Recommended nomenclature of epidote-group minerals

Deer et al., (1986) concluded in their Rock-forming Minerals: “There is no universally accepted nomenclature of the monoclinic Fe-Al members of the epidote group. Some earlier names, e.g. fouquéite for compositions containing up to about 10 mol. % of the Ca₂Fe₃Si₃O₁₂(OH)¹

The nomenclature problems discussed by Deer et al., (1986) were only partly solved after publication of the criteria for new mineral names (Nickel & Mandarino, 1987). Are these “rules” only applicable to new minerals (the corresponding paragraph was entitled “criteria for new mineral names”)? What about the “grandfathered” mineral names epidote and clinozoisite described in the past and generally believed to represent valid species names? How are the boundaries between epidote and clinozoisite defined?

Of the 22 “piemontite” analytical data reported by Deer et al., (1986) only seven meet the minimum criterion of Fe < Mn > 0.5 required to satisfy the rule given by Nickel & Mandarino (1987): “At least one major structural site is occupied by a different chemical component.”

Ercit (2002), in his article “The mess that is allanite”, asks the provocative question: What is “allanite”? He states that the name “allanite” is often incorrectly used to describe any REE-bearing epidote. Investigators were additionally confused by introduction of new mineral names for REE-bearing epidote-group minerals such as dissakisite-(Ce), dollaseite-(Ce), khristovite-(Ce), and androsite-(La). Until 1988, the mineralogy of REE-rich epidote-group minerals was very simple because there was only the one official root name allanite with the end-member formula CaREEAl₂Fe²⁺[Si₂O₇][SiO₄]O(OH). An additional source of confusion (Ercit, 2002) is that cations in the formula of REE-rich epidote-group minerals are commonly grouped rather than assigned to specific sites. However, specific site assignments determine the species. In the absence of appropriate guidelines, Ercit (2002) discusses several approaches for using electron-microprobe analytical data to calculate formulae of REE-bearing epidote-group minerals. Then he uses analytical data from the literature to obtain a simplified end-member formula corresponding to a mineral name. There are two major criticisms to his approach: (1) Analytical data from the literature that had been obtained on metamict materials were used, which do not fulfill the criterion that a mineral be crystalline; (2) Standard guidelines for mineral nomenclature (Nickel & Grice, 1998) were uncritically applied, resulting in end-member formulae [e.g., CaLaAl₂V³⁺[Si₂O₇][SiO₄]O(OH)] that are not charge-balanced and thus meaningless.

The Commission on New Minerals and Mineral Names (CNMMN) of the International Mineralogical Association (IMA) established at the beginning of 2003 the Subcommittee on Epidote-Group Mineral Nomenclature. This subcommittee defined following aims: (1) Development of a consistent nomenclature system to minimize proliferation of unrelated mineral names; (2) to explain appropriate use of existing and new names within this mineral group by defining simple rules for nomenclature based on chemistry and ion (cation and anion) order; (3) to provide a simple but powerful scheme to derive a “working name” of an epidote-group mineral based on electron-microprobe analytical data.

Finally, this report supplies an Appendix with specifically selected chemical data of epidote-group minerals representing either borderline cases, or incomplete or erroneous analytical data, or data originating from partly metamict minerals. These examples have been chosen to demonstrate naming of epidote-group minerals in non-trivial cases.

It is not the intent of this nomenclature recommendation to provide a complete view of chemical variations reported for epidote-group minerals. For additional references on this subject we refer to Liebscher & Franz (2004).

Epidote, a monoclinic mineral with the idealized formula Ca₂Al₂Fe³⁺[Si₂O₇][SiO₄]O(OH), was named by Haüy (1801). The type locality is Bourg d'Oisans, Dauphiné, France. The name comes from the Greek epi over plus dotós given (verbid of didóni), i.e., given besides, “increased”. A related noun is epidosis, i.e. “increase”. This refers to Haüy's observation that the base of the mineral's prism has one side longer than the other. The rationale for the name epidote may appear quaint to us today, but it has to be understood in the context of criteria for distinguishing minerals at the beginning of the 19^th century (mainly crystal forms, density, and optical properties, as well as chemical composition). Dana (1896) stated that Haüy set aside several older names: Thallite, derived from the Greek noun thallos meaning young twig (alluding to the green color), was rejected because of color variation; delphinite, and arendalite (also named akanticone) were rejected because they were derived from specific localities. A synonym for epidote, pistacite, comes from the Greek for pistachio nut, a reference to the distinctive yellowish-green color of some epidote. The name pistacite was introduced by Ludwig (1803–1804), who extended the systematic nomenclature of Abraham Gottlob Werner (1750–1817). This timing gives epidote precedence. Pistacite is not listed among accepted mineral names (in spite of its frequent use) by Hey (1955) but referenced as synonym of epidote. Other obsolete names (Dana, 1896) are oisanite, puschkinite (six other spellings have also been used, e.g., pushklinite in Dufrénoy, 1856), achmatite, beustite and escherite. Withamite was used for a strongly pleochroic deep crimson and straw yellow epidote-group mineral from Glen Coe, Strathclyde (Argyllshire), Scotland, UK and sometimes referred to as piemontite. However, chemical analyses gave only 0.24 wt.% MnO – equivalent to 0.27 wt. % Mn₂O₃-_- (Dana, 1896). Thus this low Mn³⁺ content is already sufficient for piemontite-like pleochroism. Subsequent chemical analyses of “withamite” (Hutton, 1938) clearly suggest its identity with epidote. Withamite from the Yamanaka mine (Japan) described by Yoshimura & Momoi (1964) is manganian clinozoisite.

Weiss (1820) presented the first complete indexing of crystal faces of monoclinic epidote, which may be considered the beginning of systematic crystallographic work on this mineral group. Dufrénoy (1856) used epidote as the group name and distinguished thallite (green iron-bearing epidote) and zoisite on the basis of different cleavage and habit. Ito (1950) and Ito et al., (1954) first deduced the correct atomic arrangement in epidote, which they showed to contain both single SiO₄ tetrahedra and double-tetrahedral Si₂O₇ groups (cf. Ito, 1947). Systematic structural studies by Dollase (1968, 1969, 1971) clarified the crystal-chemical relationships of the epidote group.

The name clinozoisite was given by Weinschenk (1896) to Fe-poor epidote from Prägratten, Tyrol, Austria. The name is for the polymorphism with zoisite, Ca₂Al₃[Si₂O₇][SiO₄]O(OH) (see below). However, Lacroix (1889) was probably the first to describe this mineral under the name fouquéite. We quote Dana (1896): “Composition like zoisite from which it differs in form; it appears to be an epidote essentially, containing but little iron … Occurs in anorthite-gneiss at Salem, and less often at Kandy, Ceylon. The rock also contains ordinary epidote (but not immediately associated with fouquéite) … Named for M. Fouqué.” The analysis reported by Lacroix (1889) is given in Table 1.

