Two possible models are available for the generation of komatiitic mel
ts from upper mantle sources: (1) large extents of melting at relative
ly low pressures and (2) low extents of pseudoinvariant melting at hig
h pressures. For the partial melting of olivine-dominated upper mantle
and generation of komatiitic magmas, the [Mg]-[Fe] diagram calculated
from olivine-melt K-D data is potentially useful in estimating the ph
ysical conditions of melting and in evaluating Fe/Mg ratios of mantle
sources. Application of this diagram to a suite of komatiitic amphibol
ites derived by different extents of melting of similar sources from t
he Kolar Schist Belt, south India, indicates that (1) their magmas wer
e derived at pressures ranging from 3 to 7 GPa by different extents of
melting; (2) melt generation occurred in adiabatically rising deep ma
ntle diapirs; and (3) diapiric mantle sources had higher Fe/Mg ratios
than the undepleted Lesotho garnet Iherzolite (PHN 1611). Komatiitic r
ocks from other Archean terranes in the [Mg]-[Fe] diagram suggests a r
ange of pressures and degree of melting that agree with those proposed
from high pressure experimental studies. Although some komatiitic roc
ks have geochemical features and conditions of melting that may requir
e residual garnet in the magma source regions, such geochemical featur
es can also be inherited from sources without much residual gamet. Thi
s is possible if the sources had undergone high pressure melt addition
at the time of melting.