We now consider in more detail one of the problems with potassium-argon dating, namely, the branching ratio problem. Here is some relevant information that was e-mailed to me.
There are some very serious objections to using the potassium-argon decay family as a radiometric clock. This is harmful to the position of those holding to the theory of sea-floor spreading since their time scale has been calculated using K40/Ar40 dates mainly. About 11% of K40 decays by electron capture and gamma ray emission to Ar40 and the remaining 89% of the K40 decays by B-particle emission to form Ca40. The geochronologist considers the Ca40 of little practical use in radiometric dating since common calcium is such an abundant element and the radiogenic Ca40 has the same atomic mass as common calcium.
"Juggling" is also performed by geochronologists in this K-Ar system. Here the actual observed branching ratio is not used, but rather a small ratio is arbitrarily chosen in an effort to match dates obtained method with U-Th-Pb dates.
The branching ratio that is often used is 0.08, while the true value is probably about 0.12. This means that K-Ar dates computed with the lower branching ratio are a third too large, that is, the actual K-Ar date should be 2/3 of the computed date. Thus we have another source of error for K-Ar dating.
How Errors Can Account for the Observed Dates
Thus there are a number of sources of error. We now consider whether they can explain the observed dates. In general, the dates that are obtained by radiometric methods are in the hundreds of millions of years range. One can understand this by the fact that the clock did not get reset (if one accepts the fact that the magma "looks" old, for whatever reason). That is, we can get both parent and daughter elements from the magma inherited into minerals that crystallize out of lava, making these minerals look old. Since the magma has old radiometric dates, depending on how much the clock gets reset, the crust can end up with a variety of younger dates just by partially inheriting the dates of the magma.
Thus any method based on simple parent to daughter ratios such as Rb-Sr dating is bound to be unreliable, since there would have to be a lot of the daughter product in the magma already. And Harold Coffin's book Creation by Design lists a study showing that Rb-Sr dates are often inherited from the magma.
Even the initial ratios of parent and daughter elements in the earth do not necessarily indicate an age as old as 4.5 billion years. Radioactive decay would be faster in the bodies of stars, which is where scientists assume the heavy elements formed. Imagine a uranium nucleus forming by the fusion of smaller nucleii. At the moment of formation, as two nucleii collide, the uranium nucleus will be somewhat unstable, and thus very likely to decay into its daughter element. The same applies to all nucleii, implying that one could get the appearance of age quickly. Of course, the thermonuclear reactions in the star would also speed up radioactive decay. But isochrons might be able to account for pre-existing daughter elements.
Furthermore, some elements in the earth are too abundant to be explained by radioactive decay in 4.5 billion years (such as calcium, argon, and, I believe, strontium). Some are too scarce (such as helium). So it's not clear to me how one can be sure of the 4.5 billion year age, even assuming a constant decay rate.
Why older dates would be found lower in the geologic column especially for K-Ar dating
In general, potassium-argon dates appear to be older the deeper one goes in the crust of the earth. We now consider possible explanations for this.
There are at least a couple of mechanisms to account for this. In volcano eruptions, a considerable amount of gas is released with the lava. This gas undoubtedly contains a significant amount of argon 40. Volcanos typically have magma chambers under them, from which the eruptions occur. It seems reasonable that gas would collect at the top of these chambers, causing artificially high K-Ar radiometric ages there. In addition, with each successive eruption, some gas would escape, reducing the pressure of the gas and reducing the apparent K-Ar radiometric age. Thus the decreasing K-Ar ages would represent the passage of time, but not necessarily related to their absolute radiometric ages. As a result, lava found in deeper layers, having erupted earlier, would generally appear much older and lava found in higher layers, having erupted later, would appear much younger. This could account for the observed distribution of potassium-argon dates, even if the great sedimantary layers were laid down very recently. In addition, lava emerging later will tend to be hotter, coming from deeper in the earth and through channels that have already been warmed up. This lava will take longer to cool down, giving more opportunity for enclosed argon to escape and leading to younger radiometric ages. A discussion of these mechanisms may be found at the Geoscience Research Institute site.