According to the structural studies of Ito et al., (1954) and Dollase (1968, 1969, 1971) monoclinic members of the epidote group have three distinct octahedral sites (M1, M2, M3), each contributing with the same multiplicity to the formula. If one of these sites in an unnamed mineral were found to be dominantly occupied by a chemical component that is not dominant at the equivalent site in an existing mineral, then the unnamed would be a new mineral species with its own name (Nickel & Mandarino, 1987). Application of this rule requires detailed crystal-chemical knowledge of the various mineral groups and particularly of the cation site-preference. In the epidote group, if by chance ferric iron did not order onto the largest and most distorted octahedral site M3, but were randomly distributed over the three available sites, the composition Ca₂Al₂Fe³⁺[Si₂O₇][SiO₄]O(OH) could not be given an independent mineral name. Instead, it would have to be called ferrian clinozoisite because Al would be dominant at all three octahedral sites (M1, M2, M3). However, structure refinements and spectroscopic investigations of epidote-group minerals have consistently shown a strong preference of Fe³⁺ for M3 (e.g., Ito et al., 1954; Dollase, 1971, 1973). Thus, the old name epidote is retained.

Originally this mineral was named saualpite for the locality Saualpe in Carinthia, Austria, where it occurs in eclogites. The name zoisite was chosen by A.G. Werner in 1805 to honor Siegmund Zois, Baron von Edelstein (1747–1819), the Austrian mineral collector from whom Werner obtained the holotype specimen from Saualpe (Dana, 1896). Zoisite is the orthorhombic polymorph of clinozoisite. Other obsolete names or synonyms cited by Dana (1896) are: Illuderite, lime-epidote, thulite, unionite. Hey (1955) also cited chrome-zoisite and manganese-zoisite. Tanzanite is a gem name for vanadium-bearing zoisite from Tanzania, which turns from brown into blue upon heat-treatment at 400–500°C. A dense green chromium-bearing zoisite in a zoisite amphibolite associated with mostly non-transparent ruby also from Tanzania was named “anyolite” meaning green in the language of the Massai (Eppler, 1984). Ito (1950) first proposed a structural model for zoisite, which was subsequently confirmed by Fesenko et al., (1955) and Dollase (1968). Zoisite is the only orthorhombic mineral species originally assigned to the epidote group.

A Mn-rich, epidote-related mineral from Praborna mine, St. Marcel, Aosta Valley, in the Italian Western Alps, was named piemontite (originally “Piemontit”) by Kenngott (1853). The English spelling of the Italian word Piemonte is Piedmont, and Dana (1896) arbitrarily anglicized the mineral name to piedmontite. However, the Italian region of the type locality, St. Marcel, was called Piemonte when the mineral was first described, and for this reason the name piemontite is correct. Today, St. Marcel belongs to Valle d'Aosta, which was established as an autonomous region in 1945. Dufrénoy (1856) stated that the “variété manganésifère” from Piemonte is named “piemonite”. Hey (1955) considered both piedmontite and manganepidote to be synonyms of piemontite.

Dollase (1969) demonstrated by single-crystal X-ray structure refinement that Mn³⁺ in piemontite, ideally Ca₂Al₂Mn³⁺[Si₂O₇] [SiO₄]O(OH), is preferentially ordered on the octahedral M3 site, thus confirming the interpretation of spectroscopic data by Burns & Strens (1967). Nonetheless, there seems to be a general problem with the assignment of the name piemontite (Catti et al., 1989). Dana (1896) included under the name piemontite monoclinic, dark-red epidote-group minerals with characteristic red – pink to amethyst- orange to yellow pleochroism, even if the molar Mn₂O₃ content was less than Fe₂O₃ content. Mottana & Griffin (1986) showed that piemontite from the type locality (the Praborna manganese mine) is rather heterogeneous. Nonetheless, the compositions of most samples fall in the vicinity of Al:Mn = 2:1 with a general trend of some excess Mn extending to Al:Mn = 1.3:1. In addition, Mottana & Griffin (1986) found that piemontite from St. Marcel is frequently strontian piemontite with up to 0.47 Sr pfu. Catti et al., (1989) and Ferraris et al., (1989) reported structural data on this strontian piemontite.

Strontiopiemontite, ideally CaSrAl₂Mn³⁺[Si₂O₇][SiO₄]O(OH), is the analogue of piemontite with Sr dominant in the structural site A2, as shown by X-ray single-crystal structure refinement (Bonazzi et al., 1990). It is a low-temperature metamorphic product and occurs in the manganese ore at Molinello and Cassagna mines of Val Graveglia, Northern Apennines, Italy.

Tweddillite, CaSrMn³⁺AlMn³⁺[Si₂O₇][SiO₄]O(OH), is a mineral related to strontiopiemontite (Armbruster et al., 2002). It is found as a hydrothermal alteration product of primary manganese ore in the Wessels mine of the Kalahari manganese field, Republic of South Africa. It was named tweddillite in honor of S. M. Tweddill, the first curator (from 1897 to 1916) of the Museum of the Geological Survey at Pretoria, RSA. Definition of this new mineral species was justified, as shown by structure refinement, because Mn³⁺ dominates both octahedral M1 and M3 sites, different from just M3 in strontiopiemontite and piemontite.

The name “tawmawite” was introduced by Bleeck (1907) in his description of jadeite deposits in Upper Burma (now Myanmar). Tawmaw was a major jadeite mining-district in this area at the beginning of the 20^th century. Bleeck (1907) described “tawmawite” as an emerald green, chromium-rich epidote mineral. However, the chemical analysis did not correspond to epidote-group stoichiometry, mainly due to contamination with chromite. The existence of a mineral with idealized composition Ca₂Al₂Cr³⁺[Si₂O₇][SiO₄]O(OH) has been shown by electron-microprobe analyses of epidote-group minerals occurring as inclusions in calcic plagioclase in a kyanite amphibolite from Southern Alps, New Zealand (Grapes, 1981), and in a quartzite from Outokumpu, Finland (Treloar, 1987). Chromiumrich epidote-group minerals are commonly zoned on a very fine scale (oscillatory zoning) so that determination of specific site preference of Cr using crystal-structure refinement has not been possible to date. Burns & Strens (1967) provided spectroscopic evidence that Fe³⁺, and Mn³⁺ in epidote-group minerals order preferentially on M3, but assumed that Cr³⁺ orders preferentially on M1. However, Liebscher (2004) concluded that the derived Δ_oct with the position of V_II (Burns & Strens, 1967) results in a Racah parameter B that is higher than that of the free Cr³⁺ ion and therefore physically meaningless. The interpretation of the spectra by Burn & Strens (1967) is thus open to question. Armbruster & Lahti (in prep.) have recently performed a combined electron microprobe – crystal-structure study on V³⁺-rich “tawmawite” from Outokumpu, Finland, first described by Eskola (1933). Preliminary results indicate that Cr is disordered over M3 and M1 (with a slight preference for M3), which cast doubts on the species character of “tawmawite” if the Cr content is below 1 Cr pfu. To our knowledge the highest Cr content in chromian clinozoisite was reported as 15.37 wt.% Cr₂O₃ (Treloar, 1987), corresponding to 0.98 Cr pfu, which is still slightly below the limit required for highly disordered partition of Cr between M1 and M3. On the basis of available data “tawmawite” cannot be considered a valid species. Such samples have to be described as chromian (or better Cr³⁺-rich) clinozoisite until new evidence is presented.

Mukhinite, Ca₂Al₂V³⁺[Si₂O₇][SiO₄]O(OH), was described by Shepel & Karpenko (1969) from marbles (Gornaya Shoriya, Kemerovo Oblast, Siberia, Russia). The name is for the geologist A.S. Mukhin of the West Siberian Geological Survey. In the absence of crystallographic data ordering of V³⁺ on M3 may be assumed by analogy with Fe³⁺. Structural data including cation site-distributions are highly desirable for mukhinite.

