Robert W Bales Nov0492 12:41PM Isochron Errors I The accuracy of isochron dating has been

Robert W Bales Nov-04-92 12:41PM Isochron Errors -- I The accuracy of isochron dating has been a topic in this group. Those who insist that the only idea which should be considered is that the earth is billions of years old hold that isochron dates are accurate. I am told I have not shown that isochron dates can be wrong. But, as I have pointed out, neither have those who trust isochrons presented evidece independent of the theory to establish their points. My position has been that, in the lack of definite information on either side, the possibilities that dates obtained from isochrons are and are not accurate should be considered equally. I now have some additional information on the errors which can occur with isochrons. To avoid any possible misunderstaning, I also want to emphesize that those I quote as having found descrepancies in isocron dates do not (with one partial exception) believe that ischon dating as a whole is undermined. Neither do they share my belief in a young earth. This information does not show all dates derived from isochrons to be wrong, and I am not claiming that all such dates should be absolutely discarded. However, the evidence does show that, contrary to what has been claimed, isochron methods are not completely self-calibrating and are not immune from the problems which can radiometric dating in general. In isocron dating, ages are obtained from the slope of a line based on isotope ratios measured for different minerals of the same age. The theory is that, although the different minerals have different initial amounts of the radioactive parent, the same percentage of the parent will decay in each case, with the result that themeasured isotope ratios will fall on a line. There are different isochron systems. Dalrymple [1] discusses the Rb-Sr isochron, which plots the ratio Sr(87)/Sr(86) as a function of the ratio Rb(87)/Sr(86) ratio: "When a rock is first formed, say from a magma, the Sr(87)/Sr(86) ratios in all of the minerals willl be the same, regardless of the rubidium or strontium contents of the minerals, so all of the samples will plot on a horizontal line." (page 32) However, Brooks, James, and Hart [2] have found that this statement is not always true. They studied 30 examples given in the literature, correcting (where necessary; The paper states: "For the most part the psuedoisochrons come from young volcanic terranes in which no correction of the present-day Sr isotopic composition is necessary." -- page 1087) the measured isotopic ratios "to the time of their crystalliztion by using independent information for the age of crystallization." (page 1087) According to isochron theory, the resulting diagrams, called psuedoisochrons, should be horizontal lines. However, the authors report: "Correlation theory and regression analysis indicate that most of these psuedoisochrons have slopes significantly different from zero at confidence levels up to 95 percent (in some cases up to 99.9 percent) and that they define excess "ages" ranging from 70 million years to more than 3000 million years." (page 1087) Note that the upper limit of error magnitude would give a newly formed rock an "age" 75% of the believed age of the earth. The paper states that the cause of the error could be contamination, but also states a number of reasons for believing it is not, putting forth the idea that the results represent the age of the underlying mantle that was the source of the magma. The authors' cvonclusion is: "One serious consequence of the mantle isochron model is that crystallization ages determined on basic igneous rocks by the Rb-Sr whole rock technique can be greater than the true age by many hundreds of millions of years. This problem of inheirited age is more serious for younger rocks , and there are well- documented instances of conflicts between stratigraphic age and Rb-Sr age in the literature." (page 1093) In light of the last sentance, remember that agreement with stratigraphic ages was one of the reasons given for believing radiometric dates. I'm trying to keep my articles shorter, so I'll continue this in a second article. Bob Bales Tektronix, Inc. I help Tektronix make their instruments. They don't help me make my opinions. (The list of all references is included in both articles.) [1] G. Brent Dalrymple, Ph. D., "Radiometric Dating, Geologic Time, and the age of the Earth: A Reply to 'Scientific' Creationism," February, 1982, published by U.S Geological Survey. [2] C. Brooks, D.E. James, and S.R. Hart, "Ancient Lithosphere: Its Role in Young Continental Volcanism," _Science_ Volume 193, September 17, 1976, pages 1086-1094 [3] Carter, Evensen, Hamilton, and O'Nions, "Neodymium and Strontium Isotope Evidence for Crustal Contamination of Continental Volcanos," _Science_, vol 202, November 17, 1978, pages 743-747 [4] Felix Chayes, "Use of Correlation Statistics with Rubidium-Strontium Systemics, _Science_, vol 196, June 10, 1977, pages 1234-1235 ------------------------------------------------------------------------- Robert W Bales Nov-04-92 12:42PM Isochron Errors -- 2 In my first posting on this topic, I pointed out that a study of "pseudo- isochrons" [2] gave too ancient-date for some samples. The authors of the study contended that the dates represented real dates for the earth's mantle. One respone I expect is, "Even if the dates are not those of the rock's crystallization, the earth is at least as old as its