Another factor is that rocks absorb argon from the air. It is true that this can be accounted for by the fact that argon in the air has Ar36 and Ar40, whereas only Ar40 is produced by K-Ar decay. But for rocks deep in the earth, the mixture of argon in their environment is probably much higher in Ar40, since only Ar40 is produced by radioactive decay. As these rocks absorb argon, their radiometric ages would increase. This would probably have a larger effect lower down, where the pressure of argon would be higher. Or it could be that such a distribution of argon pressures in the rocks occurred at some time in the past. This would also make deeper rocks tend to have older radiometric ages.
Recent lava flows often yield K-Ar ages of about 200,000 years. This shous that they contain some excess argon, and not all of it is escaping. If they contained a hundred times more excess argon, their K-Ar ages would be a hundred times greater, I suppose. And faster cooling could increase the ages by further large factors. I also read of a case where a rock was K-Ar dated at 50 million years, and still susceptible to absorbing argon from the air. This shows that one might get radiometric ages of at least 50 million years in this way by absorbing Ar40 deep in the earth without much Ar36 or Ar38 present. If the pressure of Ar40 were greater, one could obtain even greater ages.
Yet another mechanism that can lead to decreasing K-Ar ages with time is the following, in a flood model: One can assume that at the beginning of the flood, many volcanoes erupted and the waters became enriched in Ar40. Then any lava under water would appear older because its enclosed Ar40 would have more trouble escaping. As time passed, this Ar40 would gradually pass into the atmosphere, reducing this effect and making rocks appear younger. In addition, this would cause a gradient of Ar40 concentrations in the air, with higher concentrations near the ground. This also could make flows on the land appear older than they are, since their Ar40 would also have a harder time escaping.
Do different methods agree with each other on the geologic column?
Let us consider the question of how much different dating methods agree on the geologic column, and how many measurements are anomalous, since these points are often mentioned as evidences of the reliability of radiometric dating. It takes a long time to penetrate the confusion and find out what is the hard evidence in this area.
In the first place, I am not primarily concerned with dating meteorites, or precambrian rocks. What I am more interested in is the fossil-bearing geologic column of Cambrian and later age.
Now, several factors need to be considered when evaluating how often methods give expected ages on the geologic column. Some of these are taken from John Woodmoreappe's article on the subject, but only when I have reason to believe the statements are also generally believed. First, many igneous formations span many periods, and so have little constraint on what period they could belong to. The same applies to intrusions. In addition, some kinds of rocks are not considered as suitable for radiometric dating, so these are typically not considered. Furthermore, it is at least possible that anomalies are under-reported in the literature. Finally, the overwhelming majority of measurements on the fossil bearing geologic column are all done using one method, the K-Ar method. (And let me recall that both potassium and argon are water soluble, and argon is mobile in rock.) Thus the agreement found between many dates does not necessarily reflect an agreement between different methods, but rather the agreement of the K-Ar method with itself. For example, if 80 percent of the measurements were done using K-Ar dating, and the other 20 percent gave random results, we still might be able to say that most of the measurements on a given strata agree with one another reasonably well. So to me it seems quite conceivable that there is no correlation at all between the results of different methods on the geologic column, and that they have a purely random relationship to each other.
Let us consider again the claim that radiometric dates for a given geologic period agree with each other. I would like to know what is the exact (or approximate) information content of this assertion, and whether it could be (or has been) tested statistically. It's not as easy as it might sound.
Let's suppose that we have geologic periods G1 ... Gn. Let's only include rocks whose membership in the geologic period can be discerned independent of radiometric dating methods. Let's also only include rocks which are considered datable by at least one method, since some rocks (I believe limestone) are considered not to hold argon, for example.
Now, we can take a random rock from Gi. We will have to restrict ourselves to places where Gi is exposed, to avoid having to dig deep within the earth. Let's apply all known dating methods to Gi that are thought to apply to this kind of rock, and obtain ages from each one. Then we can average them to get an average age for this rock. We can also compute how much they differ from one another.
Now we have to be careful about lava flows -- which geologic period do they belong to? What about rocks that are thought not to have their clock reset, or to have undergone later heating episodes? Just to make the test unbiased, we will assign altitude limits to each geologic period at each point on the earth's surface (at least in principle) and include all rocks within these altitude limits within Gi, subject to the condition that they are datable.
