A Tertiary, dacitic volcanic land-form in Kelowna, British Columbia, shows layering that has not been recognized elsewhere. Layering is expressed as thin (0.5 m) layers separated by thick (4.5 m) layers exposed along a weathered fault scarp. The major elements show that both thick and thin layers are dacitic and geochemically very similar. Trace element modelling indicates that thin layers formed from thick layers via crystal fractionation involving removal of plagioclase, biotite and magnetite in the proportions 75:20:5, and with only 12% fractionation. The thin layers represent segregation veins generated during crystallization of the dacite. They formed when the crystal mush at the bottom of the upper crust successively, thermally contracted, fractured and foundered, siphoning evolved interstitial liquid from the mush into the horizontal crack. Cooling of the segregation veins led to further fracturing. Later, fluids following these fractures altered the thin layers and precipitated secondary carbonate minerals. The altered thin layers weather preferentially, thus visually accentuating the small primary chemical differences between thick and thin layers. The scale of layering, mode of formation and differentiation mechanisms appear different from those in felsic magma chambers and it is unclear how common this phenomenon is. However, similar layering is more easily identified and commonly developed in mafic lava flows.

The chemistry and textures of the minerals in the layered syenites of Nunarssuit and West Kûngnât record a complex post-crystallisation history. The initial cumulus phases were alkali feldspar, pyroxene and olivine, apatite and opaque oxides. Grain size distributions of the pyroxenes and olivines have been modified since crystallisation, possibly during transport of these grains in a density current, prior to deposition in layers. Olivines and pyroxenes re-equilibrated with intercumulus melt. The lower ratio of mafic phases to intercumulus fluid in leucocratic layers has resulted in pyroxene and olivine in these layers undergoing more re-equilibration that in melanocratic layers, consequently pyroxenes and olivines in leucocratic layers are more ferroan. Pyroxenes reacted with either intercumulus or deuteric fluid to form amphibole. In West Kûngnât pyroxene and amphibole re-equilibrated via a later fluid. Olivines reacted with deuteric fluid to produce biotites, probably at temperatures < 550°C. After crystallisation alkali feldspars exsolved, the exsolution textures coarsened in the presence of deuteric fluids with a magmatic origin, possibly at temperatures as low as 450°C. Zoning patterns of apatites and zircons were modified during interactions with either, or both intercumulus or deuteric fluids. Leucocratic layers underwent more modification than melanocratic ones, probably due to the higher concentration of intercumulus melt in the former leading to a higher concentration of deuteric fluids as water exsolved from the magma during cooling. The syenites underwent little compaction during solidification.

The rhythmic modal layering in the Nunarssuit syenite has a vertical extent of >150 m and a lateral extent of >15 km. Individual layers average 20 cm thick and grade from a relatively melanocratic base into more leucocratic syenite over a distance of up to 5 cm. The major cumulus phases are alkali feldspar, ferro-salite/hedenbergite and fayalite. Two basic stratigraphic cycles have been identified in which faint modal layering becomes more pronounced up section, each cycle terminating in a thick melanocratic zone. Slumps, slump breccias, troughs, micro-rhythmic layering and one occurrence of crossed layers were observed.

Qualitative grain size analysis indicates no size sorting in the layers. Preliminary application of crystal size distribution theory to ferro-salite/hedenbergite and fayalite from the bases of individual layers gives results which may be interpreted as indicating a relative lack of coarse grains. If the layers were deposited from density currents it would be expected that the coarsest grains would be deposited close to the source of the currents. There is no evidence in the majority of the syenite that the cumulus pile underwent compaction during crystallization.

There was little, or no, primary chemical variation across individual layers. Whole-rock compositions and the ferro-salite/hedenbergite, fayalite, biotite and amphibole present in the syenites show a slight, but statistically significant, increase in the ratio Mg/(Mg+Fetotal), from the base up to the top of the layered succession.

A model is suggested in which successive magma layers become more ferroan towards the top of the chamber. Cooling is concentrated at the top of the chamber and layers of magma crystallize sequentially, the uppermost, ferroan layers first. As layers of magma cool and crystallize they sink, as crystal-melt plumes, to the bottom of the chamber where they source density currents from which layers are deposited.