Introduction

1Melian alumen is one of the industrial minerals exploited in antiquity and is broadly categorised as an earth which includes Melian, Cimolian and Samian Earths (fig. 1) as well as sulfur (Photos-Jones et al. in review). These earths are differentiated from Industrial stone like marble, pumice and emery. The Aegean island of Melos was well known in antiquity for the industrial mineral ‘alumen’ as well as sulfur and Melian Earth. Our research has been geared to identifying these industrial minerals in the archaeological record using a geoarchaeological approach.

2This paper summarises our current understanding of the nature of Melian alumen, its origin and the potential for exploitation of deposits of alumen in antiquity.

3Alumen is mentioned by Pliny (Bostock and Riley 1857) and Dioscorides (Gunther 1934). We have adopted the latin term alumen rather than simply using its usual modern translation as ‘alum’ in English. This is because of the complex chemical nature of alum, and hence an inevitable uncertainty in exactly what substance(s) Pliny was referring to when he out-lined the properties and uses of alumen in the first century AD.

4Based on the translation of Bostock and Riley (1857) and discussions in Bailey (1932) and Healy (1999), we understand that Pliny wrote that there were several varieties of alumen which had thirty-eight ‘remedies’ or medical applications in addition to other uses as a mordant etc. « For all those maladies which we have mentioned as being treated with the other kinds of alumen, that imported from Melos, be it understood, is still more efficacious ». (Pliny XXXV 52, Bostock & Riley 1857). Its uses reflect its main characteristic: « The leading property of every kind of alumen is its remarkable astringency » (Pliny XXXV 52, Bostock & Riley 1857). It is therefore clear that in antiquity alumen is important for its medicinal properties rather than as a mordant. This role is reversed in the medieval period onwards.

5Singer (1948) reviewed the history of alum and its development as a chemical to meet mainly the needs of the dyeing industry. Although other sulfates have been used as a mordant, aluminium sulfates were preferred. The industrial production of alum (meaning the hydrated double sulfate, KAl(SO₄)₂.12 H₂O or with NH₄+ or Na+ rather than K+) was perfected during the 17^th to 19th centuries (Millard 1999) in parallel with an understanding of chemistry in general. Tournefort (1718) describes an alum mine (cavern and vaults) as well as warm springs on Melos. Leycester (1852) visited and mapped Melos and described an alum mine and also the sulfur emissions at Kalamos (fig. 2).

6Cottrell (1893) in his survey of minerals of Melos also shows an alum mining area clearly based on the earlier map of Leycester (1852).

7Taking the historical accounts of Melos and the history of the chemical alum into account, we can surmise the following. Melian alumen could be an aluminium sulfate salt, and probably one or both of the following minerals, K-alum (’modern alum) KA1(S0₄)₂.12 H₂O, or alunogen Al₂(SO₄)₃.17 (H₂O). It may have occurred naturally or been derived from rock alum (alunite) KAl₃(SO₄)₂(OH)₆. Our own work has indicated the former. It has been argued that alunite processing was not developed until the middle ages (Singer 1948 ; Cardon 2003).

8The importance of industrial minerals on Melos in antiquity has been raised in the archaeological literature since the late nineteenth century (MacKenzie 1897, Pittinger 1975). The archlogical survey of Melos (Cherry 1982) strengthened the argument by stressing that some of the sites of ‘special purpose’ may have related to mineral processing (Sparkes 1982). However, before we began our research on Melos in 1997 there had been no definitive evidence published of sites related to alumen, or other industrial mineral, production in antiquity. Further, the precise localities of the alum mines of the middle ages appear to have been lost. Our research therefore had to adopt a geoarchaeological approach from the outset which meant prospecting for both the mineral deposits and the archaeological evidence (Photos-Jones et al. 1999 ; Hall et al. 2003b).

Origin and nature of alumen deposits on Melos

9The first stage in prospecting for alumen was to tackle the question: Where could aluminium sulfates occur on Melos and how do they form ?

10Clues as to the nature of alumen deposits come from Pliny (XXXV, 52, Bostock and Riley 1857). « Every kind of alumen is a compound of slime and water, or in other words, it is a liquid product exuding from the earth ; the concretion of it commencing in winter, and being completed by the action of the summer sun... ». Pliny (op. cit.) goes on to say « Of this last kind (Melian) there are also two varieties, the liquid alumen, and the solid....”. While his first statement points to an evaporative process around a mineral spring, attempting to understand what he meant in the second statement has been less fruitful in our search for Melian alumen. It has also preoccupied many commentators who have argued that liquid alumen could only be processed alumen or alumen which has been transported/traded as a slurry in amphorae.

