The term phylum was coined in 1866 by Ernst Haeckel from the Greek phylon (φῦλον, "race, stock"), related to phyle (φυλή, "tribe, clan"). In plant taxonomy, August W. Eichler (1883) classified plants into five groups named divisions, a term that remains in use today for groups of plants, algae and fungi. The definitions of zoological phyla have changed from their origins in the six Linnaean classes and the four embranchements of Georges Cuvier.
Informally, phyla can be thought of as groupings of organisms based on general specialization of body plan. At its most basic, a phylum can be defined in two ways: as a group of organisms with a certain degree of morphological or developmental similarity (the phenetic definition), or a group of organisms with a certain degree of evolutionary relatedness (the phylogenetic definition). Attempting to define a level of the Linnean hierarchy without referring to (evolutionary) relatedness is unsatisfactory, but a phenetic definition is useful when addressing questions of a morphological nature—such as how successful different body plans were.
The most important objective measure in the above definitions is the "certain degree" that defines how different organisms need to be to be members of different phyla. The minimal requirement is that all organisms in a phylum should be clearly more closely related to one another than to any other group. Even this is problematic because the requirement depends on knowledge of organisms' relationships: as more data become available, particularly from molecular studies, we are better able to determine the relationships between groups. So phyla can be merged or split if it becomes apparent that they are related to one another or not. For example, the bearded worms were described as a new phylum (the Pogonophora) in the middle of the 20th century, but molecular work almost half a century later found them to be a group of annelids, so the phyla were merged (the bearded worms are now an annelid family). On the other hand, the highly parasitic phylum Mesozoa was divided into two phyla (Orthonectida and Rhombozoa) when it was discovered the Orthonectida are probably deuterostomes and the Rhombozoa protostomes.
This changeability of phyla has led some biologists to call for the concept of a phylum to be abandoned in favour of cladistics, a method in which groups are placed on a "family tree" without any formal ranking of group size.
A definition of a phylum based on body plan has been proposed by paleontologists Graham Budd and Sören Jensen (as Haeckel had done a century earlier). The definition was posited because extinct organisms are hardest to classify: they can be offshoots that diverged from a phylum's line before the characters that define the modern phylum were all acquired. By Budd and Jensen's definition, a phylum is defined by a set of characters shared by all its living representatives.