Is there a chance that the anisotropies in the cosmic microwave background detected by COBE are an artefact of COBE's image reconstruction?

Abstract

The COBE time ordered data (TOD) was reanalysed for anisotropies in the cosmic microwave background with a novel technique that is much simpler and more robust than used in the official analysis of the COBE TOD. The technique extends the statistical concept of rejecting the null hypothesis to image reconstruction by asking if the TOD is consistent with no anisotropies outside the galactic band. If it is assumed that the instrumentation noise is ideal, as appears to be the case in the official analysis, the null hypothesis can be rejected to 28 sigma, a highly significant detection. However, the official analysis presents no information supporting the ideal noise assumption to the level required to reject the null hypothesis. The interaction of COBE's calibration error of roughly 3.0% with the brightest regions of the galactic band in combination with the reconstruction algorithm may have introduced artefacts roughly of the size and spectral signature of the 10ppm anisotropies reported by COBE. Data from follow up missions have similar problems with calibration errors that may have resulted in similar sky maps to COBE. It is pointed out that either a few hours of repetitive circular scans using the WMAP satellite or the single beam antenna of the soon-to-be-launched Planck satellite should provide definitive evidence as to the existence or non existence of the 10ppm anisotropies without the use of image reconstruction.