A dwarf galaxy may be defined in terms of absolute magnitude and/or linear size. There is, however, a continuous transition between giant, average, and dwarf galaxies. The separation must rest on some arbitrary demarcation lines. As far as we know, lenticulars and spirals in the Sa-Sc range are all brighter than MT -16; dwarf systems fainter than -16 occur only among ellipticals and late-type spirals Sd-Sm or Magellanic irregulars Im, but of course not all E or Im systems are dwarfs. For instance, by this definition the Large Magellanic Cloud (MT = -18.1) is certainly not a dwarf and the Small Cloud (MT = -16.0) is a borderline case, but M32 (MT = -15.6) and NGC 205 (MT = -15.8) qualify as dwarfs. Similar systems in the Virgo cluster are NGC 4486B, the bright compact dE0 companion of M87, and IC 3475, the prototype of the low-luminosity diffuse ellipticals described by Reaves (1956), which are both at MT = -16.0. The low-luminosity dwarf ellipticals of the Local Group exemplified by the Sculptor and Fornax systems are much fainter, at MT = -11.2 and -12.9, respectively, and the extreme dwarf globular-like systems discovered by Wilson (1955) and Zwicky (1957) are fainter still-for example, Leo II is at MT = -9.1.
The lower end of the scale is indefinite, and for all we know (or rather do not know) ``pygmy'' systems of even smaller populations down to isolated star-cluster size might exist and remain undetected throughout intergalactic space as Zwicky has often argued (1957).