Hancockite is another old name. Penfield & Warren (1899) named a Pb-rich monoclinic epidote-group mineral discovered at Franklin, N.J., USA after Elwood P. Hancock (1836–1916) of Burlington, N.J., a collector of Franklin minerals. At the type locality the mineral contains appreciable amounts of Sr, and of Mn³⁺, the latter being responsible for the strong red color (Dunn, 1985). Holtstam & Langhof (1994) reported a second occurrence of this very rare species from Jakobsberg, Filipstad, Sweden (skarn enclosed in dolomitic marble). Hancockite, CaPbAl₂Fe³⁺ [Si₂O₇][SiO₄]O(OH), may be considered an epidote (sensu stricto) with Ca on A2 substituted by Pb.

Niigataite, CaSrAl₃[Si₂O₇][SiO₄]O(OH), is related to clinozoisite but with Ca on the A2 site substituted by Sr (shown by single-crystal X-ray structure refinement). The name is for the Japanese prefecture where the mineral was discovered. Niigataite was found in a boulder of prehnite rock, where the mineral occurs interstitially with diaspore and chlorite in close association with strontian clinozoisite (Miyajima et al., 2003).

In the following text REE represents the lanthanides (elements 57 to 71) and Y because of its chemical similarity to the lanthanides.

Historically, “allanite” (Thomson, 1810) is the second monoclinic mineral of the epidote group with a name that is still in use. The name is for Thomas Allan (1777–1833), a Scottish mineralogist who discovered the mineral. The type material is allegedly from Iglorsoit, East Greenland. However, studies of Giesecke's diaries from 1806 suggest that the type locality is most likely Qaqarssuatsiaq, Aluk, East Greenland (Petersen & Johnsen, 2005). Dana (1896) listed following synonyms or varieties of “allanite”: Cerine, bucklandite, tautolite, uralorthite, bagrationite, orthite, xanthorthite, pyrorthite, and wasite. Those names describe partly altered “allanites” or solid-solution members between “allanite” and epidote from specific localities with more or less characteristic crystal forms. In addition, the Y-rich minerals muromontite (5.52 % BeO) and bodenite were considered varieties of “allanite” (Dana, 1896). The latter two names are nowadays no longer listed among minerals and their relation to “allanite” is also under question [for a discussion on muromontite and Be in “allanite” cf. Grew (2002)]. Nagatelite is supposedly a phosphatian “allanite” (Iimori et al., 1931). However, X-ray diffraction data are mandatory to confirm this relationship. Hey (1955) referenced also treanorite as synonym for “allanite” and additionally the following “allanite” varieties: Cerepidote, cerorthite, yttrio-orthite, magnesium-orthite, and mangan-orthite. The “orthite” vs. “allanite” controversy was resolved in 1986 by the Commission on New Minerals and Mineral Names, IMA, in favor of “allanite”. Taking into account Levison's nomenclature (1966) for rare-earth minerals, the mineral from Greenland originally described by Thomson (1810) is actually allanite-(Ce) and, citing that priority, the CNMMN formally approved the name allanite-(Ce) (Nickel & Mandarino, 1987). Cerium is the dominant REE in most “allanite”. However, chemical data are available in the literature for La- and Y-dominant “allanite”, which should be properly considered as distinct mineral species. Their status has not until recently been formally approved, although the corresponding mineral names allanite-(La) and allanite-(Y) have been in circulation (Levinson, 1966) and are included in the IMA list of minerals. With the aim of securing official status for such unapproved epidote-group minerals, Orlandi & Pasero (2006) recently defined the “new” mineral allanite-(La), which has now the status of an IMA-approved species (proposal # 2003–065).

Lombaardite was first described by Nels et al., (1949) from the Zaaiplaats tin mine, central Transvaal, and re-examined by Neumann & Nilssen (1962) reporting ca. 10–15wt.% (REE)₂O₃. Neumann & Nilssen (1962) suggested that lombaardite is very similar or even identical to a Y₂O₃ dominant “allanite” (with 22.16 wt.% (REE)₂O₃ including 5.39 wt.% Y₂O₃) from a pegmatite at Åskagen, Värmland, Sweden. Thus the Åskagen, Värmland, sample represents allanite-(Y).

“Allanite” is characterized by one coupled heterovalent substitution where the higher charge of (REE)³⁺ replacing Ca on the A2 site (Dollase, 1971) is compensated by ferrous iron occupying the octahedral M3 site. Peacor & Dunn (1988) restudied material first investigated by Geijer (1927), which was originally named “magnesium orthite” and therefore considered the Mg-dominant analogue of “allanite”. However, chemical analyses accompanied by crystal-structure refinement showed that the true composition of this material is close to CaREEMgAlMg[Si₂O₇][SiO₄]F(OH), an epidote-group mineral characterized by two different types of coupled heterovalent substitutions: (1) ^A2(REE)³⁺ + ^M3Mg → ^A2Ca + ^M3Al and (2) ^M1Mg + ^O4F− → ^M1Al + ^O4O²⁻. This new type of REE-rich epidote-group mineral was given the root name dollaseite in honor of Wayne Dollase for his crystal chemical research on epidote-group minerals. The type material for dollaseite-(Ce) is from the Östanmossa mine, Norberg district, Sweden, originally studied by Geijer (1927).

However, Geijer (1927) also analyzed Mg-rich material with F < 0.5 apfu, which seemed close to the Mg analogue of “allanite”. Subsequently, in a literature review Grew et al., (1991) found several reports of “allanite” analogues with Mg > Fe²⁺ and compositionally distinct from “dollaseite”. The corresponding mineral with the end-member formula CaCeAl₂Mg²⁺[Si₂O₇][SiO₄]O(OH) was named dissakisite-(Ce) from the Greek dissakis = “twice over” in reference to the Mg equivalent of “allanite” being described twice. The type material was found in marble from Balchen Mountain in the eastern Sør Rondane Mountains, Queen Maud Land, Antarctica. The type material of the newly discovered dissakisite-(La) is from a peridotite body of the Ulten zone, Austroalpine domain, Eastern Alps (Tumiati et al., 2005).

Sokolova et al., (1991) described the structure of a Ce-rich epidote-group mineral from the Inyl'chek Massif, Kyrgyzstan (former Kirghiz SSR), which was also F-rich and thus closely related to “dollaseite” but with the octahedral M3 site dominated by Mn²⁺. Pautov et al., (1993) later defined this mineral as a new species, khristovite-(Ce), CaCeMgAlMn²⁺[Si₂O₇][SiO₄]F(OH), which was named after the Russian geologist Evgenia Vladimirovicha Khristova.

It has been known for a long time that piemontite may incorporate significant amounts of REE (for a literature review see Bonazzi et al., 1992). Bonazzi et al., (1996) defined a new end-member from Andros Island, Cyclades, Greece named androsite-(La) with the end-member composition Mn²⁺LaMn³⁺AlMn²⁺[Si₂O₇][SiO₄]O(OH). This REE-rich epidote-group mineral with La on A2 has the smaller of the two A sites (A1) occupied by Mn²⁺ and the largest octahedral site (M3) also occupied by Mn²⁺, whereas the M1 octahedron is dominated by Mn³⁺. In addition, epidote-group minerals with compositions corresponding to vanadoandrosite-(Ce) and (mangani)androsite-(Ce) (Cenki-Tok et al., 2006) have been approved by CNMMN (IMA 2004–015 and IMA 2002–049).