The measurements should be done in a double-blind manner to insure lack of unconscious bias.
For each geologic period and each dating method, we will get a distribution of values. We will also get a distribution of averaged values for samples in each period. Now, some claim is being made about these distributions. It is undoubtedly being claimed that the mean values ascend as one goes up the geologic column. It is also being claimed that the standard deviations are not too large. It is also being claimed that the different methods have distributions that are similar to one another on a given geologic period.
The only correlation I know about that has been studied is between K-Ar and Rb-Sr dating on precambrian rock. And even for this one, the results were not very good. This was a reference by Hurley and Rand, cited in Woodmorappe's paper. As far as I know, no study has been done to determine how different methods correlate on the geologic column (excluding precambrian rock).
The reason for my request is that a correlation is not implied by the fact that there are only 10 percent anomalies, or whatever. I showed that the fact that the great majority of dates come from one method (K-Ar) and the fact that many igneous bodies have very wide biostratigraphic limits, where many dates are acceptable, makes the percentage of anomalies irrelevant to the question I am asking. And since this agreement is the strongest argument for the reliability of radiometric dating, such an assumption of agreement appears to be without support so far.
The question of whether different methods correlate on the geologic column is not an easy one to answer for additional reasons. Since the bulk of K-Ar dates are generally accepted as correct, one may say that certain minerals are reliable if they tend to give similar dates, and unreliable otherwise. We can also say that certain formations tend to give reliable dates and others do not, depending on whether the dates agree with K-Ar dates. Thus we can get an apparent correlation of different methods without much of a real correlation in nature. It's also possible for other matter to be incorporated into lava as it rises, without being thoroughly melted, and this matter may inherit all of its old correlated radiometric dates. Coffin mentions that fission tracks can survive transport through lava, for example. It may also be that lava is produced by melting the bottom of continents and successively different layers are melted with time, or there could be a tendency for lighter isotopes to come to the top of magma chambers, making the lava there appear older. But anyway, I think it is important really to know what patterns appear in the data to try to understand if there is a correlation and what could be causing it. Not knowing if anomalies are always published makes this harder.
It is often mentioned that different methods agree on the K-T boundary, dated at about 65 million years ago. This is when the dinosaurs are assumed to have become extinct. This agreement of different methods is taken as evidence for a correlation between methods on the geologic column. One study found some correlated dates from bentonite that are used to estimate the date of the K-T boundary. I looked up some information on bentonite. It is composed of little glass beads that come from volcanic ash. This is formed when lava is sticky and bubbles of gas in it explode. So these small particles of lava cool very fast. The rapid cooling might mean that any enclosed argon is retained, but if not, the fact that this cooling occurs near the volcano, with a lot of argon coming out, should guarantee that these beads would have excess argon. As the gas bubble explodes, its enclosed argon will be rushing outward along with these tiny bubbles as they cool. This will cause them to retain argon and appear too old. In addition, the rapid cooling and the process of formation means that these beads would have Rb, Sr, U, and Pb concentrations the same as the lava they came from, since there is no chance for crystals to form with such rapid cooling. So to assume that the K-Ar dates, Rb-Sr dates, and U-Pb dates all reflect the age of the lava, one would have to assume that this lava had no Sr, no Pb, and that all the argon escaped when the beads formed. Since the magma generally has old radiometric ages, I don't see how we could have magma without Pb or Sr. In fact, I doubt that there is fresh uncrystallized lava anywhere on earth today that has zero U/Pb and Rb/Sr ages, as would be required if bentonite gave an accurate date for the K-T boundary. So to me it seems to be certain that these ages must be in error.
Furthermore, the question arises whether bentonite always gives correlated ages, and whether these ages always agree with the accepted ages for their geologic period. I believe that bentonite occurs in a number of formations of different geologic periods, so this could be checked. If bentonite does not always give correlate and correct ages, this calls into question its use for dating the K-T boundary.
