Earlier in the year, Dr. Andrew Snelling of Answers in Genesis gave us ample reason to believe that radiometric dating of meteorites is solid. Why does this matter? By dating certain types of meteorites, geologists arrive at the most precise age estimate of our Earth: 4.56 billion years. One may try to dispute whether the Earth formed simultaneously with these meteorites (we have strong reason to believe it did), since the oldest minerals that we can date directly are only ~4.3 billion years old. Regardless, even Snelling recognizes that multiple independent methods consistently tell us the Earth is billions—not thousands—of years old.
But how precisely do we know these ages? One of the key assumptions in radiometric dating is that we can know how fast the radioactive parent element decays into the radiogenic daughter element. Just like calibrating an hourglass, where sand falls at a constant rate, experimental observers might begin by trying to count directly how much sand falls each second (i.e. how many atoms decay into the daughter element). If 1 gram of sand falls per second, and the hourglass contains 3.6 kg of sand, then it should take exactly 1 hour (3,600 seconds) for all the sand to drain from top to bottom. Once this rate is observed over a limited period of time, we can test the accuracy of our observation by extrapolation and/or comparing it to other timekeeping methods, such as a sundial. If the observed rate is correct, then the shadow on our sundial should move from 1 PM to 2 PM in the same amount of time it takes for the hourglass to empty.
As I type on my laptop calibrated to satellite-based clocks, I use these ‘primitive’ examples because of their inherent uncertainties and relevance to radiometric dating. Given the subjectivity of our observation, sundials are not accurate to the second, so there is some ‘wiggle room’ in their prescribed length of an hour. The very process of sand grains falling can be very difficult to measure and is somewhat stochastic, just like radioactive decay. Thirteen grains might fall in one second, but only eleven grains the next, so we are looking for an average rate. In any case, our measurement should improve with better technology. What if you tried to count sand grains with your naked eye—could you even come close? Perhaps you might weigh a second’s worth of ‘sandfall’ after collecting it with a spoon—this estimate will be substantially better. Ultimately, however, you’ll want to use the best that technology can offer: a laser counter and imaging system automated by a high-end computer.
When physicists and geologists need to measure the decay rate of a radioactive element, they employ similar techniques. On the one hand, they could set up counters that track how many particles decay over a known period of time (when atoms decay, they emit characteristic bursts of energy; see below). This technique is limited, however, by the accuracy of the counters and the short time frame of the experiment, and it may even suffer from interference by other “energy-emitting” processes. Alternatively, radiochemists could prepare a sample of known mass of a radioactive element. After an extended period of time (up to several decades), a mass spectrometer is used to measure how much of the daughter element has accumulated in the sample. This approach is potentially more accurate, but no mass spectrometer is perfect, and so each estimate comes with error bars.
When multiple techniques provide us with similar numbers, however, then we gain confidence that we are closing in on the correct answer. Similarly, we might expect our estimate to get better with time with technological advancements. If more recent determinations converge toward a single value for a physical constant—whether the speed of light, gravitational constant, distance to the sun, or the decay rate of a particular radioisotope—then we know that science is working in our favor. The nature of research is to reevaluate, reprove, and retest, which means there will always be uncertainty. Thus our goal is to minimize this uncertainty, not to eliminate it.
Radioactive decay of 87-Rubidium: Snelling reminds us that science works
Last week, Snelling continued to falsify the scientific claims of Answers in Genesis with a new article on the Determination of the Radioisotope Decay Constants and Half-Lives: Rubidium-87 (87Rb). Therein, Snelling summarizes the careful attempts of radiochemists and geochronologists over the past six decades to determine the rate at which 87-Rubidium decays into 87-Strontium. If Snelling were preparing this paper for an undergraduate geochemistry course, then I would say that it reads quite well and is worth your time. He is very thorough and helpful in tracing both the methods and conclusions of researchers through time, which unsurprisingly have varied. If you can picture the disparity between electronics and computers around 1948 vs. 2012, then you can immediately understand why the estimated half-life of 87Rb reported in scientific literature has been ‘modified’ and updated repeatedly.