Ferriallanite-(Ce), ideally, CaCeFe³⁺AlFe²⁺[Si₂O₇][SiO₄]O(OH), is the analogue of allanite-(Ce) with Fe³⁺ dominant in the octahedral M1 site. It is of metasomatic origin and occurs in an alkaline granitic pegmatite of Mount Ulyn Khuren in the Altai Range, Mongolia (Kartashov et al., 2002). Furthermore, ferriallanite-(Ce), previously reported as “cerine” or iron-rich “allanite” is the most common lanthanide mineral next to cerite-(Ce) at the Bastnäs Fe-Cu-REE deposit, Skinnskatteberg, Västmanland, Sweden (Holtstam et al., 2003).

As exemplified by the minerals gatelite-(Ce) (Bonazzi et al., 2003) and västmanlandite-(Ce) (Holtstam et al., 2005) the epidote-type structure (E) easily matches with that of törnebohmite, (REE)₂Al[SiO₄]₂(OH), (T) to form a polysomatic series (E, ET, T). Sequences in addition to ET are to be expected because E and T modules fit together in any order and the E-T interface does not require significant structural distortions. Both gatelite-(Ce) and västmanlandite-(Ce) represent iso-topological ET type polysomes distinct by space-group symmetry and composition of the epidote-type module. The E module in gatelite-(Ce) is of dissakisite-(Ce) composition whereas the E module in västmanlandite-(Ce) is of dollaseite-(Ce) composition. To clarify the structural relationship between E and T modules Bonazzi et al., (2003) have chosen a unit cell setting for gatelite-(Ce) where a_gat. ≈ 2a_epi. ≈ [201]_tör.; b_gat. ≈ b_epi. ≈ b_tör.; c_gat. ≈ (c_epi. + c_tör.); β_gat. ≈ β_epi. Unfortunately, a corresponding setting was not used for västmanlandite-(Ce) by Holtstam et al., (2005). Their original setting, a = 8.939, b = 5.706, c = 15.855 Å, β = 94.58°, space group P2₁/m may be transformed by the matrix [−100 | 010 | 101] to obtain a setting in space group P2₁/m with a = 8.939, b = 5.706, c = 17.568 Å, β = 115.90° corresponding to gatelite-(Ce) with a = 17.770, b = 5.651, c = 17.458 Å, β = 116.18° (space group P2₁/a).

The oldest, still accepted mineral name of the mineral group under review is epidote (Haüy, 1801). For this reason the name epidote is used to not only describe a mineral species of idealized Ca₂Al₂Fe³+[Si₂O₇][SiO₄]O(OH) composition but also to designate the entire group. The subcommittee discussed the point whether zoisite should be regarded a member of the epidote group (e.g., Franz & Liebscher, 2004). However, given that ca. 20 mineral species are presently recognized as isostructural with epidote but zoisite stands alone, the subcommittee decided to limit the epidote group to closely related species having monoclinic symmetry so as to avoid having to mention repeatedly the one exception having orthorhombic symmetry. Exclusion of zoisite from the epidote group has a historical precedent (Tschermak, 1905) and was followed in the most recent edition of Strunz Mineralogical Tables (Strunz & Nickel, 2001).

An epidote-group mineral is described with the generic formula A₂M₃[T₂O₇][TO4](O,F)(OH,O). The monoclinic crystal structure is composed of T₂O₇ (usually Si₂O₇) and TO₄ (usually SiO₄) units linked to two kinds of chains (parallel to the b-axis) built by edge-sharing octahedra (Fig. 1). One chain consists of M2 octahedra while the other chain is formed by M1 octahedra with M3 octahedra attached on alternate sides along its length. M octahedra are mainly occupied by trivalent ions such as Al, Fe³⁺, Mn³⁺, Cr³⁺, V³⁺. Divalent cations (e.g., Mg, Fe²⁺, Mn²⁺) may occupy M sites (preferentially M3) if various heterovalent substitutions come into play. M2 has a strong preference for Al whereas the occupancy of M1 and M3 depends on competing ions. Usually an OH group is bonded to the M2 cation. The overall structural arrangement gives rise to two types of cavities, a smaller one named A1, usually occupied by Ca or Mn²⁺ and a larger one named A2, usually occupied by Ca, Sr, Pb, and REE. The resulting connectivity (topology) is consistent with space group P2₁/m.

The P2₁/m structure of epidote-group minerals displays ten symmetry independent anion sites, commonly labelled O1 – O10. In natural samples the site O10 represents O bonded to H (hydroxyl group) whereas the O4 site may be occupied by O²⁻ (clinozoisite and allanite subgroups) or by F- (dollaseite subgroup).

Zoisite is a polymorph of clinozoisite Ca₂Al₃[Si₂O₇][SiO₄]O(OH) of orthorhombic symmetry (space group Pnma) with only one type of edge-sharing octahedral chain and M3 octahedra attached only on one side along its length. Alternatively, zoisite and clinozoisite may even be interpreted to have polytypic relations (Ito, 1950; Merlino, 1990).

For better and easier distinction of epidote-group minerals we introduce three subgroups based on the existing accepted mineral species. Additional subgroups may be defined if new species are not accommodated in the following three.

Members of the clinozoisite subgroup are derived from the mineral clinozoisite Ca₂Al₃[Si₂O₇][SiO₄]O(OH) solely by homovalent substitutions. The key cation and anion sites are A1 = M²⁺, A2 = M²⁺, M1 = M³⁺, M2 = M³⁺, M3 = M³⁺, O4 = O²⁻, O10 = (OH)−. In other words, the dominant valence as listed above must be maintained.

Members of the allanite subgroup are REE-rich minerals typified by the eponymous mineral “allanite”. This subgroup is derived from clinozoisite by homovalent substitutions and one coupled heterovalent substitution of the type ^A2(REE)³⁺ + ^M3M²⁺ → ^A2Ca²⁺ + ^M3M³⁺. Thus the valences on the key sites are: A1 = M²⁺, A2 = M³⁺, M1 = M³⁺, M2 = M³⁺, M3 = M²⁺, O4 = O²⁻, O10 = (OH)−.

Members of the dollaseite subgroup are REE-rich minerals typified by the eponymous mineral “dollaseite”. This subgroup is derived from clinozoisite by homovalent substitutions and two coupled heterovalent substitutions of the type ^A2(REE)³⁺ + ^M3M²⁺ → ^A2Ca²⁺ + ^M3M³⁺ and ^M1M²⁺ + ^O4F− → ^M1M³⁺ + ^O4O⁻². Thus the valences on the key sites are: A1 = M²⁺, A2 = M³⁺, M1 = M²⁺, M2 = M³⁺, M3 = M²⁺, O4 = F-, O10 = (OH)−.

