Monday, September 22, 2008

John Hartnett's Cosmos. 1. Introduction

This is the first entry in what will certainly be a multi-part series of articles on the claims of Australian creationist John Hartnett (see Hartnett's entry in Conservapedia). Since my first piece on Hartnett's work appeared (My Visit to the Creation Forum at McLean Bible Church and a Review of "Hubble, Bubble, Big Bang in Trouble"), I have received a number of inquiries about his work.

In this series, I'll start with a discussion of Hartnett's claims of redshift periodicities. The claim that extragalactic redshifts are distributed in fixed steps, i.e. quantized, is not new. William Tifft of the Steward Observatory at the University of Arizona is probably the originator of the modern redshift quantization movement back in the 1970s. If real, such periodicities or quantization would be very difficult to explain for conventional Big Bang cosmology. This is why you find redshift quantization supported by advocates of other alternative cosmologies such as the Electric Universe (see Electric Cosmos).

I choose to examine the redshift periodicities issue largely because my own graduate work focussed on time-series analysis of very noisy datasets and I accumulated a fair amount of experience working with Fourier series, transforms and power spectra.

Dr. Hartnett has published two papers on the Cornell Preprint server covering the topic of extragalactic redshift periodicities. I'll refer to them as Paper I and Paper II.

  • Paper I: "Galaxy redshift abundance periodicity from Fourier analysis of number counts $N(z)$ using SDSS and 2dF GRS galaxy surveys" by John G. Hartnett, Koichi Hirano (arXiv:0711.4885)

  • Paper II: "Redshift periodicity in quasar number counts from Sloan Digital Sky Survey" by John G. Hartnett (arXiv:0712.3833)

It's interesting that the recently posted third version (v3) of paper I not only has an additional author, but seems to advocate a radically different cosmological model than the second version (v2). In the v2 paper, Hartnett advocated Moshe Carmeli's 5-dimensional cosmological model where the Hubble expansion was made part of the metric. Hartnett published several additional papers based on this model claiming it could explain Dark Matter as well. In paper v3, Hartnett has switched to another model developed by Hirano, Kawabata, and Komiya. This may be because the Hirano et al. cosmology explicitly tries to explain alleged redshift periodicities. This is another reason for me to examine the quantization claims first, as Hartnett appears to be in the process of changing his cosmological model, but quantization is a common component in both of them.

Hartnett is apparently using these works to gain him credibility in the creationist community as a professionally-published cosmologist. In his creationist publications, he has invoked redshift quantization as evidence of Galactocentrism, a feature of his young-universe cosmology. Some graphics similar to those from earlier versions of Hartnett's papers appear in his book Starlight, Time and the New Physics.

Since the advent of the Fast Fourier Transform and the availability of fast personal computers, the ability to compute the power spectral density (PSD) of a dataset has become much easier. Unfortunately, this increased ease of use does not come with an increased understanding of just what the PSD does. In cases with large amounts of low noise data, the PSD can identify well-defined frequencies in the data with relative ease. In other cases, such 'intuitive' understandings of the PSD can easily lead one astray.

Hartnett makes numerous erroneous statements on properties of PSD, suggesting he is relying on his 'intuition' on how the PSD works instead of actually testing the claim. Most researchers, myself included, must demonstrate that our test protocols work for datasets of known content before making such grandiose claims when applying the test to datasets of unknown content. In the abstract for Paper I, Hartnett states that his results "indicate that this is a real effect and not some observational artifact." Yet he has apparently conducted no tests to determine which characteristics of his results are analysis artifacts.

In future entries in this series, I will present an overview of some of Dr. Hartnett's errors in these two papers. In later entries, I'll include some samples of how scientific tests are themselves tested. Some components may be difficult to communicate in this blog since there is limited support for graphics and mathematical notation. Some of these components may take a while to assemble as I may present code snippets (using Python, numpy, scipy, and similar tools) so the reader may explore the analysis themselves. Comments and feedback are certainly welcome.