Possible other sources of correlation
Note that if there are small pockets in crystals where both parent and daughter product can accumulate from the lava, then one can inherit correlated ages from the lava into minerals. Thus even the existence of correlations is not conclusive evidence that a date is correct.
Anomalies of radiometric dating
If a date does not agree with the expected age of its geologic period, and no plausible explanation can be found, then the date is called anomalous. But if we really understand what is going on, then we should be able to detect discrepant dates as they are being measured, and not just due to their divergence from other dates.
Geologists often say that the percentage of anomalies is low. But there are quite a number of rather outstanding anomalies in radiometric dating that creationists have collected. These anomalies are reported in the scientific literature. For example, one isochron yielded a date of 10 billion years. A Rb-Sr isochron yielded a date of 34 billion years. K-Ar dates of 7 to 15 billion years have been recorded. It's also not uncommon for two methods to agree and for the date to be discarded anyway. Samples with flat plateaus (which should mean no added argon) can give wrong dates. Samples giving no evidence of being disturbed can give wrong dates. Samples that give evidence of being disturbed can give correct dates. The number of dates that disagree with the expected ages is not insignificant. I don't know what the exact percentage is.
Many dates give values near the accepted ones. But even these often differ from one another by 10 or 20 percent. And quite a few other dates are often much, much farther off. Whatever is making some of these dates inaccurate could be making all of them inaccurate.
It's interesting to note that in a few cases, old radiometric dates are above young ones.
The fact that different methods often give different dates is noted by geologists. Here are some quotes from
http://hubcap.clemson.edu/spurgeon/books/apology/Chapter7.html "It is obvious that radiometric techniques may not be the absolute dating methods that they claimed to be. Age estimates on a given geological stratum by different radiometric methods are often quite different (sometimes by hundreds of millions of years). There is not absolutely reliable long-term radiological "clock". The uncertainties inherent in radiometric dating are disturbing to geologists and evolutionists... 
As proof of the unreliability of the radiometric methods consider the fact that in nearly every case dates from recent lava flows have come back excessively large. One example is the rocks from the Kaupelehu Flow, Hualalai Volcano in Hawaii which was known to have erupted in 1800-1801. These rocks were dated by a variety of different methods. Of 12 dates reported the youngest was 140 million years and the oldest was 2.96 billion years. The dates average 1.41 billion years. "
Another source said that about 5 or 6 of the historic lava flows give ages in the hundreds of thousands of years. Geologists explain the Kaupelehu date by the lava being cooled rapidly in deep ocean water and not being able to get rid of its enclosed argon.
Here are some quotes from John Woodmorappe's paper, "Radiometric Geochronology Reappraised," Creation Research Society Quarterly 16(2)102-29, p. 147, September 1979, that indicate that radiometric dates are scattered, and that anomalies are often not reported:
"Improved laboratory techniques and improved constants have not reduced the scatter in recent years. Instead, the uncertainty grows as more and more data is accumulated ... " (Waterhouse).
"In general, dates in the `correct ball park' are assumed to be correct and are published, but those in disagreement with other data are seldom published nor are discrepancies fully explained." (Mauger)
" ... the thing to do is get a sequence of dates and throw out those that are vastly anomalous." (Curtis et al)
" ... it is usual to obtain a spectrum of discordant dates and to select the concentration of highest values as the correct age." (Armstrong and Besancon).
"In general, strong discordances can be expected among ages deduced by different methods." (Brown and Miller)
Woodmorappe also mentions that very self-contradictory age spreads in the Precambrian era are common.
In addition, Woodmorappe gives over 300 sets of dates "that are in gross conflict with one another and with expected values for their indicated paleontological positions." This table is limited to dates that approach 20% discrepancy, too old or too young. This does not include dates from minerals that are thought to yield bad dates, or from igneous bodies with wide biostrategraphic ranges, where many dates are acceptable. He states that the number of dates within range are less than the number of anomalies, except for the Cenozoic and Cretaceous. When one adds in the fact that many anomalies are unreported, which he gives evidence for, the true distribution is anyone's guess. He also combines evidence from the literature to conclude that "somewhat less than half of all dates agree with 10% of accepted values for their respective biostratigaphic positions." I believe this estimate even includes igneous bodies with very wide biostrategraphic limits, and does not include unpublished anomalies.