Gieré & Sorensen (2004) reviewed Th⁴⁺ and U⁴⁺ contents in “allanite”. According to their literature review the maximum reported ThO₂ content is 4.9 wt.% corresponding to ca. 0.07 Th pfu. The maximum UO₂ content is 0.82 wt.% in a crystal containing additionally 1.09 wt.% ThO₂, corresponding to 0.02 apfu each. Thus, Th and U are minor constituents of the actinides (ACT) group of elements. The recommended chemical criterion to assign minerals to the allanite and dollaseite subgroups is REE +ACT > 0.5 pfu. The major argument for adding U and Th to REE is that the coupled heterovalent substitution mechanism is presumably: ^A2(Th⁴⁺, U⁴⁺) + 2 ^M3M²⁺ → ^A2Ca + 2 ^M3M³⁺.

Comment: The IMA CNMMN guidelines for new minerals also list Sc³⁺ together with Y³⁺ and lanthanides as candidates for which the Levinson extension should be used. However, Sc³⁺ has also a strong preference for octahedral sites (E.g., bazzite is a beryl analogue where Sc replaces Al). For this reason a special treatment of Sc-rich epidote-group minerals seems necessary, and, if such compositions were found in nature, the Sc distribution will have to be studied.

Considering members of the clinozoisite subgroup, most frequent are homovalent substitutions on M3 by Fe³⁺, V³⁺, Cr³⁺, and Mn³⁺ (replacing Al) as well as on A2 by Sr, Pb²⁺, and less commonly, Ba (replacing Ca). These substitutions, in principle, yield 20 combinations each corresponding to a distinct mineral species. If we further consider that homovalent substitutions may also occur on M1 as exemplified by the mineral tweddillite (Armbruster et al., 2002) and that Mn²⁺ can also substitute for Ca on A1, the number of potential species increases further. For this reason we believe that the free choice of a mineral name needs to be restricted in order to avoid proliferation of unrelated mineral names. Relationships among minerals should be evident in the names. For example, the chemical relation between piemontite and strontiopiemontite is obvious, whereas the name tweddillite obscures its relation to strontiopiemontite.

In the recommended new nomenclature the key cation-sites M3 and A1 (and, in principle, M2) determine the root name. If the dominant cations at A1, M3 (and M2) exactly match those of an approved species, the same root name must be given. If at one of these sites the dominant cation is different a new root name may be suggested.

In both clinozoisite and allanite subgroups no prefix is added to the root name if M1 = Al. In the dollaseite subgroup no prefix is added to the root name if M1 = Mg. Otherwise a proper prefix derived from the name of a chemical element must be attached. The prefixes ferri, mangani, chromo, and vanado indicate dominant Fe³⁺, Mn³⁺, Cr³⁺, and V³⁺ on M1, respectively (clinozoisite and allanite subgroups). The prefixes ferro and mangano indicate dominant Fe²⁺ and Mn²⁺ at M1, respectively (dollaseite subgroup). The dominant cation on A2 in the clinozoisite subgroup (other than Ca) is defined as extended Levinson suffix (Levinson, 1966; Bayliss & Levinson, 1988). Note that A2 in the allanite and dollaseite subgroups is occupied by REE. Thus for those minerals Levinson suffixes per se apply to the dominant REE. This simple nomenclature requires complete renaming of following species:

• Niigataite (old) = clinozoisite-(Sr) (new),

• hancockite (old) = epidote-(Pb) (new),

• tweddillite (old) = manganipiemontite-(Sr) (new).

Fortunately, two of these species were very recently described and all three species are rare, so that the miner-alogical community is not familiar with the original names. Minor modifications are necessary for following species:

• Strontiopiemontite (old) = piemontite-(Sr) (new),

• androsite-(La) (old) = manganiandrosite-(La) (new).

Comments: “Androsite” is an example for a root name derived from the occupancy of two key sites A1 (Mn²⁺) and M3 (Mn²⁺). Additional root names are required for corresponding compositions with either A1 occupied by Ca or M3 occupied by any other divalent cation.

Kartashov et al., (2002) list analytical data for a mineral of idealized composition CaCeFe³⁺₂Fe²⁺ [Si₂O₇][SiO₄]O(OH) but the available data were not considered sufficient for naming a new species. On the one hand, such a hypothetical mineral would qualify for a new root name due to its unique occupancy of M3 and M2. The strong relation to ferriallanite-(Ce) could also be expressed by the possible name “ferriferriallanite-(Ce)” where the dominant occupancy on M2 is expressed by a second prefix. If the two prefixes are identical an alternate choice could be the name “diferriallanite-(Ce)”. Because there are yet no accepted examples for such species with M2 ≠ Al we defer the naming decision to the authors proposing the new species and to the CNMMN members reviewing the proposal. We believe that awkward doubly prefixed names should be the exception rather than the rule in epidote-group minerals.

Occupancy of the key sites and recommended names for all hitherto defined epidote-group minerals and for some examples of possible compositions not yet found are summarized in Tables 2–4. In addition, cell dimensions of approved species are given in Table 5.

Comment: In particular, our subcommittee discussed the use of proper prefixes for V³+ and Cr³⁺. The ending for both prefixes should be -o because V³+ and Cr³⁺ are the predominant low valences of V and Cr in mineral formulas (V²⁺, although reported in minerals, is not considered a potential constituent because it is so rare). We have chosen the prefix “chromo” for Cr³⁺ by analogy with the common use of this Greek root meaning color in scientific terms (chromophore, chromosome, etc.). For V³⁺ we have agreed on “vanado” although we were aware that the element name is derived from the goddess Vanadis in Norse mythology. “Vanado” was preferred over “vanadio” because this prefix implies “vanadii” would be the prefix for V⁴⁺. Both prefixes would lead to awkward mineral names, in particular if the root name begins with a vowel. We also discussed the possibility of hyphenation between prefix and root name for certain epidote-group mineral names and concluded that hyphenation should be avoided. Nevertheless, we concur with the IMA CNMMN guidelines (Nickel & Grice, 1998) and recommend that a hyphen be used if an unhyphenated name is awkward and the hyphen assists in deciphering the name.

The definition of mineral species of the epidote-group of minerals depends on the preferred occupancy of various structural sites. Therefore, submission of any new species to CNMMN must be accompanied by crystal-structure refinement and/or spectroscopic experiments indicating the extent of order of a chemical species on a given structural site. Exceptions are permissible only for new species distinguished by REE cation, or for simple compositions where the cation distribution is unambiguous from the chemical composition alone. In such cases high-quality chemical data would be considered sufficient. Given the role of F in the dollaseite subgroup, a fluorine analysis is mandatory for epidote-group minerals.

It is emphasized that all new species, for which names (e.g., “clinozoisite-(Pb)”, “epidote-(Sr)”, for additional examples see Tables 2–4) are already defined in this recommended nomenclature, require approval of a formal new-mineral proposal submitted to IMA CNMMN.

In order to name an epidote-group mineral, priority should be given to the choice of subgroup, then root name, and last the specific name. The starting point will be a table reporting the cations on A1, A2, M1, M2 and M3 and the anions on O4 and O10 (key sites).

For a proper subgroup assignment one has to determine the dominant valence at M3, M1 and A2 expressed as M²⁺ and M³⁺, not just specify a single dominant ion (i.e., Fe²⁺ or Mg or Al). In addition, the dominant valence on O4: X-or X²⁻ must be evaluated.

• [M²⁺]_A2 > 0.50, [M³⁺]_M3 > 0.50 → clinozoisite subgroup,

• [M³⁺+ M⁴⁺]_A2 > 0.50, [M²⁺]_M3 > 0.50 → allanite subgroup,

• {[M²⁺]_M3+M1 − [M³⁺+ M⁴⁺]_A2} > 0.50 and [X-]_O4 > 0.5 → dollaseite subgroup.

