A number of diorite complexes occur within the Channel Islands region, notably on Jersey, Alderney, and particularly Guernsey. Much of northern Guernsey is made up of the largest of these complexes (fig. S1), the Bordeaux diorite. In the north-western part of this diorite, around Chouet, a complicated association of plutonic rocks occurs. Although the field relationships in this area are sometimes difficult to interpret—this is often the case in diorite complexes—three separate groups of rocks may be distinguished within the association: a diorite group; a granodiorite group; and an inhomogeneous suite of rocks (fig. S2).

The widespread diorite group consists predominantly of an even-grained diorite, which is relatively homogeneous but which occasionally grades into an acicular diorite, the latter often containing pods and veins of appinite. The granodiorite group is the least common, occurring as bodies which are interpreted as intrusive sheets and bosses within the even-grained diorite, but occurring as angular blocks within the inhomogeneous suite of rocks. The granodiorite invariably contains rounded diorite xenoliths. The inhomogeneous suite consists of a variety of rocks from patchy, dark diorite, through quartz diorite to tonalite. Commonly, these rock types are intimately associated, often showing gradational contacts with each other and frequently with the more basic portions occurring as ‘xenolithic’ material within the more acidic portions. At contacts between the inhomogeneous suite and the even-grained diorite certain features (e.g. lobate margins and pipe-like structures) indicate that the diorite must have been close to its solidus temperature at the time of emplacement of the inhomogeneous suite. The field relationships between the three groups are interpreted as indicating that the diorite group was emplaced first, followed by the granodiorite group, with both of these clearly pre-dating the inhomogeneous suite.

Fifty-nine specimens, chosen to give a representative sample of each of the three rock groups, have been analysed for major, minor, and a selection of trace elements and thirteen of these specimens have been analysed for REE. The chemistry of the analysed rocks confirms the division into three groups, with each group showing distinctive characteristics. Furthermore, chemical plots (e.g. Al2O3, P2O5, Cr, and Ni v. SiO2) show discontinuities and areas of overlap between each group which cannot be explained within the constraints of a single genetic model relating the three groups to each other. This argument is particularly strong for the relationship between the diorite group, which spans the range 50 to 59% SiO2, and the inhomogeneous suite, spanning the range 53 to 68% SiO2. In the area of overlap (53 to 59% SiO2) the two groups are geochemically different. Therefore, for a variety of reasons, including emplacement order, the geochemical characteristics of the groups and the lithological inhomogeneity which is associated only with the chemically intermediate members of the association (the inhomogeneous suite), three quite different and genetically unrelated liquids are required to generate the three groups of rocks.

The even-grained diorite shows chemical variation (e.g. with increasing SiO2, decreasing Al2O3, MgO, CaO, Sc, V, Cr, and Ni, and increasing Na2O, La, Nd, and Y) consistent with amphibole + plagioclase fractionation up to 55% SiO2. At 55% SiO2 several elements show a change of slope (e.g. FeO + Fe2O3, TiO2, Rb, Ba, and Zr) indicating the introduction of biotite as a fractionating phase. Increasing total REE content with increasing SiO2 throughout the even-grained diorite supports the contention that amphibole is an important fractionating phase. The higher TiO2, P2O5, Sr, La, Ce, Nd, and Y contents and negligible Cr and Ni contents of the acicular diorite suggest an origin by delayed crystallization of volatile-enriched portions of the diorite group magma.

The granodiorite group shows little geochemical variation. Members of this group contain detectable amounts of Cr and Ni, unlike virtually all members of the inhomogeneous suite. For this reason, and because of the field relationships, the granodiorite is considered to be genetically unrelated to members of the inhomogeneous suite and a separate liquid is thus required for its genesis. This liquid may have been the fractionated derivative of some other magma (though if this is so the ‘parent’ is entirely unrepresented at the present erosion level) or it may represent a direct crustal melt. Diorite xenoliths within the granodiorite are chemically similar to the even-grained diorite.

Despite the lithological complexity of the inhomogeneous suite, its geochemical unity is clearly established in that, for instance, virtually none of the members of the suite (including even the most SiO2-poor) contain detectable Cr and Ni. Moreover, geochemical variation within the group is rational (with the possible exceptions of Sr, Zr, and Ba) and may be explained in terms of a crystal fractionation model. However, the fractionation must have acted on a liquid itself unrelated to either the diorite or granodiorite group magmas. An additional complication is that later derivative liquids intrude into and partly digest earlier-formed semi-solids of the suite to produce much of the observed inhomogeneity. The phases which have controlled fractionation within the suite include plagioclase (established petrographically as well as geochemically) and hornblende. The role of apatite is uncertain. The fractionation of hornblende is particularly useful in explaining the change in REE contents within the inhomogeneous suite. Total REE contents increase from the dark diorite to the quartz diorite, but decrease from the quartz diorite to the tonalite with concomitant relative HREE depletion. This is taken to be a reflection of the changing hornblende/liquid partition coefficients for REE with increasing SiO2, which are less than one for liquids of basaltic and andesitic composition but greater than one for liquids of dacitic composition.