In Table 1 of Snelling’s paper, he provides a comprehensive list of decay-rate estimates for 87Rb, reported between 1948 and 2012. The accompanying figures 1 and 2 display these estimates graphically according to when they were published. It should be immediately apparent that scientists had a more difficult time counting particle decay prior to the 1970’s, when nearly all estimates were made via liquid scintillation counting. From 1950–1970, liquid scintillation counters underwent rapid technological advances, such as interference reduction and computer automation, driven especially by opportunities in nuclear medicine. Since liquid scintillation was not invented to address the needs of geoscientists (it derived from wartime attempts to image radioactive fluids), we can expect early analyses to have been relatively crude.
From a scientific perspective, it is therefore encouraging to note that half-life estimates for 87Rb varied only by ~9.6% over six decades, despite early technological challenges. Another feature of this plot is that values converge toward a single value as we approach the modern day—strong evidence that science is working. Since 1975, estimated half-life varies by only 0.38%, which is already better than the precision of mass spectrometers used for radiometric dating (i.e. small enough that it doesn’t affect age estimates). If we narrow the overview only to the last 14 years—the era of high-end computing and ultra-sensitive electronics—then the estimated half-life varies by less than 0.28%. In geology, it rarely gets better than this.
At this point, we should therefore expect Dr. Snelling to concede that since we know the decay rate of 87Rb very accurately and precisely, radiometric dating is equally reliable unless a different key assumption may be called into question. But citing the existing discrepancy between half-life estimates over the past ~30 years, Snelling writes:
These discrepancies only serve to highlight that the Rb-Sr dating method cannot be absolute when the 87Rb decay rate has not been accurately determined… Therefore, Rb-Sr dating cannot be used to discredit the young-earth creationist timescale.
I am now partially convinced that Andrew Snelling has long abandoned the young-Earth worldview, but continues to write for them because it is his favorite job. After presenting pages of evidence in favor of conventional radiometric dates, he inserts a de facto pronouncement to the contrary, as though he’s tempting his colleagues to catch on. For the record, I claim no authority as a judge of Snelling’s sincerity, but personally I find it difficult to explain his writings otherwise.
In one graphic, Snelling proves the accuracy of radiometric dating
Though in Table 1, Snelling reports the error estimates associated with half-life determinations, he does not display them in the figure. Furthermore, he telescopes the y-axis (depicting half-life) so as to exaggerate the discrepancies between published studies. If we plot Snelling’s summary data in a more honest and objective manner, we can better envision good science in action. Note that with time, the error bars get smaller (due to more sensitive equipment), and the data converge on the modern accepted half-life of 87Rb. Nearly all estimates are within a 2-σ window of each other (the 95% confidence interval, or two error bars from the data point):
More impressive, however, is the precision with which modern observational techniques (decay counting over several decades) can predict historical scientific applications. Some of the more recent half-life estimates were obtained by comparison to other radiometric dating techniques, much like calibrating the hourglass to the sundial. To accomplish this, geologists took the U-Th-Pb ages of meteorites and various Earth rocks, which were also dated by the Rb-Sr method. Since the half-lives of uranium isotopes are known more precisely, they asked: what half-life of 87Rb would cause the Rb-Sr age to be the same as the U-Th-Pb age?
Using my analogy from before, this is like asking: how many grains of sand must fall every second for the whole hourglass to empty in exactly one hour (as determined by the sundial)? To answer, we must assume that the sundial can accurately measure that hour for us. As it turns out, if we assume the accuracy of U-Th-Pb dates from meteorites, then the estimated half-life of 87Rb is within 0.2% of that calculated by direct counting of beta particles and long-term (“in growth”) decay experiments. Modern observations for one decay system accurately predicted the age estimation according to the other. In this manner, historical science is testable, repeatable, and falsifiable, but Rb-Sr dating has passed that test. Since 0.2% is less than the analytical uncertainty of mass spectrometers used to measure the modern isotopic ratios, this minute discrepancy between half-life estimates is too small to affect the estimated age of minerals. Therefore, the Rb-Sr may be considered an ‘absolute’ geochronometer, which thoroughly discredits the young-Earth paradigm.