11Our initial searches for aluminium sulfates in SE Melos in the vicinity of Aghia Kyriaki, a site of special purpose (Sparkes 1982), led to the discovery of efflorescences at sulfurous fumaroles in a setting reminescent of Pliny’s despcription of alumen depositssits. But there were additional occurrences such as low temperature efflorescences in caves/mines in white rock and efflorescences at sulfurous fumaroles intersected in caves/mines. On the other hand, alunite (rock alum) was also recognised to be quite widespread as a component – of hydrothermally altered rocks for which we adopted the convenient name ‘white rock’ in SE Melos. Three industrial minerals of antiquity on Melos (Pittinger 1975), Melian Earth (white pigment) probably kaolin+ silica+ alunite (McNulty 2000), native sulfur and alumen, as well as most of the industrial minerals exploited on Melos in modern times (Stamatakis et al. 1996) are clearly related to the same geological/ geothermal processes that transform existing rocks, whether they be metamorphic schists, volcanic lavas or alluvial sediments into a ‘white rock’ (Hall et al. 2003b). Therefore, understanding the alteration of rocks to produce ‘white rock’ is the first stage in understanding the origin of alumen. Alunite forms as a result of acid sulfate alteration of rocks-hot sulfuric acid seeps through the rock, removing soluble elements. Although evident from field observations, the process is best understood using computational aqueous geochemical modelling as demonstrated in figure 4 of Hall et al. (2003a). Recent quarrying has revealed the nature of the process in places where fumaroles have been intersected. White alteration zones can be seen to have peripheral zones of red rock result from the leaching of elements such as K, Na, Ca, Mg and Fe from the rocks and reprecipitation of some iron locally on oxidation and cooling of fluid. The iron in the orginal rock is mainly in the reduced or ferrous state(Fe⁺⁺) while the red stain results from the precipitation of oxidised iron(Fe⁺⁺⁺) as ferric oxyhydroxides (especially fine hematite, Fe₂O₃) and possibly sulfates. The chemistry of this process which is archaeologically significant as a method of producing a potential red hematitic pigment in addition to the white kaolinite/silica pigment, is explained in more detail in figure 6 of Hall et al. (2003a).

12Although it is relatively difficult to produce alum from alunite, a method involving roasting and repeated slaking was used in medieval times (Singer 1948 ; Cardon 2003). A change in ratios of components and removal of Al⁺⁺⁺ presumably as an insoluble aluminium hydroxide is required as shown in the following possible reaction (Hall et al. 2003b):

13No archaeological evidence for alunite processing (structures, fuel, waste) has been found in SE Melos (McNulty 2000). It therefore seems unlikely that alunite was exploited on Melos. However, preliminary consideration of our detailed survey of mining features in SE Melos in September 2003 indicates that this view may need to be revised as we located concentrations in veinlets in altered ‘white rock’ of small alunite-rich pellets. These occurred close to an archaeological feature that could be mining related (Photos-Jones et al. in progress). Alunite is a rather characterless mineral but the weathering of the pellets to a distinctive yellowish brown colour could have provide a means of recognising this material in antiquity.

14The discovery of aluminium sulfates associated with active fumaroles, for example near Aghia Kyriaki (Photos-Jones et al. 1999) has been particularly important in our search for Melian alumen. Soft, fluffy to compact, white botryoidal masses of soluble salts located near sulfurous fumaroles, both underground and subaerially proved to be rich in alunogen, Al₂(SO₄)₃.17 H₂O, (Hall et al. 2003a). Other sulfates are present as ‘impurities’ in the alunogen and include K-alum and artinite (Hall et al. op. cit.). However, it is bulk chemical rather than mineralogical analyses that best reflects the chemical nature of the seemingly pure aluminium sulfate deposits, for example K₂O = 0.04-3.93wt. % and Fe as Fe₂O₃ 0.04-0.36wt. % (Hall et al. op. cit., table 4). Iron is of particular significance because of its deleterious effect on colour hue and vividness when alum is used as a mordant (Cardon 2003).

15We have confirmed that the white seemingly pure, aluminium sulfate deposits can contain sufficient iron (more than about 100 ppm) to darken pomegranate juice, a ‘quality control’ test (Hall and Photos-Jones 2002) mentioned in Pliny (Pliny XXXV 52, Bostock & Riley 1857) and an indirect hint of the importance of alumen in the Roman textile industry.

16XRD, SEM with EDAX, ICP-AES, stable isotope analyses and computer modelling of chemical solutions have been used to characterise and understand the origin and properties of the efflorescent salts. The account given here is based on that detailed in Hall et al. (2003a).

17Sulfurous fumaroles and other manifestations of the high geothermal gradient (hot subsurface) under SE Melos are widespread through the island especially in a zone extending from Adamas to Paleochori (fig. 1). The heat is probably the result of relatively recent magmatism, about 100,000 years ago. Sulfur is precipitated from hot vapours that espcape from small orifices in the soil or from subterranean fractures if the fumarole is intersected by tunnelling underground (fig. 3). The groundwater around sulfurous fumaroles becomes acidic due to the formation of sulfuric acid on oxidation of the native sulfur. This sulfuric acid will promote low temperature acid sulfate alteration of local rocks. The more soluble cations present in silicate minerals that make up the rock are progressively lost from the system. Even relatively insoluble aluminium is dissolved from silicates (Hall et al. 2003a, fig. 4), while silica remains in the form of residual minerals such as cristobalite. The significant outcome is that groundwater becomes enriched in aluminium, an element normally considered to be geochemically immobile. Because of the arid climate and the high local geothermal gradient, aluminium sulfates and some other salts are precipitated as efflorescences in the evaporitic conditions around the fumaroles (fig. 3).