There have been criticisms of John Woodmorappe's study, but no one has given any figures from the literature for the true percentage of anomalies, with a definition of an anomaly, or the degree of correlation between methods. Steven Schimmrich's review of this study often concerns itself with John W's presentation of geologists explanation for anomalies, and not with the percentage of anomalies; the later is my main concern.
Here are a couple of more quotes about anomalies:
"Situations for which we have both the carbon-14 and potassium-argon ages for the same event usually indicate that the potassium-argon `clock' did not get set back to zero. Trees buried in an eruption of Mount Rangotito in the Auckland Bay area of New Zealand provide a prime example. The carbon-14 age of the buried trees is only 225 years, but some of the overlying volcanic material has a 465,000-year potassium-argon age."
[Harold Coffin, Origin by Design, page 400.]
A similar situation is reported in the December 1997 issue of Creation ex nihilo in which lava with a K-Ar age of about 45 million years overlays wood that was carbon dated by 3 laboratories using AMS dating to about 35,000 years.
Still another evidence for problems with radiometric dating was given in a recent talk I attended by a man who had been an evolutionist and taken a course in radiometric dating. The teacher gave 14 assumptions of radiometric dating and said something like "If creationists got a hold of these, they could cut radiometric dating to pieces."
Another evidence that all is not well with radiometric dating is given in the following quote from Coffin p. 302:
"We find that most primary radioactive ores that have not been exposed to weathering exist in secular equilibrium. Many sedimentary uranium ores are not."
Since equilibrium should be reached in 1 million years, this is a problem for sediments that are assumed to be older than 1 million years.
On another point, if we can detect minerals that were not molten with the lava, as has been claimed, then this is one more reason why there should be no anomalies, and radiometric dating should be a completely solved problem. But that does not appear to be the case, at least (especially) on the geologic column.
I'm not claiming that anomalous results are being hidden, just that the agreement of a mass of results, none of which has much claim to reliability, does not necessarily mean much.
Picking out a few cases where radiometric dates appear to be well-behaved reminds me of evolutionary biologists focusing on a few cases where there may be transitional sequences. It does not answer the overall question. And as I said above, I'm also interested to know how much of the fossil-bearing geologic column can be dated by isochrons, and how the dates so obtained compare to others.
Concerning K-Ar anomalies, here is a quote from Woodmorappe's paper cited above, p. 122:
"K-Ar ages much greater than inferred earth age are also common. Gerling et al called attention to some chlorites yielding K-Ar dates of 7 to 15 b.y. It had been noted that some minerals which yield such dates (as beryl, cordierite, etc.) can be claimed to have trapped excess argon in their channel structures or to have fractioned the Ar isotopes, but none of this can apply to the simple mica-like structures of chlorite. They also pointed out that for the anomalies to be accounted for by excess argon, unreasonably high partial pressures of Ar during crystallization would have to be required. They concluded by suggesting some unknown nuclear process which no longer operates to have generated the Ar."
This implies that excess argon is coming from somewhere. Here is another quote from Woodmorappe about isochrons, since some people think that mixing scenarios or other age-altering scenarios are unlikely:
Shafiqullah and Damon said: "The Ar40/Ar36 vs. K40/Ar36 isochrons are valid only when all samples of the system under consideration have the same non-radiogenic argon composition. If this condition does not hold, invalid ages and intercepts are obtained. Models 2-9 yield isochron ages that are too high, too low, or in the future, sometimes by orders of magnitude."
from Woodmorappe, "An Anthology of Matters Significant to Creationism and Diluviology, Report 1," Creation Research Society Quarterly 16(4)209-19, March 1980, p. 218.
The fact that the only "valid" K-Ar isochrons are those for which the concentration of non-radiogenic argon (Ar36) is constant, seems very unusual. This suggests that what is occuring is some kind of a mixing phenomenon, and not an isochron reflecting a true age.
