Possible oscillations in high-precision measurements of Newton's gravitational constant -A reassessment based on the generalized Lomb-Scargle periodogram

• Published in: Europhysics letters Vol. 113; no. 2; p. 20001
• Main Authors: Scholkmann, F., Sieber, O. D.
• Format: Journal Article
• Language: English
• Published: Les Ulis EDP Sciences, IOP Publishing and Società Italiana di Fisica 01.01.2016; IOP Publishing
• Subjects: 04.80.-y; 06.20.Jr; 06.30.Ft; Bayesian analysis; Data analysis; Datasets; Error analysis; Gaussian; Gravitational constant; Noise measurement; Oscillations; Phase relationships; Random noise; Spectra
• ISSN: 0295-5075; 1286-4854
• DOI: 10.1209/0295-5075/113/20001

The presence of an oscillation with a period of 5.9 yr in measured values of Newton's gravitational constant G over more than three decades, and of a correlation with a 5.9 year oscillation in the length of day (LOD) variability, was recently reported by Anderson et al. (EPL, 110 (2015) 10002). A reanalysis based on an improved data set of measured G values was conducted by Schlamminger et al. (Phys. Rev. D, 91 (2015) 121101(R)) with the result of supporting the finding of a low-frequency oscillation present in the G measurements (with a period of ). A subsequent reanalysis by Anderson et al. (arXiv:1505.01774 [gr-qc]) using the improved data set of Schlamminger et al. confirmed the presence of the oscillation. However, the phase relationship changed (G and LOD not in phase anymore). In an additional analysis, Pitkin showed by Bayesian model selection that the oscillation is most probably due to chance since the data can be modelled at best with the assumption that the scattering of values is caused by measurement errors and an additional Gaussian noise term overlaid. In order to add to the analysis of possible oscillation in G data sets the aim of our work was to reanalyze the data based on the data sets compiled by Schlamminger et al. using the generalized Lomb-Scargle (GLS) periodogram (and the Lomb-Scargle (LS) periodogram, as a control) with additional bootstrapping-based statistical testing. We found periods of and in all the investigated data sets; however, the corresponding peaks in the spectra did not reach statistical significance. We therefore conclude that there is not enough statistical evidence that these oscillations are not due to chance -a finding in agreement with the work of Pitkin.