18When the salts are dissolved in water (rainwater, fumarole condensate or springwater) they can be transported a short distance and may recrystallise on evaporation of ponded solutions just as described by Pliny.

19The relatively narrow range of values close to zero of sulfur isotope analyses, indicates that the sulfur in native sulfur (d³⁴S = –0.30 to +6.10 ‰ V-CDT) and sulfates (+5.60 to +6.76 ‰) from fumarole sites is dominantly of igneous origin (Hall et al. 2003a). The oxygen in the alunogen sulfate is enriched in d¹⁸O, a result consistent with the sulfate having originated on sulfur oxidation by atmospheric oxygen which is itself enriched in d¹⁸O (Hall and Fallick in prep). In summary, the mineralisation at sulfurous fumaroles is a secondary process resulting from acid sulfate alteration/weathering in a geothermal setting, and this process would have formed readily workable deposits in antiquity in the post-volcanic but geothermally active landscape of SE Melos.

Potential for exploitation of Melian alumen in antiquity

20Pliny’s term alumen therefore was, or included, aluminium sulfate minerals and in the case of Melos, alumen was mainly alunogen (probably with K-alum) which most likely came in Classical/ Roman times, or earlier, from fumaroles where it could easily be ‘harvested’. Such alumen could be called a renewable resource in modern parlance. It is not difficult to imagine that underground mining into fumaroles would have taken place in pursuit of sulfur, alumen and possibly Melian Earth at depth. While we have not yet confirmed that such mining took place in antiquity we have located several potential early mining sites in SE Melos including undated caverns that appear to have been excavated and shaped for the exploitation of a fumarole. One of these, the ‘Fyriplaka cave’ is illustrated in figures 4-9.

Conclusions

21Our main observations related to the origin and nature of Melian alumen are, in summary:

1. Aluminium sulfates formed and are still forming at sulfurous fumaroles and also as underground efflorescences in the post-volcanic geothermal setting of SE Melos.
2. Oxidation of sulfur in the weathering or groundwater environment results in sulfuric acid which reacts with aluminium silicates and mobilises aluminium sulfate ; this process is enhanced in high heat flow settings on Melos.
3. Aluminium sulfates precipitated as efflorescences at fumarole sites are recycled ; potentially lowering the iron impurity content.
4. Melian alumen was an alunogen-rich (aluminium sulfate) salt which probably contained potassium sulfate (K-alum and/or K-sulfate) and minor Fe, Na, Ca and Mg sulfates. These elements may relate to the properties ascribed to Melian alumen in the classical references.
5. Sulfur and oxygen isotope analyses may help in fingerprinting Melian alumen and distinguishing it from chemically produced alum, but alumen from similar geological settings worldwide is likely to have the same isotopic signature.
6. The diverse occurrences of alumen (aluminium sulfates) on Melos (fig. 1) which would all have been readily exploitable in antiquity are :
   1. As effloresescences around sulfurous fuma-roles and probably as evaporitic precipitates from ponded run-off fluids [e. g. Aghia Kyriaki ; Kalamos, Paleochori]
   2. As efflorescences close to sulfurous fumaroles in near-surface caverns and probably as evaporitic precipitates from ponded hydrothermal fluids [e. g. Fyriplaka Cave ; Paleochori caves ; Kalamos]
   3. As precipitates and efflorescences in caverns in areas of recent or active acid sulfate alteration by geothermal fluids [e. g. the 18th-19th century Zephyria alummine ?]
   4. As efflorescences in caves and tunnels unrelated to active fumaroles and formed from migrating groundwater. These are widespread minor occurrences in SE Melos and are expected in areas with near-surface sulfur deposits [e. g. Fyrlingos ; Theorycheia].

22Our understanding of the origin and occurrence of alumen can guide further surveys for industrial mineral mining sites on Melos and future excavation of processing sites as well as contribute to the clarification of issues relating to the transport/ trade of this substance.

23We sincerely thank the organisers of the colloquium on “L’Alun de Méditerranée”, Naples, June 2003 for an opportunity to present our account of the origin of Melian alumen. This paper benefited greatly from discussions with participants at the meeting. We are also grateful for funding for our Melos research from: the British Academy ; the UK NERC (NER/B/S/2000/00301) ; the Silver and Baryte Ores Mining Co., Athens ; the British School at Athens ; the Carnegie Trust for the Universities of Scotland ; and the Society of Antiquaries of London.