1In the long history of metallurgy some information on the use of alum in several ancient metallurgical processes can be found in the few ancient texts which discuss such topics, however not all of them are clear and the interpretation of some passages can be problematic. In this paper some examples of metallurgical processes and treatments, involving the use of alum will be discussed and illustrated, where possible, by observations on analysed objects.

2Some of the ancient texts, such as Pliny’s Naturalis Historia, the workshop recipes of the Leyden and Stockholm papyri and the texts of the Greek alchemists are rather well known. Other, such as the Syriac or Arabic manuscripts, practical treatises on alchemy, have been less well studied, however they contain the translations of much more ancient texts and can be rather important in the interpretation of earlier manuscripts. Of course, we have to be aware, that some of them, for example De investigatione perfectionis, the Liber fomacum or De inventione veritatis, contain later interpolations and additions (Berthelot 1893, I, p. 343). Therefore, such texts are to be taken into account only when the passages of interest can be recognized in older manuscripts. However there are also late interpolations, which can be very useful to establish the continuity in the use of earlier techniques in Late Antiquity or in Medieval times.

Alum in the purification of gold and in depletion gilding

3The use of alum in the refining of gold has been recently thoroughly discussed by Ramage and Craddock (2000), in their study on the materials from the excavation of Croesus’ gold refinery at Sardis. In their historical survey of gold refining methods they quote and discuss many passages in the ancient literature which therefore do not need to be discussed again here. The most famous quotations are the paragraphs written by Pliny (N. H., 33, 84 ; 34,121 ; 35, 183) and the passage in which Strabo (Geogr., 3, 2, 8) describes the method of gold refining employed in Gallia. More descriptions are to be found in the texts of the Alexandrian alchemists, in the already mentioned Leyden and Stockholm papyri, but also in later texts such as the De perfecto magisterio and in the Summa perfectionis (Ramage and Craddock 2000, p. 35-39).

4In the ancient parting process (cementation process), electrum, the natural alloy of silver and gold or gold alloys with copper and silver, were hammered out into thin foils, put into a crucible, together with alum and salt, and heated to a temperature below 800°C. During the heating salt and silver formed silver chloride and the gold was left in the crucible as thin pure gold foils. The silver chloride debris were smelted with lead and the silver was finally recovered by cupellation. This is of course a simplified description of the process, for more details see Ramage and Craddock (2000, p. 200-210).

5Alum was also used on debased gold alloys for removing the base metals and silver from the upper layers of the objects, to enhance the percentage of gold on the surface. The porous upper layer was then hammered and burnished to obtain a compact surface and give the impression that the object was made of better quality gold. Surface enrichment process, or depletion gilding, seems to have been known since very early periods. The gold rings found at the Nahal Qanah Cave, are dated to the 4^th Mill. BC (Shalev 1993) and are the earliest examples of surface enrichment identified up to now. Similar processes were also employed on silver alloys.

Surface enrichment on silver alloys

6Debased silver could be improved by pickling the objects made of copper and silver alloys in a pickling bath with alum, salt and organic acids. The base metal corroded from the surface layers and the object was then burnished and polished to achieve a shiny and silvery surface. Several recipes in the Leyden and Stockholm papyri give detailed instructions for this process (Halleux 1981 ; La Niece 1993, p. 201-210; Giumlia-Mair 1998, p. 247; 2001a, p. 770-771). This method was also widely used in Roman times, particularly in the periods of crisis between the years 63 and 260 AD, on the “silver” denarii, which were in circulation in inflation times and contained often only 12-18% of silver (Cope 1972; Burnett 1991; Reece 1997), however this technique is found also in Medieval times, for example in one of the recipes collected in the Syriac version of Zosimos’ books, which in Berthelot s translation into French suggests to “purify” silver in vinegar, with the addition of burnt salt and alum (Berthelot 1893, II, p. 217).

7The result of this treatment on debased silver can be appreciated by looking at the SEM micrographs of a very diluted silver coin from Palestine, dated to 1190-1191 AD (fig. 1). The superficial layer of the coin contains around 20% of copper, while in the inside around 42% of copper was determined. The micrograph shows a detail of the coin in an area where the upper layer, richer in silver, has partly flaked away, because of the development of corrosion products inside the coin, revealing the copper-rich nucleus underneath. The stratified layers of the burnished metal on the surface are clearly visible. The treatment with acids had apparently produced a porous and spongy surface, which had been compressed and solidified into a silver-rich layer by burnishing and polishing it. Fig. 2 shows a section of the coin seen in the SEM: the bulk is still metallic and contains around 42% of copper, the dark areas under the compact looking surface are extremely fragile, porous and spongy and contain mainly copper corrosion products. The analysis of the thin burnished metallic layer on the surface has determined a copper content of only around 20%.

