Astronomy Editing and Proofreading Services

A detection is announced at three sigma. It may be real. It may also be one of the twenty things you looked for, in which case finding one at three sigma is exactly what noise does — and the difference between those two papers is not the data. It is whether the search was specified in advance and whether the non-detections were reported with the same care as the detection.

We edit what astronomers write — journal articles and their results sections, detection, non-detection and upper limit reporting, uncertainty and error budget documentation, observing proposals and their justification, instrument and pipeline description sections, survey and catalogue papers, transient and follow-up circulars, data release documentation, simulation and modelling papers, and press and public communication of results. Our editors work on the reporting of what was not found.

Upper limits and the honest treatment of significance are what separate a durable astronomical result from a retracted one, and the failure is a detection reported without its search. A three-sigma feature in one of twenty spectra is not a three-sigma result. We work through these so the number of independent trials is stated wherever a significance is claimed, since this single number converts a marginal detection into either evidence or noise and it is routinely omitted; so upper limits are reported with their confidence level and the assumption behind them, because an upper limit that assumes a line width or a spectral shape is conditional and the condition travels with it; so the error budget is itemised into statistical and systematic components with the dominant term named, given that a result whose uncertainty is dominated by calibration is improved by better calibration and not by more integration; so the non-detections are reported in the same table as the detections, as a survey reporting only its positives cannot be used by anybody modelling the population; so the pipeline version and the calibration applied are stated, since reprocessing changes results and the literature contains disagreements that are entirely versioning; so any cut applied to the data is stated with how many objects it removed; and so a marginal result is described as marginal in the abstract rather than in the discussion. Papers written this way survive reanalysis.

Everything you send is treated in confidence, including unpublished data, proposals and pipeline documentation. We are editors rather than astronomers or statisticians, and we offer no view on detections, significance or instrumentation. What we can do is make the search as visible as the finding.