Sunday, September 7, 2008

Radioactive decay rates depend on Earth-Sun distance?

Reports are floating around the scientific blogs (such as at arXiv Blog) about a new paper recently published on the Cornell Preprint server. Evidence for Correlations Between Nuclear Decay Rates and Earth-Sun Distance, by Jenkins, Fischbach, et al., has created a stir in the physics community. It may eventually rank up there with the Pioneer Anomaly in the 1990s and reports of a Fifth Force in the 1980s.

The results are intriguing as they involve measurements conducted at two different sites, Brookhaven National Laboratory (BNL) and Physikalisch-Technische-Bundesandstalt (PTB) in Germany, as well as two different isotopes, 32Si (a beta emitter) and 226 Ra (an alpha emitter), respectively. 36Cl (a beta emitter) was used as a calibration source in the BNL experiment. Two independent sites rule out a local variation at the experimental site. The BNL experiment also exhibits this annual variation in the ratio of 32Si to 36Cl, suggesting there must be an isotopic difference in the response of the nuclei to this effect. The authors suggest a solar effect, reported in another paper, Perturbation of Nuclear Decay Rates During the Solar Flare of 13 December 2006.

I've not yet heard of any Creation “Scientists” claiming that their theories predict this type of annual effect, but I suspect it is only a matter of time before they do. Reports of even the most miniscule changes in radioactive decay rates excite creationists a great deal since much of radioisotope dating depends on relatively constant rates of radioactive decay. I've examined a number of creationists claims of variation in radioactive decay rates (see Claims of Accelerated Radioactive Decay but have found none that specifically predict variation based on the Earth-Sun distance. Like psychics, such successful predictions by creationists will be ex post facto. Lest the creationists get too excited, it should be noted that both experiments exhibit this variation at the 0.2% level (above and below a constant value), nowhere near large enough to turn a 4.5 billion year old Earth into a 6000 year old Earth.

There are a number of interesting aspects of the Jenkins et al. analysis that hint this is a measuring systematic. One is that the amplitude of the variations seems to vary significantly from one year to the next. It is not perfectly correlated with the Earth-Sun distance. Another interesting issue is that the oscillation appears to cover the same range (0.2%) when examining the 226Ra decay as well as the ratio of the 32Si/36Cl decay. Is the 36Cl calibration source unaffected by this phenomenon? Perhaps 32Si is unaffected and 36Cl has an inverse correlation? If 32Si and 36Cl were identically affected, we would expect them to cancel out. As noted by the authors, there also appears to be a significant phase shift between the Earth-Sun distance and decay variation. This suggests an effect driven by another variable which may be coupled to the annual variation. Finally, the data are from the 1980s so it may be difficult to re-examine some of the techniques used to test for other systematics. Of course, the fact that this process seems to occur over a relatively short annual time scale will make it easier to verify today.

On a similar historical note, it was realized in the 1970s that no one had examined if the Newtonian inverse-square law operated at laboratory distances as well as it worked on solar system scales and larger. Previous researchers had determined the value of Newton's 'G' under the assumption that the inverse-square law was valid. When the historical data were finally examined with that assumption removed, the data suggested that at short ranges, gravity was not an inverse-square law (see Why do we believe Newtonian gravitation at laboratory dimensions? by Daniel R. Long). This launched a search in the 1980s and 1990s for a possible 'fifth force', in addition to the four fundamental forces known to physics.

It is interesting that the second author in the Jenkins paper, Ephraim Fischbach, is one of the researchers involved in the Fifth Force search (see Ten Years of the Fifth Force by Ephraim Fischbach). Interest in a fifth force died down as improvements in measurement technology seem to verify inverse-square at very short ranges (see Laboratory tests of gravity).

I suspect there are similar systematic measuring errors in the current Jenkins work as well. It is nonetheless an intriguing observation that definitely requires verification from other sources.