8The alloys used for the production of objects, which had to be “improved” by this kind of treatment could also be rather complex and made out of different metals with appropriate characteristics.

9The Leyden and Stockholm papyri, written as workshop manuals of Egyptian artisans after the 3^rd c. AD give several recipes of alloys for the imitation of silver, which contain variable amounts of silver, tin and “white copper”, i. e. arsenical copper and other minerals.

10In only one single case up to now, one of the alloys described in the alchemistic recipes has been recognized in the alloy employed for a Roman object (Giumlia-Mair 2001b, p. 15-16), dated to the 1^st c. BC-1^st c. AD, found in the Roman town Emona, todays Ljubljana, the capital of Slovenia. This is a rather large, silvery-coloured Almgren-type fibula (fig. 3), made of two unusual alloys: The pin contains 49. 4% Cu, 2. 63% Sn, 0. 04% Zn, 0. 77% Pb, 0. 03% Ni, 0. 13% Fe, 35. 1% Ag, 0. 8% Sb and 0. 5% As. Cobalt, bismuth and cadmium were not detected. The body of the fibula is made of an alloy with 40. 1% Cu, 1. 79% Sn, 0. 19% Zn, 0. 65% Pb, 0. 03% Ni, 0. 25% Fe, 37. 3% Ag, 1. 2% Sb and 3. 83% As.

11It is quite clear that the artisan who made the fibula was a skilled metalworker and knew at least some of the alchemistic alloys in circulation in Roman times, which up to this point were only known from the mentioned literature. The alloy chosen for the pin is a debased silver to which the low tin content gives excellent properties of malleability: the spring is still elastic today.

12The alloy chosen for the body is to be found in two different recipes, which describe alloys for the imitation of silver, containing variable amounts of silver, tin and “white copper” (Leid. 4 ; 39 ; Holm. 3 ; 4.). As explained in one of the recipes of the manual (Leid. 22) with the title χαλκοῦ λεύκωσις – the Medieval dealbatio aeris, the whitening of copper – white copper is made with 2 drachmas of “rotten sandaraca” and one mina of molten Cyprian copper (pure copper). Also 5 drachmas of lamellar stypteria are mentioned in the recipe. Sandaraca (Pl., Ν. H., 34, 176; 177; 178; 35, 30; 39; 40; 177) is the name of realgar (arsenic sulphide AsS), a red crystalline mineral, which exposed to sunlight “goes rotten”, becoming dark and earthy and is found in association with yellow orpiment (As₂S₃) and with stibnite, a grey antimony mineral. Arsenic and antimony give a silvery appearance to copper, because of the phenomenon of inverse segregation (Meeks 1993, p. 267-261) and this is the reason why sandaraca was added to the alloy of the fibula body. Antimony came apparently into the alloy as stibnite, together with arsenic sulphide. Indeed the recipe specifies that “rotten sandaraca of the colour of iron” –σανδαράχης τῆς σαπράς τῆς σιδηριζούσης – should be used and stibnite has a metallic dark grey colour. It has to be noted that both elements are not present in the alloy of the pin: their presence would have hindered the finishing by hammering, as already very small amounts of both elements greatly increase the hardness of the copper alloy and makes it fragile. Stypteria was of course used to remove the copper from the surface layer and to make the fibula look like real silver. After this treatment the surface looked rather porous and had to be consolidated by burnishing, as described before.

Alum in artificial patination processes

13The existence of artificially patinated copper based alloys, containing small amounts of gold, and silver in antiquity, has been recognized only in the last decade (Giumlia-Mair and Craddock 1993; Craddock and Giumlia-Mair 1993; Giumlia-Mair 1996, 1990; 2000; 2001a; Giumlia-Mair and Riederer 1998; Giumlia-Mair and Lehr 2003). The material is similar to the modern Japanese artificially patinated alloy shakudo, which is made of copper with around 1% of gold and 1% of silver, is patinated in a chemical bath containing copper salts, alum and sometimes vinegar, and becomes blue-black or purple-black after the treatment. These objects are usually also inlaid with gold and silver which contrast in a pleasant way with the dark background of the patina.

