Many readers will recall frustration on reaching the last paragraphs of an exciting paper only to stumble across an anticlimactic referral to enticing 'unpublished data' or 'data not shown' in the discussion section. Individual papers can only achieve conceptual advances of finite size and a good study will raise as many questions as it answers. We like when the discussion of a paper clearly outlines some of the next key questions to be addressed, and realize that these questions are often already being addressed. However, adding 'data not shown' to key implications can come across as somewhat disingenuous: if the data adds to the conceptual advance presented in the published paper, it should be included in the paper and formally peer reviewed. Conversely, if it is not essential, why mention it at all? The reader is left feeling that the authors are trying to stake a claim on a topic without having data that would hold up to the scrutiny of their peers. It is much better to just flavour the discussion with the next big questions — even if the work is already in progress.

An altogether different category of 'data not shown' relates to the core data of the paper; for example, data for one cell line is presented and the supporting data from four others is 'not shown'. Why, especially when many journals host online supplementary information? If the data warrants being mentioned to support the conclusions made, then it should usually be presented to back up the claim. How else can the reader judge if the generality of the conclusions hold, or if the authors have viewed their supporting data through rose tinted glasses. Even worse is when controls (such as for antibody specificity) are 'not shown'. This is as unacceptable as the excuse 'your format restrictions are too tight'.

This leaves data that is submitted for the eyes of peer reviewers and editors only. Again, we would discourage this approach. If a referee requests unnecessary or irrelevant data, surely the author should argue their case, rather than trying to appease an off-base referee (unless the data happens to be available). If the data is important, but part of another planned study, the author should publish the other study first or 'bite the bullet' and transfer the data. Our readers have a right to judge the same data set that was evaluated by our referees.

Finally, a note on 'personal communications'; assuming these are attributed and permission has been obtained, these have to be admissible, but only where they underscore further reaching statements in the discussion, or as an entry point into a study — not to bolster core data in the paper.

Online publishing should have made 'data not shown' largely a thing of the past and we encourage our authors to show what is necessary to present a definitive and well rounded piece of work of the sort we hope you expect to find in this journal.