Comment: The structural refinement of both dollaseite-(Ce) and kristovite-(Ce) convincingly showed F- to be ordered at the O4 site. Thus, the presence of more than 0.5 F per formula unit corresponds to the dollaseite subgroup. Assigning F to O4 and not to O10 has major implications; substitution of F- for O²⁻ reduces total negative charge and thus must be balanced by decreasing the M³⁺: M²⁺ value.

The above definition of the dollaseite subgroup may at first glance appear unduly complex. A simpler scheme based solely on cation occupancy would be:

• [M²⁺]_A2 > 0.5, [M³⁺]_M1 > 0.5, [M³⁺]_M3 > 0.5 → clinozoisite subgroup (with [X-]_O4 < 0.5 implicit),

• [M³⁺+ M⁴⁺]_A2 > 0.5, [M³⁺]_M1 > 0.5, [M²⁺]_M3 > 0.5 → allanite subgroup (with [X-]_O4 < 0.5 implicit),

• [M³⁺+ M⁴⁺]_A2 > 0.5, [M²⁺]_M1 > 0.5, [M²⁺]_M3 > 0.5 → dollaseite subgroup (with [X-]_O4 > 0.5 implicit).

The above alternate scheme is not recommended although it is correct for ideal end-members. For members with 0.5 < (REE)³⁺ << 1.0 the scheme might fail. There are only two structure refinements addressing cation order on M1 and M3 in dollaseite subgroup minerals (Peacor & Dunn, 1988; Sokolova et al., 1991). These are not sufficient for us to provide a more rigorous definition of the dollaseite subgroup.

As mentioned in the introduction, the traditional distinction between clinozoisite and epidote is based on their optical character. Clinozoisite was defined as optically positive whereas epidote was shown to be optically negative. The change over occurs at ∼ 40 % epidote component (Ca₂Al₂Fe³⁺[Si₂O₇][SiO₄]O(OH)). In a similar way, the name piemontite has been commonly assigned to monoclinic, Mn-bearing epidote-group minerals exhibiting the characteristic pleochroism, even if the molar Mn₂O₃ content is less than Fe₂O₃ content.

There is a wealth of optical data on epidote-group minerals, recently summarized by Franz & Liebscher (2004). Crystal optics is a fundamental tool in analyzing and describing rocks and minerals, and good optical data may be used in some cases to identify a species or even to quantify a solid-solution member. Nevertheless, binary behaviour should not be assumed a priori for complex systems such as the epidote-group minerals. If the chemical composition becomes too complex optical data are ambiguous and have their limitations in mineral identification. As an example, different species of the allanite and dollaseite subgroups are very difficult to distinguish on optical grounds. For a proper evaluation of capabilities and limitations of optical methods for the identification of epidote-group minerals comparative studies are necessary.

The subcommittee, therefore, strongly recommends that optical criteria alone not be used for distinguishing epidote-group minerals. Although optical tests (e.g., determination of the optical sign, optical orientation or pleochroism) are useful for preliminary species identification, we recommend that electron-microprobe analytical data be used to determine a mineral name of a solid-solution member. The dominant cation at key structural sites, with the restriction described below, determines the name (e.g., Nickel & Grice, 1998; Nickel, 1992).

If only homovalent substitutions take place, the dominant cation at key structural sites simply determines the name. Coupled heterovalent substitutions in epidote-group minerals require a special treatment of the so-called 50 % rule in a solid-solution series. Even for the REE-bearing minerals, the A2 cation appearing in the suffix has to be selected from among the divalent cations in order to name a species belonging to the clinozoisite group. The root name depends on the dominant trivalent cation on M3.

Example: Consider the A2 occupancy (0.35 Ce, 0.05 La, 0.30 Ca, 0.20 Sr, 0.10 Pb). Because (REE)³⁺ < 0.5 the mineral belongs to the clinozoisite subgroup. Although Ce is the dominant cation on A2, the critical cation is Ca, the dominant divalent cation. No suffix is needed in this case because a suffix is only added for a dominant A2 cation other than Ca.

The above hypothetical mineral might have the M3 site occupied by 0.4 Mg, 0.25 Fe³⁺, 0.35 Al. Because this mineral is a member of the clinozoisite subgroup the dominant M³⁺ ion (i.e. Al, not Mg) is decisive for the root name of the species. Formal derivation of the proper name gives clinozoisite. The adjectival modifiers cerian and magnesian (Schaller, 1930; Bayliss et al., 2005) may be added to emphasize the high Ce and Mg contents.

This deviation from the common procedure of naming minerals is necessary because strict adherence to the rule based on the dominant ionic species leads to inconsistencies and unbalanced formulas. For example, the above mineral would have the idealized formula CaCeAl₂Mg[Si₂O₇][SiO₄]O(OH), which is the same as the formula for dissakisite-(Ce), a member of the allanite subgroup. If 0.1 Fe²⁺ were to replace 0.1 Mg, M3 composition becomes 0.3 Mg, 0.1 Fe²⁺, 0.25 Fe³⁺, 0.35 Al, and the idealized formula becomes CaCeAl₃[Si₂O₇] [SiO₄]O(OH). The latter formula is not charge balanced and thus nonsensical. Furthermore, the requirement to name clinozoisite-subgroup minerals on the basis of dominant M²⁺ on A2 and dominant M³⁺ on M3 reduces proliferation of unwanted mineral names.

For the sites involved in the charge compensation of a heterovalent substitution in A2 (i.e. M3 in the allanite subgroup, and also M1 in the dollaseite subgroup), the identification of the relevant end-member formula must take into account the dominant charge-compensating octahedral cation (M²⁺) and not the dominant cation in these sites.

Example: An allanite-subgroup mineral where M3 is not dominated by a single divalent cation but by several, so that a trivalent cation is the most abundant one: e.g., Ca(La_0.6Ca_0.4)Al₂(Fe²⁺_0.3Mg_0.2 Mn²⁺_0.1Al_0.4)[Si₂O₇][SiO₄]O(OH) [One might be tempted to write a meaningless, non-charge-balanced end-member CaLaAl₃[Si₂O₇] [SiO₄]O(OH)].

Within the framework proposed above, the correct way to proceed is: First assign the mineral to the allanite subgroup ([M³⁺]_A2 > 0.5, [M²⁺]_M3 > 0.5, [F-]_O4 < 0.5). Thus the end-member formula is: CaLaAl₂M²⁺[Si₂O₇][SiO₄]O(OH). Fe²⁺ is dominant among the M²⁺ cations (i.e. Fe²⁺ is the dominant charge-compensating cation). Thus the mineral would properly be named allanite-(La). Likewise, another allanite-subgroup mineral with the same formula except for 0.4 Al replaced by 0.4 Cr must be given the same name, i.e. allanite-(La). The adjectival modifier Cr-rich (chromian) may be added to distinguish this mineral from Cr-poor allanite-(La).

One of the pitfalls in epidote-group mineralogy has been the derivation of a mineral name from chemical data.

