Granite plan: granite granitic rocks are classified according to the qapf diagram for coarse

Granitization[edit]
Granitization is an old, and largely discounted, hypothesis that granite is formed in place through extreme metasomatism. The idea behind granitization was that fluids would supposedly bring in elements such as potassium, and remove others, such as calcium, to transform a metamorphic rock into granite. This was supposed to occur across a migrating front. However, experimental work had established by the 1960s that granites were of igneous origin.^[36] The mineralogical and chemical features of granite can be explained only by crystal-liquid phase relations, showing that there must have been at least enough melting to mobilize the magma.^[37]
However, at sufficiently deep crustal levels, the distinction between metamorphism and crustal melting itself becomes vague. Conditions for crystallization of liquid magma are close enough to those of high-grade metamorphism that the rocks often bear a close resemblance.^[38] Under these conditions, granitic melts can be produced in place through the partial melting of metamorphic rocks by extracting melt-mobile elements such as potassium and silicon into the melts but leaving others such as calcium and iron in granulite residues. This may be the origin of migmatites. A migmatite consists of dark, refractory rock (the melanosome) that is permeated by sheets and channels of light granitic rock (the leucosome). The leucosome is interpreted as partial melt of a parent rock that has begun to separate from the remaining solid residue (the melanosome).^[39] If enough partial melt is produced, it will separate from the source rock, become more highly evolved through fractional crystallization during its ascent toward the surface, and become the magmatic parent of granitic rock. The residue of the source rock becomes a granulite.
The partial melting of solid rocks requires high temperatures and the addition of water or other volatiles which lower the solidus temperature (temperature at which partial melting commences) of these rocks. It was long debated whether crustal thickening in orogens (mountain belts along convergent boundaries) was sufficient to produce granite melts by radiogenic heating, but recent work suggests that this is not a viable mechanism.^[40] In-situ granitization requires heating by the asthenospheric mantle or by underplating with mantle-derived magmas