr 14, 2011 at 6:21
  • 1
    Should be moved to physics?
    – Sklivvz
    Commented Apr 14, 2011 at 6:39
  • 2
    In addition to your 3 possible explanations, I'd add: 4) The conditions under which the magma formed were insufficient to "reset" the K-Ar ratio, or 5) The conditions under which the sample formed led to contamination by mixing with nearby material (Wikipedia says this is not uncommon), or 6) The samples were somehow contaminated after being collected.
    – SigmaX
    Commented Apr 2, 2012 at 20:54

6 Answers 6


Statistically significant disparity in measured ages is inconsistent with the assumptions required to make radiometric dating predictions. Therefore no prediction of the theory has been contradicted.

Drawing inferences from radiometric dating requires at least two basic assumptions:

  1. At the time the rock as formed, no daughter product was present.
  2. No daughter or parent product has entered the sample since then.

Igneous rock often forms under conditions that favor (1). But not always -- contamination of both types is common.

To falsify K-Ar dating, as the claim is attempting to do, one must not only show that the predictions of K-Ar measurements lead to incorrect years, but that both assumptions (1) and (2) hold.

The claim shows one sample that dated to 3.5 +/- 0.2 million years before present. While other answers are correct that the half-life of Potassium-40 is very large, the 0.2 MY error bars indicate that the measurements were accurate enough to establish the "age" with a high degree of certainty. The claim "the real ratio of elements would indicate a 50 year old sample" lies more than 17 standard deviations away from the mean measurement, so it is very unlikely that the sample matched that prediction.

But here's a problem. The same flow yielded samples that dated to 1.3 +/- 0.3 MYA, 0.8 +/- 0.2 MYA, and 1.2 +/- 0.2 MYA. These are also significant numbers. (One sample also dated to less than 270,000 years ago, which appears to be the minimum measurable value, so is not a significant aberration).

Radiometric dating would predict that, if the assumptions (1) and (2) hold, samples from the same flow would have the same age. But this is not what is observed, so the theory rejects (1) and/or (2). Therefore, radiometric dating does not predict that the rocks would date accurately.

Therefore this data cannot be used to falsify K-Ar dating, because it does not violate a prediction of radiometric dating. Instead, the hypothetical contamination scenarios proposed by other answers/comments gain support.

Other methods, such as Isochron dating could potentially be used to show that the data are still consistent with current geological theory.

Note: This answer has been completely rewritten. I wrote the first draft thinking the entire range of reported dates represented the error bars on one sample, implying that the perceived dating error was not statistically significant. I later realized that this was not the case.

  • 4
    this is the best answer. it goes directly in line with bertrand russell's three points on expert opinion from "the will to doubt": (1) when the experts are agreed, the opposite opinion cannot be held to be certain; (2) when they are not agreed, no opinion can be regarded as certain by a non-expert; and (3) when they all hold that no sufficient grounds for a positive opinion exist, the ordinary man would do well to suspend his judgment. where (2) is applicable here. Commented Apr 5, 2012 at 0:20
  • 1
    I'm coming in late, but I don't understand this argument. Isn't it the No True Scotsman fallacy? If the sample gathered by these skeptics cannot be measured accurately (allegedly due to assumption violations), how can we trust results given on other samples?
    – Oddthinking
    Commented Mar 3, 2014 at 7:00
  • 1
    @Oddthinking: Dismissing claims of anomalous data is a No True Scotsman if it is done in a cavalier fashion. In the philosophy of science, this danger is described by the Durhem-Quine Thesis (en.wikipedia.org/wiki/Duhem%E2%80%93Quine_thesis). Since all data carries the possibility of error, the solution is not to "pick and choose" what data you deem reliable, but to use converging lines of evidence to synthesize all the data and its uncertainty. This is how Bayes filters work to let robots reason about evidence under uncertainty. Unfortunately science is less formal than robotics.
    – SigmaX
    Commented Mar 4, 2014 at 14:41
  • 1
    The converging lines of evidence for radiometric dating are outside the scope of this question. But basically, it's like seeing a TV image in a noisy signal. Each pixel is somewhat unreliable, but when you piece together the whole picture, the predictions of the hypothesis "that's Mr. Rogers' face" are met so well that it's unambiguously true that Mr. Rogers is playing on TV. We may also be able to put bounds on the unreliability of each pixel once we know what Mr. Rogers' face looks like. But it's easier to just sample more pixels: get lots and lots of rocks from that mountain.
    – SigmaX
    Commented Mar 4, 2014 at 15:00
  • 1
    @SigmaX: Uh, so what I took from that was "K-Ar is a noisy method and can't trusted on individual samples, but with sufficient samples a strong signal can be determined." If that is true, it seems to me that THAT is a key part of the answer - it isn't reliable without a large number of samples.
    – Oddthinking
    Commented Mar 4, 2014 at 22:29

The accuracy of the K-Ar dating is dependent upon the following:

So there are reasons why their test could came back the way it did.

