Drafting Papers: The Study ​Anomalies

Anomalies comprise a class of novelties (and/or limitations, depending on the context) that need to be especially carefully addressed. MiltonWainwright, of Sheffield University, often used to remark that ‘anomalies should be cherished’. By this he meant that when results that conflict with previous results, or orthodox views, are obtained they should be carefully recorded and considered. In such cases people often try to find technical reasons (e.g., faulty instruments or inconsistencies in sampling or handling samples) that may explain the anomalies. For example, scientists using a Dobson Spectrophotometer detected the ‘hole’ in the ozone layer believed to be caused by Chlorofluorocarbons (CFCs) before it was detected by satellite data from the Total Ozone Mapping Spectrometer (TOMS) and the Solar Backscatter Ultraviolet (SBUV) instrument using more robust technology (NASA 2009), but they decided that their readings were due to instrumental error, thus missing a chance for an extremely high impact publication.

Like limitations, when one records anomalies, they need to be treated carefully, for several reasons. If they conflict with strongly entrenched views, they are clearly potentially highly important, but many referees who hold the entrenched views may raise fierce objections to them and seek to repress their publication. There may even be strong objections from non-scientific quarters. For this reason Galileo faced fierce opposition from many philosophers and clerics for supporting Copernicanism, and when he published hisDialogue concerning two chief world systems (1632) he was found to be vehemently suspect of heresy and spent the rest of his life under house arrest. Similarly, Darwin delayed publication of On the Origin of Species (1859) for several decades, because he knew it strongly conflicted with Creationist of the universe and thus would cause an outcry in religious communities that hold such views.

In other cases, anomalies may conflict with results that your own research group has already published, suggesting that the previous results were incorrect or at least that more factors than those considered affect the observed phenomena. In such cases, ideally one would conduct further studies to resolve the apparent conflicts, but this may not be possible (e.g., if the project has finished and there are no funds for further experiments).

There are two possible strategies that can be followed in such situations. One is to present all of the findings, including the anomalies, and postulate reasons for the observed discrepancies with previous results. This strategy has the great virtue that it might provide good grounds for further grant applications. However, there is often a considerable risk that referees might decide that the postulated hypotheses are interesting, that they should be tested and the results of the tests should be incorporated in the paper before it is accepted (which, as mentioned, may not be possible). In such cases, the chances of publication may be increased by making the conflict the central focus of the paper (i.e., one could state in the Abstract and Introduction that Previous studies have found that levels of cruciferoids (a fictitious class of plant growth regulators) form lateral gradients, across cambial meristems, and vertical, acripetal gradients in stems of crucifers. However, in the study presented here, although we found corroborative lateral gradients, basipetal rather than acripetal gradients were found in the stems. Possible reasons for the discrepancy with previous findings are discussed.

The other strategy is to publish selective results (e.g., the corroborative results regarding the lateral gradient, perhaps accompanied by morphological observations or analyses of the effects of manipulating it using external applications of cruciferoids), ignoring conflicting results.

We know of cases in which both strategies have been successfully applied. The latter can be easily criticized for being unscientific, but alternatively it could be validly argued that no incorrect data are being presented, and it is highly possible that explanations for the anomalies will be forthcoming. Furthermore, if one acquires funding, one can re-examine the gradients and test possible explanations for the discrepancies (and hence publish another paper).

It should also be noted that interesting anomalies or novelties are sometimes ‘buried’ in papers, that is, mentioned very briefly, without emphasis. If a study has numerous novel features, this will not matter. However, for studies that lack much novelty, highlighting any novel (or strange, anomalous) features may be sufficient to attract a referee’s or editor’s attention sufficiently to decide that they should be published. For instance, if a study of factors associated with diabetes only detected the expected correlations between rates of the disease and weight, age, gender and so on, it would be unlikely to be published in a highly ranked journal. However, if the study also found that  there was a consistent, weak but highly significant correlation between rates of the disease and the house numbers of the residences of the subjects, this should be highlighted, although the associated variation in rates was weak, since it would be so surprising. Hence, the paper would be more likely to be accepted, provided of course that the statistical analysis was sound.


Blackwell, J., & Martin, J. (2011). A scientific approach to scientific writing (pp.16-17). Springer Science & Business Media.

Comments are closed.