The following quote is from http://www.pathlights.com/ce_encyclopedia/Index.htm "Processes of rock alteration may render a volcanic rock useless for potassium-argon dating . . We have analyzed several devitrified glasses of known age, and all have yielded ages that are too young. Some gave virtually zero ages, although the geologic evidence suggested that devitrification took place shortly after the formation of a deposit." *J.F. Evernden, et. al., "K / A Dates and Cenozoic Mannalian Chronology of North America," in American Journal of Science, February 1964, p. 154.
Why a low anomaly percentage is meaningless
One of the main arguments in favor of radiometric dating is that so many dates agree with each other, that is, with the date expected for their geologic period. But it's not evident how much support this gives to radiometric dating. If a rock dates too old, one can say that the clock did not get reset. If it dates too young, one can invoke a later heating event. Neither date would necessarily be seen as anomalous. If lava intrudes upon geologic period X, then any date for the lava of X or later will not be seen as anomalous. And even if the date is one or two geologic periods earlier, it may well be close enough to be accepted as non-spurious. If one does not know the geologic period of a rock by other means, then of course one is likely to date it to find out, and then of course the date agrees with the geologic period and this will not be seen as anomalous. So it is difficult to know what would be a reasonable test for whether radiometric dating is reliable or not. The percentage of published dates that are considered as anomalous has little bearing on the question.
The biostrategraphic limits issue
The issue about igneous bodies may need additional clarification. If a lava flow lies above geologic period A and below B, then allowable ages are anything at least as large as A and no larger than B. This is called the biostratigraphic limit of the flow. Now, according to Woodmorappe's citations, many lava flows have no such limits at all, and most of them have large limits. For example, a flow lying on precambrian rock with nothing on top would have no limits on its dates. And such flows often have a large internal scatter of dates, but these dates are not considered as anomalies because of the unrestricted biostratigraphic limit. Other flows with wide biostratigraphic limits have weak restrictions on allowable dates. This is one reason why just reporting the percentage of anomalies has little meaning.
John W. states that very many igneous bodies have little or no biostrategraphic limits, so just about any age is acceptable. Thus these ages, though they generally have a considerable scatter, are not considered as anomalies. He cites another reference that most igneous bodies have wide biostrategraphic limits. Thus just by chance, many dates will be considered within the acceptable ranges. If the igneous body is constrained to have a date between that of geologic period X1 and X2, with times T1 and T2, and if we regard any date within 20 percent as non-anomalous, then any date between T1/1.2 and T2*1.2 will be considered as non-anomalous, and this will include a considerable portion of geologic history. Again, the percentage of anomalies means nothing for the reliability of radiometric dating.
Now, igneous bodies can be of two types, extrusive and intrusive. Extrusive bodies are lava that is deposited on the surface. These cool quickly and have small crystals and form basalt. Intrusive bodies are deposited in the spaces between other rocks. These cool more slowly and have larger crystals, often forming granite. Both of these tend on the average to have wide biostrategraphic limits, meaning that a large spread of ages will be regarded as non-anomalous. And if we recall that most radiometric dating is done of igneous bodies, one sees that the percentage of anomalies is meaningless. Thus we really need some evidence that the different methods agree with each other.
To make the case even stronger, "Many discrepant results from intrusives are rationalized away immediately by accepting the dates but reinterpreting the biostrategraphic bracket," according to John Woodmorappe. This of course means that the result is no longer anomalous, because the geologic period has been modified to fit the date. Finally, the fact that the great majority of dates are from one method means that the general (but not universal) agreement of K-Ar dating with itself is sufficient to explain the small percentange of anomalies (if it is small).
Preponderance of K-Ar dating
Now, the point about agreement is that whatever figure is given about how often ages agree with the expected age, is consistent with the fact that there is no agreement at all between K-Ar and other methods, since so many measurements are done using K-Ar dating. And one of the strongest arguments for the validity of radiometric dating is that the methods agree. So I'm very interested to know what data there is about how often _different_ methods agree.
So when one combines all of the above figures, the statement that there are only 10 percent anomalies or 5 percent or whatever, does not have any meaning any more. This statement is made so often as evidence for the reliability of radiometric dating, that the simple evidence that it has no meaning, is astounding to me. I don't object to having some hard evidence that there are real agreements between different methods on the geologic column, if someone can provide it. The precambrian rock is less interesting because it could have a radiometric age older than life, but this is less likely for the rest of the geologic column.