There are four major problems. (1) Rules of finding a proper root name depend on the subgroup because of interference between homovalent and coupled heterovalent substitutions (discussed above). Thus, before a mineral name can be assigned, the subgroup has to be determined. (2) Epidote-group minerals are frequently mixed-valence Fe and/or Mn compounds and the charge of the respective cations has to be calculated on the basis of assumed negative charges. (3) Frequent oscillatory zoning may be masked if the zoning is perpendicular to the incident electron beam. (4) Minerals of the allanite and dollaseite subgroups are commonly metamict, in some cases with complete loss of crystallinity (for a review, see Gieré & Sorensen, 2004). Strongly metamict materials significantly deviate in composition and stoichiometry from their non-metamict equivalents. Metamictization is accompanied with hydration, swelling, and selective leaching. The transition from an ideal crystal to amorphous material during metamictization is continuous. The question of crystal quality remains open if electron-microprobe analytical data are applied without accompanying diffraction experiment.

A central question intimately associated with the problem of metamictization concerns the existence of significant vacancies, in particular on the A sites. This issue becomes important when discussing how the mineral formula should be normalized. Ercit (2002) states two arguments in favor of A-site vacancies: (1) the structure refinement of Sokolova et al., (1991) and (2) the electron-microprobe analyses by Peterson & MacFarlane (1993) and Chesner & Ettlinger (1989).

Ad (1). We believe that the structure refinement by Sokolova et al., (1991) on a very Mn-rich sample does not unambiguously demonstrate the presence of vacancies on A sites if the cation distribution is critically scrutinized. Normalization of the corresponding electron-microprobe analysis to Si = 3 leads to an excess of M-type cations. Coordination of the A1 site, for which vacancies are assumed, is approximately six-fold, which is characteristic of Mn²⁺ on A1 (Bonazzi & Menchetti, 2004). Furthermore, the determined scattering power at A1 is 23 electrons, which is intermediate between Ca (20 electrons) and Mn (25 electrons).

Ad (2). Allanites from granitic pegmatites and uraniferous calcite veins in the Grenville Province of the Canadian Shield (Peterson & MacFarlane, 1993) contain up to 4.16 wt. % ThO₂ and are mostly metamict. Thus the analyzed deficit of A-site cations might be associated with partial metamictization. Interestingly, significant A-site vacancies were also calculated from electron-microprobe analytical data (Chesner & Ettlinger, 1989) on young (75 000 y. to 1.2 m.y) volcanic allanites (ThO₂ up to 2.17 wt.%). These allanites appear optically rather fresh and metamictization is certainly not advanced. Thus the A-site substitution vector (REE)³⁺_2/3 + Vac_1/3 ⟷ Ca should not be ignored.

We recommend normalization of electron-microprobe analytical data on the basis of Σ(A+M+T) = 8. This method, however, is inadequate whenever A-site vacancies are present or if not all elements have been analyzed. In either case, the assumption σ cations = 8.0 leads to Si > 3.00 apfu. If Si becomes > 3.05 apfu, the formula may be renormalized on Si = 3. One should be aware that normalization to 3 Si transfers all errors of the Si determination to the other cations in the formula, resulting in larger absolute errors on the number of cations (Ercit, 2002). We discourage normalization on Σ (M+T) = 6 because Mn²⁺ and probably also Fe²⁺ (to a lesser extent) may occupy M1, M3, and A1 (Bonazzi et al., 1996)

Because refinement of the dollaseite-(Ce) and kristovite-(Ce) structures convincingly showed F- to be ordered at the O4 site, the total sum of negative charge should be assumed as follows: Σ (anion charge) = 2(12−x) + x+1, where x = F + Cl (apfu).

Comment: Significant amounts of Cl have been rarely reported, except for Cl up to 0.86 wt.% (0.126 apfu) in the halogen-bearing “allanite” from the Hemlo area, Ontario (Pan & Fleet, 1990). We have no knowledge about the role of Cl in the epidote structure. By analogy with F, we recommend the simplest scheme: Assign Cl to the O4 site.

The Fe²⁺/Fe³⁺ and Mn²⁺/Mn³⁺ values are varied until Σ (cation charge) equals Σ (anion charge), oxidizing first Fe²⁺, then Mn²⁺, in order to account for their different redox potentials. The assignment of all Mg to the M sites also contributes to the balancing of excess positive charge in the REE-bearing members. One must be aware that this step introduces considerable errors if not all cations have been analyzed (one possible indicator is Si > 3.05 apfu). Furthermore, this step assumes that there is no “oxyallanite” component.

• All Si is assigned to the T sites. If Si < 3.00 apfu (or Si⁴⁺ + P⁵⁺ + Be²⁺ + B³⁺ < 3.00 apfu), [^IV]Al is calculated as 3 -Si. However, if Si is significantly²

  2

  A good and complete electron-microprobe analysis of an epidote-group mineral should reproduce the stoichiometric constrains within 1 % relative, i.e. Si = 2.97 to 3.03 pfu. Thus, analyses yielding < 2.97 Si pfu may indicate T-site substitution. However, Si values below 2.9 pfu should be regarded as anomalous and probably erroneous. Si values > 3.03 pfu may indicate either A-site vacancies (partial metamictization?) or incomplete analyses.

  below 3.00, the analysis is probably not reliable.

• If there is sufficient Al after step i), the M2 site is fully occupied by Al. Otherwise, any deficit (1 − Al) is to be compensated by Fe³⁺.

• Any excess (oct − 2) of octahedral cations (oct = Al, Fe³⁺, Mn³⁺, V³⁺, Cr³⁺, Ti⁴⁺, Sn4⁺, Fe²⁺, Mn²⁺, Mg, Cu²⁺,….) is to be assigned to A1. Priority must be given to Mn²⁺. If the amount of Mn²⁺ is not sufficient preference should be given to Fe²⁺ or other available cations with large ionic radii.

• The A1 site is filled with Ca to sum up to 1.0 apfu. (minor Na will also be assigned to A1). Exception: If Σ(REE+ACT+Ba+Sr+Pb²⁺) > 1.0, excess cations may be assigned to A1 (preferentially those with the smaller ionic radii).

• All (REE)³⁺ (together with Th⁴⁺ and U⁴⁺) are assigned to A2. Larger divalent cations such as Sr, Pb²⁺, Ba, and K are also added to this site. Remaining Ca is added to A2.

  Comment: In the absence of knowledge on octahedral REE in allanite and dollaseite-subgroub minerals we ignore this possibility in this simplified procedure. Cressey & Steel (1988), on the basis of L_III edge extended X-ray absorption fine structure (EXAFS), suggested that Lu in synthetic dissakisite-(La) is located at the M3 site and that Gd and Er reside at the A2 and A1 sites, respectively. These results, while interesting, require confirmation.

• If any, F and Cl must be assigned to the O4 site.

• The remaining 2.00 octahedral cations are assigned to M3 and M1. In the lack of structural information, a sequence based on decreasing ionic radii could be written to fill first M3 and then M1. Octahedral ionic radii according to Shannon (1976) are listed in Table 6.

  Table 6.

  Effective octahedral ionic radii in Å (Shannon, 1976).

  

  View largeDownload slide

  View Large

Comment: In particular, we do not know how M²⁺ is ordered between M1 and M3 if O4 is partly occupied by F. Because M1 has two bonds to O4 whereas M3 has only one, we would expect that M1 would have a rather strong selectivity for M²⁺ if O4 = F. The method must be certainly refined if additional information on cation order in minerals of this subgroup is available.