  • thanks for the links - that's some good knowledge! but according to @ElendilTheTall's answer, the results are never going to be accurate for a rock aged less than 100,000 years. so it sounds like the inaccuracies that you mention, come into play when dating rocks which fall in this age range and are not the reason for the inaccuracies in this particular case Commented Apr 14, 2011 at 9:25

The rock that came out of the volcano isn't magically reformed into brand new material: it's just some ancient rock from the mantle that's been brought up.

Have a read of Radiometric Dating, Geologic Time, And The Age Of The Earth: A Reply to "Scientific" Creationism by G. Brent Dalrymple. U.S. Geological Survey Open file report 86-110.

Your exact question is on page 26. These (flawed) studies cite examples of "anomalous" ages from specific lava floes:

Their claims:

"Volcanic rocks produced by the lava flows which occured in Hawaii in the years 1800-1801 were dated by the potassium-argon method. Excess argon produced apparent ages ranging from 160 million to 2.96 billion years." (Kofahl and Segraves, 1975, p.200)

And the author's response:

These authors cite a study by Funkhouser and Naughton (1968) on xenolithic inclusions in the 1801 flow from Hualalei Volcano on the Island of Hawaii. The 1801 flow is an unusual flow because it carries very abundant inclusions of rocks foreign to the lava. These inclusions, called xenoliths (meaning foreign rocks), consist primarily of olivine, a pale-green, iron-magnesium silicate mineral. They come from deep within the mantle and were carried upward to the surface by the lava.

In the field, they look like large raisins in a pudding, and even occur in beds piled one on top of the other, glued together by the lava. The study by Funkhouser and Naughton (1968) was on the xenoliths, not on the lava. The xenoliths, which vary in composition and range in size from single mineral grains to rocks the size of basketballs, do indeed carry excess argon in large amounts. Funkhouser and Naughton were quite careful to point out that the apparent "ages" they measured were not geologically meaningful. Quite simply, xenoliths are one of the types of rocks that cannot be dated by K-Ar techniques.

Funkhouser and Naughton were able to determine that the excess gas resides primarily in fluid bubbles in the minerals of the xenoliths, where it cannot escape upon reaching the surface. Studies such as the one by Funkhouser and Naughton (1968) are done to determine which materials are suitable for dating and which are not, and to determine the cause of sometimes strange results. They are part of the continuing effort to learn.

There have been two extensive K-Ar studies on historic lava flows (Dalrymple, 1969; Krummenacher, 1970) that showed that excess argon is not a serious problem for dating lava flows. An exception is the lava from the 1801 Hualalei flow, which is so badly contaminated by the xenoliths that it is not possible to obtain a completely inclusion-free sample.

  • That is one possible explanation, but I don't see any evidence to confirm it.
    – SigmaX
    Commented Apr 2, 2012 at 21:20

K-Ar dating is sometimes tricky, but if you understand what you are doing, the results are generally reliable. Geologists avoid dating glassy, polydeformed or very altered rocks with K-Ar, because these are known to sometimes be problematic . A lot of the difficulties with the K-Ar method outlined by the other user here are addressed by the newer 40Ar/39Ar method which allows for:

  1. stepwise degassing and the generation of isochrons - initial argon composition and concentration become less important
  2. much better uncertainties
  3. measurement of K and Ar on the same sample aliquot
  4. degassing patterns which can indicate whether an aliquot has lost or gained Ar

Therefore, the 40Ar/39Ar method yields results which are considered to be more robust.

Here are the original claims behind the Ngauruhoe volcanic rocks from the Institute for Creation Research.