It's not surprising that K-Ar dates often agree with the assumed dates of their geological periods, since the dates of the geological periods were largely inferred from K-Ar dating.
By the way, Ar-Ar dating and K-Ar dating are essentially the same method, so between the two of them we obtain a large fraction of the dates being used.
Some information from an article by Robert H. Brown at the Geoscience Research Institute site confirms the preponderance of K-Ar dating:
History of the Radioisotope based Geologic Time Scale
Before the discovery of radioactivity in the late nineteenth century, a geological time scale had been developed on the basis of estimates for the rates of geological processes such as erosion and sedimentation, with the assumption that these rates had always been essentially uniform. On the basis of being unacceptably old, many geologists of the time rejected these early twentieth century determinations of rock age from the ratio of daughter to radioactive parent (large). By 1925, increased confidence in radioisotope dating techniques and the demands of evolution theory for vast amounts of time led to the establishment of an expanded geological time scale. With the K-Ar dating techniques developed after World War II, this time scale was refined to the standard Geologic Time Scale adopted in 1964. The construction of this time scale was based on about 380 radioisotope ages that were selected because of their agreement with the presumed fossil and geological sequences found in the rocks. Radioisotope ages that did not meet these requirements were rejected on the basis of presumed chemical and/or physical modifications that made the "ages" unreliable indicators of real time. About 85% of the selections were K-Ar date s, 8% rubidium-strontium dates, and 4% uranium-lead dates. Igneous rocks are particularly suited to K-Ar dating. The crucial determiners are therefore volcanic (extrusive igneous) rocks that are interbedded with sediments, and intrusive igneous rocks that penetrate sediments.
This verifies what I said about almost all of the dates used to define correct ages for geologic periods being K-Ar dates. Also, the uncertainty in the branching ratio of potassium decay might mean that there is a fudge factor in K-Ar ages of up to a third, and that the occasional agreements between K-Ar ages and other ages are open to question.
So the point is that there is now no reason to believe that radiometric dating is valid on the geologic column.
Excuses for anomalies
Another issue is that sometimes the geologic periods of rocks are revised to agree with the ages computed. This also makes data about percentages of anomalies less meaningful.
It sometimes seems that reasons can always be found for bad dates, especially on the geologic column. If a rock gives a too old date, one says there is excess argon. If it gives a too young date, one says that it was heated recently, or cannot hold its argon. How do we know that maybe all the rocks have excess argon? It looks like geologists are taking the "majority view" of K-Ar dating, but there is no necessary reason why the majority of rocks should give the right date.
The following quote is from the article by Robert H. Brown, cited earlier:
What is a Radioisotope Age?
The relationship of a radioisotope age with real-time must be based on an interpretation. A discussion of rubidium-strontium ages in the Isotope Geoscience Section of the journal, Chemical Geology, specifically states that a “radioisotope age determination "does not certainly define a valid age information for a geological system. Any interpretation will reflect the interpreters presuppositions (bias).
Need for a double-blind test
Concerning the need for a double blind test, it would seem that there are many places where human judgment could influence the distribution of measured radiometric dates. It could increase the percentage of anomalies, if they were regarded as more interesting. It could decrease them, if they were regarded as flukes. Human judgment could determine whether points were collinear enough to form an isochron. It could determine whether a point can justifiably be tossed out and the remaining points used as an isochron. It could determine whether one should accept simple parent-to-daughter K-Ar ratios or whether some treatment needs to be applied first to get better ages. It could influence whether a spectrum is considered as flat, whether a rock is considered to have undergone leaching or heating, whether a rock is porous or not, or whether a sample has been disturbed in some way.
Since one of the main reasons for accepting radiometric dates (at least I keep hearing it) is that they agree with each other, I think that geologists have an obligation to show that they do agree, specifically on the geologic column. Since we do not know whether or how much human judgment is influencing radiometric dating, a double blind study is most reasonable. And it should not be restricted to just one or two well-behaved places, but should be as comprehensive as possible.