Before a name can be given to a mineral species the subgroup has to be determined (see above). In case of dollaseite-subgroup minerals, assigning an individual name may fail for compositions 0.5 < REE + ACT ≪ 1 and 1 ≫ (F on O4) > 0.5. This shortcoming results from our limited knowledge about order of M²⁺ between M1 and M3 for this subgroup. In the Appendix several examples are presented to illustrate how a “working name” can be derived from an electron-microprobe analysis. The term “working name” is used to emphasize that a strongly simplified scheme for the naming procedure was applied. The “working name” may be used in any mineralogical or petrological investigation as long as the derived mineral name agrees with the recommendations of this IMA CNMMN subcommittee. No quotation marks are needed in case the name is already that of an approved species; otherwise they are mandatory. Of course, the “working name” procedure is not sufficient to define a new species of the epidote-group minerals.

The term “pistacite” is not an accepted mineral name and should therefore not be used as a synonym for epidote. There is a tradition among petrologists to name the hypothetical composition Ca₂Fe³⁺₃[Si₂O₇][SiO₄]O(OH) “pistacite” component in clinozoisite – epidote solid solutions. If the term “pistacite” is used in this sense it should be written in quotation marks accompanied by the word component to indicate its theoretical meaning [“pistacite” component]. We firmly discourage from using “pistacite” component and recommend an alternative, less confusing expression X_Fe = Fe³⁺/(Fe³⁺ + Al) to quantify a solid-solution member.

It has been shown by several investigators (e.g., Dollase, 1973; Bonazzi & Menchetti, 1994) that “allanite” dehydroxylates in air between ca. 600 and 700°C where charge balance is maintained by oxidation of ferrous to ferric iron according to: ^M3Fe²⁺ + ^O10OH- → ^M3Fe³⁺ + ^O10O²⁻ + ¹/₂ H₂. This substitution has been achieved experimentally, but it has not yet been demonstrated to occur in nature. The term “oxyallanite” may be used for the theoretical end-member CaREEAl₂Fe³⁺[Si₂O₇][SiO₄]O₂. If the term “oxyallanite” is used in this sense it should be written in quotation marks accompanied by the word component to indicate its theoretical meaning [“oxyallanite” component]. A recommended more descriptive term is oxidized or dehydroxylated allanite-(REE).

Metamictization is a major problem in assigning a proper species name, in particular for REE-bearing epidote-group minerals. Metamict samples tend to be more reactive than well-crystallized minerals (for a review see Gieré & Sorensen, 2004) and exhibit anion- and cation-exchange properties (e.g., possibly leading to vacancies on A sites). There is no sharp borderline between a completely X-ray amorphous substance (due to metamictization) and a mineral with a well-ordered crystal lattice. Subsequent heat-treatment under inert conditions to reestablish an ordered crystal lattice seems to be questionable if the resulting “mineral” adopts a composition that is characteristic of a partly ion-exchanged poorly crystalline (metastable) substance, but not of the original mineral. There is at least some suspicion that such “mineral” compositions are influenced by the experimenter and are not an unaltered product of nature. These problems are not specific of epidote-group minerals but are much more prominent in other mineral groups with higher concentrations of radioisotopes. For this reason the issue of metamictization is out of the scope of this subcommittee and should be treated in a general way by a different working group. We recommend exercising caution with compositions of “partly” metamict epidote-group minerals in naming new species, even if the “faulty” lattice has been mended by subsequent heat treatment.

Some selected electron-microprobe analytical data cited from the literature are summarized in Table 7. Those data have been selected because they represent either borderline compositions (approximately intermediate between two species), or the data are incomplete and/or erroneous, or they originate from partly metamict samples. The aim of this appendix and the data in Table 7 is to demonstrate how a mineral name (Table 8) can be derived in non-trivial cases, based on the formal ion assignment to the key sites as discussed in Chapter 4.

The selected analytical data by Chessner & Ettlinger (1989) and Treloar & Charnley (1987) had to be normalized to Si = 3 because normalization on Σ (A + M + Z) yielded Si > 3.05. The Chessner & Ettlinger (1989) data suggest either A-site vacancies or incomplete analyses of A-site cations, whereas the Treloar & Charnley (1987) analytical data on dissakisite-(Ce) (originally described as allanite) from Outokumpu (Finland) indicate significant M-site vacancies, which are most probably related to incomplete analyses. In particular, the presence of V in the Outokumpu deposit suggests that V³⁺ could have been incorporated in the epidote-group minerals, but Treloar & Charnley (1987) did not analyze V. Another indication of missing cations in the analytical data of Treloar & Charnley (1987) is the sum of positive charges, which is insufficient to balance 25 negative charges even if all iron is calculated as ferric. A less likely alternative is significant F replacing O on O4 and thus lowering the sum of negative charges. F on O4, however, would not explain the low sum of M cations. There is also a deficit of positive charges (with all Fe as ferric) in some analytical data of “epidote-(Sr)” reported by Ahijado et al., (2005) (e.g., analysis Nr. 8). Furthermore, analysis ACNF-8 (Table 7) gives only 2.847 Si pfu, which is anomalously low and probably is an analytical artifact.

What to do if subgroup and/or mineral name assignment fails?

(1) Poor or incomplete analyses or analyses from metamict materials are not expected to lead to the correct species name. There are several indicators for questionable analytical data such as Si > 3.0 or Si < 2.9, vacancies on M sites, or unbalanced charge (Δ charge). (2a) Border-line case (REE + ACT > 0.5, close to 0.5 apfu): It may happen that, due to small amounts of Na, and/or vacancies on the A sites, and/or minor amounts of tetrahedral Al, the charge balance of the formula requires ^M3[M³⁺] to slightly exceed ^M3[M²⁺] even if ^A2[REE +ACT] > 0.5. In other words, the dominant valence on M3 is not in agreement with the definition of the allanite subgroup (example: Analysis PF6 in Tables 7 and 8).

Solution. Priority must be given to the A2 key site. Because ^A2[REE +ACT] > 0.5 and ^O4F < 0.5 the mineral is assigned to the allanite subgroup and the species name is determined by the dominant M²⁺ on M3.

(2b) Border-line case (REE + ACT < 0.5, close to 0.5 apfu); it may happen that, due to small amounts of Th⁴⁺ on the A2 site, and/or Si slightly exceeding 3.00, and/or minor amounts of octahedral Ti⁴⁺, Sn⁴⁺, and/or significant F on O4, the charge balance of the formula requires ^M3[M²⁺] to slightly exceed ^M3[M³⁺] even if ^A2[REE³⁺] < 0.5 (example: Analysis PF2 in Tables 7 and 8).

Solution. Priority must be given to the A2 key site. Because ^A2[REE +ACT] < 0.5 the mineral is assigned to the clinozoisite subgroup and the species name is determined by the dominant M³⁺ on M3. (3) ^A2[REE +ACT] partly charge-balanced by vacancies: We may suspect a certain degree of metamictization (examples OTT16–1 and MTT8–1 in Tables 7 and 8).

Solution: In analogy to (2a).

We are highly indebted to E.S. Grew (Orono, Maine) for improving the English. Former members of this subcommittee are thanked for their contributions. The revised report benefited from comments and corrections by the members of IMA CNMMN. We highly appreciate the supporting comments by Stefano Merlino on the draft version of this report.