Publication of the results of research is an essential part of science. The number of publications is sometimes used to assess the work of a scientist and as a basis for distributing research funds (see "publish or perish"). In order to achieve a high rank in such an assessment, there is a trend to split up research results into smaller parts that are published separately, thus increasing the number of publications. Semi-ironically, this process has been described as splitting the results into the smallest publishable units.
"Salami publication," or "salami slicing," is a variant of the smallest-publishable-unit strategy. In salami slicing, data gathered by one research project are separately reported (wholly or in part) in multiple end publications. Salami slicing, apparently named by analogy with the thin slices made from larger pieces of salami meat, is generally considered questionable when not explicitly labeled, as it may lead to the same data being counted multiple times as apparently independent results in aggregate studies.
When data gathered in one research project are partially reported as if a single study, a problem of statistical significance can arise. Scientists typically use a 5% threshold (0.95 probability) to determine whether a hypothesis is supported by the results of a research project. If multiple hypotheses are being tested on a single research project, 1 in 20 hypotheses will by chance be supported by the research. Partially reported research projects must use a lower threshold when testing for statistical significance but often do not do this.
There is no consensus among academics about whether people should seek to make their publications least publishable units. Particularly for people just getting started in academic publication, writing a few small papers provides a way of getting used to how the system of peer review and professional publication works, and it does indeed help to boost that all-important publication count. But publishing too many LPUs is thought not to impress peers when it comes time to seek promotion beyond the assistant professor (or equivalent) level. Also, LPUs are not an efficient way to pass on knowledge, because they break up ideas into small pieces, forcing people to look up many cross-references. On the other hand, a small piece of information is easily digestible, and the reader may not need more information than what is in the LPU.