mantle, so, 'so what?'" But not everone thinks that the ancient dates mean something. Another paper [3] (which reference the above paper but does not study all of the same cases) states that, in the cases studied by the authors: [garbled] rubidium-strontium 'psuedoisochrons' of different ages in these contaminated continental volcanos indicates that they are artifacts of the contamination processes and have no temporal significance with respect to mantle fractionation events." (page 743 -- abstract of paper) Another objection to my coclusion might be drawn from the characterization of the results which appears in [3]: "It has been observed in several Sr isotope studies that certain continental volcanic sequences display positive correlation between 87Sr/86Sr and 87Rb/86Sr, simulating poorly defined isochrons that imply ages significantly greater than the ages of eruption." (page 743) I've been in these discussions long enough to know that someone is almost sure to say, "Those are 'poorly defined' isochrons. 'Well defined' isochrons are accurate." However, the isocrons are claimed by some, and were seriously considered by others, to represent real ages. No one said, "These isochrons should be rejected because they are ill-defined." Any isochron sufficiently well defined to establish an age is sufficiently well defined to be a counter-example. In reference [4], the author questions whether the results reported in [2] are statistically significant. (The quick-witted :-) will notice that if this latter author is correct, then [2] does not necessarily show excess age, and so my claims of isochron errors might appear to vanish. However, they don't, since the particular calula- tions in question are not the only ones which show isochrons to be wrong -- see [2] and [3].) The author makes a most interesting statement: "In most discussions of the Sr-Rb procedure for determining geologic age the variables are given as (87)Sr/(86)Sr and Rb/(86)Sr, a clear example of correlation between two ratios having a common denominator; even if the three terms of the ratio are completely uncorrelated, the ratios will be positively correlated. In the current work, the denominator of the second ratio is Sr rather than (86)Sr, which complicates but does not mitigate the situation. It is possible, of course, that the common element bias is negligibly small, but this is a matter requiring demonstration or convincing rationalization in each specific case. That the denominator may perhaps be serving only as a convienient scaling device provides no escape; use of an uncorrelated scaling variable whose varience is sufficiently large will impose strong positive correlation on *any* pair of variables whose negative correlation is less than perfect. That the scaled data cluster closely about a regression line is then mearly a consequence of the choice of scaling variable and says nothing about the nature of the relation between the variables being scaled." (page 1234, emphesis in original.) It has been stated a number of times in this group that isochrons are self- checking and thata "good" isochon line means the data is accurate. It seems, however, that a trend line may be entirly the project of the methodolgy, and may give *no information* on age!! The authors of [2], in a response published as part of [4], recognize the implications of the above: "Chayes's criticism, on face value, not only threatens to undermine the manner in which we treated Rb-Sr mantle isochrons, but puts in question the very essence of Rb-Sr geochronology -- namely, the regression treatment of Rb-Sr isochrons. Thus, in replying to Chayes, we must defend not only our recent Rb-Sr mantle isochron proposal, but also the Rb-Sr isochron concept in general. We believe that we can successfully do both." (page 1235) The last statement indicates that, as a minimum, isochron accuracy is not guarrenteed, that it is a valid subject of discussion, and that in *each* case in which isochron accuracy is claimed, the claim needs to be suppoted by *positive* evidence, and not just accepted by default unless disproven. Bob Bales Tektronix, Inc. I help Tektronix make their instruments. They don't help me make my opinions. (The list of all references is included in both articles.)] [1] G. Brent Dalrymple, Ph. D., "Radiometric Dating, Geologic Time, and the age of the Earth: A Reply to 'Scientific' Creationism," February, 1982, published by U.S Geological Survey. [2] C. Brooks, D.E. James, and S.R. Hart, "Ancient Lithosphere: Its Role in Young Continental Volcanism," _Science_ Volume 193, September 17, 1976, pages 1086-1094 [3] Carter, Evensen, Hamilton, and O'Nions, "Neodymium and Strontium Isotope Evidence for Crustal Contamination of Continental Volcanos," _Science_, vol 202, November 17, 1978, pages 743-747 [4] Felix Chayes, "Use of Correlation Statistics with Rubidium-Strontium Systemics, _Science_, vol 196, June 10, 1977, pages 1234-1235 ======================================================================== In article <1992Nov4.225323.16178@scic.intel.com>, sbradley@scic.intel.com writes: > Bob, in the 6 years I've been reading this newsgroup, I > and several others have asked you many, many times to > present _any_ mechanism whereby several different dating > methods can yield the same incorrect (according to you) > ancient date for a single sample. You have never uttered a > single word in response. Are you ever going to address > this issue, or will we continue to see a never-ending > stream of hand-waving?