If you look at the data in Table 4 you will see that only a very very small portion of the 40Ar is radiogenic (Column 40Ar*) between 1.7 and 4 percent, relative to the total 40Ar. The rest is assumed to be atmospheric; the atmospheric 40Ar is subtracted from the total 40Ar: 40Ar* = 40Ar(tot)-295.5*36Ar (because in atmosphere the 40Ar/36Ar ratio was assumed to be 295.5 during these determinations)

There have been some adjustments to the known exact composition of atmospheric argon used for corrections, it is now considered to be 40Ar/36Ar = ~298.56 rather than 295.5

Normally this does not matter too much because we work with old rocks which host much more 40Ar* relative to 40Ar(tot) and this correction is somewhat negligible. However for very young rocks like these it is a problem. If you redo the maths and calculate the ages with the ratio of 298.56 and propagate errors properly, you will find that most or all of them overlap with 0 age at the two sigma level, thus correctly dating the rocks of this volcano at their known age. Which means that in this case, the K/Ar method just measured zero ages. Because it is not precise enough for young rocks.

  • Would you kindly link to a notable early release article about the newer 40Ar/39Ar method, so we can guesstimate whether a given report from the field was done with the better method or not? Commented Aug 16, 2018 at 19:21
  • The 40Ar/39Ar method is used in 100s of labs, so there are many many publications. For a full overview I recommend this book: McDougall, I. and Harrison, T.M., 1999. Geochronology and Thermochronology by the 40Ar/39Ar Method. Oxford University Press on Demand. For articles discussing relatively young rocks, try a google scholar search like this: scholar.google.no/…
    – Geochron
    Commented Sep 3, 2018 at 6:36

Definitely not a creation website:

Problems with K-Ar dating

  • Complex and separate procedures to measure amounts of K and Ar
  • Measurements of K and Ar subject to errors on order of a percent – error in ratio is even larger
  • K and Ar must be measured on different parts of sample – inhomogeneity errors
  • No way of knowing if age is correct, or sample has lost or gained K or Ar

geo.arizona. edu/~reiners/.../kelly2002:

One of the fundamental assumptions of K-Ar dating (assumption 3, above) is that after correcting for atmospheric argon, all 40 Ar in the sample is the result of the in situ decay of 40 K, an assumption which is not always valid.

My first post was deleted for not being referenced, but think about why K-Ar dating is not used for "young" rock. It's assumed there is not enough 40Ar in the sample to be measured. However enough 40Ar was found in recently formed samples of volcanic rock. The excuse of the wrong test doesn't hold up. Also the "excess Argon" excuse does not hold up either. It's that the amount of Argon is not what they expected. BTW I not trying to say the Earth is young or old; I'm just saying K-Ar doesn't work because the assumptions are not valid.


K-Ar dating relies on very long half-lives, and hence should not be used for dating new rocks. Wikipedia says "the technique is most applicable for dating minerals and rocks more than 100,000 years old. For shorter timescales, it is likely that not enough Argon 40 will have had time to accumulate in order to be accurately measurable."

As an aside, this article from the Creation Institute admits that K-Ar's effective range starts at about 400,000 years old. That should refute your creationist friends.

  • everything you say is true, but i think you missed the point of K-Ar dating - the idea is to use K-Ar to determine the age of the rock without any external knowledge. Say one person (the creationists in this case) knows the age of the rock and another person (the various labs in this case) do not know when the rock was formed, the results from each independent source of knowledge should not contradict each other. this is otherwise known as a blind test and its one of the best ways to assess the accuracy of a trial. the creationists may be misguided but... Commented Jul 2, 2012 at 2:09
  • at least they did use a valid scientific method to attempt to prove their theories. the fact that it backfired only adds credibility to the K-Ar dating mathod. Commented Jul 2, 2012 at 2:14
  • "less than 270, 000 years or up to 3.7 million years old." - all the figures agree that the rocks are very young. But the scientists should have been more careful in pointing out that the rocks were too young to date reliably. Also, note that the last entry in the table is dated 1975. 40 years progress since then should give us better accuracy today. Creationists often use very old data to prove their point!
    – hdhondt
    Commented Jul 2, 2012 at 23:54
  • This is the correct answer. K-Ar has a lower bound, below which error and noise return a meaningless result. “the idea is to use K-Ar to determine the age of the rock without any external knowledge.” You can’t. Just like you can’t use phenolphthalein to determine the exact pH of an unknown mixture. Tests have limits. We use multiple methods to analyze a sample and look for corroboration. Commented Jan 22, 2023 at 0:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .