In recent years, speculation on the origin of “alpinotype” ultramafic rocks has tended to ignore the field relationships of the rocks, and in particular, their very common association with basic volcanic rocks, or greenschists. In New Zealand, the almost invariable association of ultramafic intrusions and basaltic volcanics is so striking that it has dominated the thinking of field geologists on the origin of these rocks, and some have proposed that they represent ultramafic submarine flows contemporaneous with the Permian volcanics. However, pronounced mineralogical layering, and other textural features suggest that gravitational differentiation has played a part in their formation, and high temperature metamorphism at unfaulted contacts excludes cold, solid intrusion. Geophysical data suggest that they have a lopolithic form and a thickness of 5,000–10,000 ft. for the Dun Mountain and Red Hill masses. A detailed study of both ultramafic rocks and the surrounding volcanics has shown that there are chemical and mineralogical similarities between the ultramafic and volcanic rocks in any particular area. In the Permian sequence in New Zealand, large thicknesses of andesitic-basaltic volcanics form pronounced belts on both limbs of a syncline, marginal to the main Upper Paleozoic and Mesozoic New Zealand geosyncline, and a number of individual ultramafic bodies are found within both belts. On the eastern limb, olivine-poor tholeiitic basalts with very low K 2 O Na 2 O ratios of 0.09, enclose the ultramafic intrusions of Dun Mountain and Red Hill in which the clinopyroxenes show features common to pyroxenes from tholeiitic basalts and their derivatives (low Al z , K 2 O, Na 2 O and TiO 2 ). The pegmatitic phase in these ultramafic rocks is a pyroxenite, and there is very little sulphide mineralisation associated. On the western limb, the basic volcanics have an average K 2 O Na 2 O ratio of 0.40, and the associated ultramafic bodies show more alkaline features with Ti-rich pyroxenes and primary hornblende. The pegmatitic phase in these rocks is a hornblendite, and much of the known sulphide mineralisation and platinum in New Zealand is associated with ultramafics of the western belt. I have previously suggested that the New Zealand ultramafic masses represent the magma chambers of basaltic and andesitic volcanoes in a Permian double volcanic arc, analogous to the present-day arcs of the Circum-Pacific Belt. The depths of the magma chambers, and hence the levels now exposed by erosion, are dependent on the composition of the original magma, and the crustal thickness in the area. It is possible that in more deformed areas, the ultramafic portions of the magma chambers could be displaced, either horizontally or vertically, by faulting.