[IUCr Home Page] [Acta C] [CIF Home Page]

Co-editor Preview of Proposed Data-Validation Tests

Details of the data checks, which will soon be applied to all CIF submissions before a DVN is assigned, are available here. Co-editors are asked to look at the detail of these tests before they are fully implemented, and made available to authors.

Feedback on these tests is welcomed. However, before firing off your particular preferences on the various error thresholds currently set in these tests, PLEASE take into account that they represent the instilled wisdom (and compromises) of several people with particular knowledge on the topic and an understanding of the current practices. Those of us who have been labouring on this aspect of the tests understand only too well that it is unlikely that we will achieve unanimous agreement on all error thresholds; some will find the current levels too high, others too low. Nevertheless, if you feel strongly that the present limits are unrealistic or unworkable, I would like to hear from you immediately.

It should also be emphasized that the DVN tests are designed to meet the standards expected for studies based on current technologies and methodologies. As such they may be a challenge to some authors using older techniques - or for those not varying the diffraction experiment to suit the compound under study. One of the reasons for the DVN tests is to raise the overall standard of studies in the structural science community.

Note that the alert levels used in these tests mean the following.

ALERT (Level C)
indicates a marginal outlier which the Co-editor (or the author during a pre-check) should look at carefully. Discretion should be exercised in seeking its correction because it may reflect an insensitivity in the test. When correlated with alerts from other checks it may signal a more serious problem.

CAUTION (Level B)
indicates a failure at a more serious level. The value is a definite outlier and should be examined carefully, and probably corrected. It may constitute grounds, individually or collectively, for a significant revision. If coupled with other CAUTIONs it may constitute grounds for a rejection.

WARNING (Level A)
indicates the test has failed and the value is well below the acceptance standard. In the robotic mode it may cause the automatic return of the submission, and Co-editors may not get to see submissions containing these warnings. However, until this mode is implemented, Co-editors should use this signal, unless there are mitigating circumstances, as probable grounds for rejecting the submission or for requesting a major revision.

Syd Hall
syd@crystal.uwa.edu.au