14The earliest objects identified up to now, made of these patinated alloys are the famous crocodile statuette of the god Sobek, now in the Ägyptische Sammlung in Munich, and the statuette of pharaoh Amenemhat III (1842-1794 BC) now in a private collection in Geneva, both found at el-Fayum and dated to the 19^th c. BC (Giumlia-Mair 1996), however many more examples, dated to later times, until the fall of the Roman Empire or even somewhat later, were identified in the last years (fig. 4). This kind of alloys were produced by specialized workshops. The artisans jealously kept the secret of the recipes for these precious alloys, which were called hmty-km in ancient Egyptian, kuwano in Mycenaean, kyanos in Greek, and corinthium aes in Latin (Giumlia-Mair 2000; 2001a, p. 772).

15The Syriac version of the book of the alchemist Zosimos, who lived in the 2^nd c. AD and wrote a famous alchemistic treatise, gives several recipes for the production of black patinated alloys (Berthelot 1893, II, p. 223-230) and several “salts” and “minerals”, among them also alum, are mentioned for the preparation of the chemical bath for achieving the black patination (Giumlia-Mair 2002 ; Hunter 2002).

Alum on iron

16In the ancient alchemistic texts or in the surviving workshop manuals, iron is almost never mentioned, except as material for tools or, seldom, as ingredient for some of the recipes. In Roman times it was considered “the impure” or the evil metal, not worth to be discussed. Even Pliny, who attempts a defence of iron at the beginning of the paragraphs dedicated to this metal in book 34 of the Naturalis Historia, gives only a very short description of its ores, reduction process and working operations. It is only in the Renaissance, with the work of Vannoccio Biringuccio from Siena, that we have the first more detailed description of the blacksmith s work.

17Before the publication of the manual “De la pirotechnia” in 1540, there is only one notable exception: a recipe on iron working and treatments found in the text of Zosimos of Panopolis (Berthelot 1888, II, V, V, 1, 10-21 ; 2, 1-7, 347-348; III, 332), in which the production of the very special ferrum Indicum is described. Ferrum Indicum is a very special material, apparently similar to the wootz steel from Sri Lanka or the Persian pulat or the Russian bulat (Craddock 1995, p. 275-283; Gilmour 1996, p. 118-120, Giumlia-Mair and Maddin 2004, p. 130-133).

18In some few instances in the Mappae Clavicula, a work, dated to the 8^th – 9^th c. AD which collects ancient alchemistic recipes and later additions (Smith and Hawthorne, 1974) iron is worth mentioning, but only when it has to be gilded. This is either done by employing resin on heated iron, by mechanically inserting gold foil into a keying on the surface, previously carefully degreased (with an alum solution or other substances) or by cleaning and roughening the part to be gilded in a mixture of copper filings, vinegar, salt and alum. The paste, with the consistency of honey, has a corrosive action and produces a copper coating on the surface by electrolytic replacement. The piece can be gilded on the copper coating, by applying an amalgam of gold and mercury and by heating it to drive off the mercury. The yellowish and porous gold layer has finally to be polished with a burnishing tool. It is just possible that the short mention by Pliny, aceto aut alumina inlitum fit aeri simile (N. H., 34, 149) might refer to this technique.

19In Roman times there are many instances of small iron objects with some gilt decoration (fig. 5), however their number would never be enough to explain the presence of a huge amount of the typical amphorae used as alum containers in Roman blacksmith workshops (see this volume “Alun et artisanat en Gaule romaine”, p. 323-334). Because of the lack of direct information we can only hypothesize the uses for which alum was employed in iron working.

20Iron, like almost all metals, forms an oxide on the surface and, in order to be hammer-weld, as it was the case in antiquity, the oxide must be eliminated or reduced so that there can be metal-to-metal contact. As we have seen with the previously discussed applications of alum in metallurgical processes, this material was apparently well established already in early times, as etching, de-scaling, pickling and degreasing agent.

21In hammer-welding during ancient times, there were different ways of reducing the oxide to permit metal-to-metal contact. One of these ways was for example the raising of the hammering temperature to temperatures in excess of about 700°C, however this could only be done by very skilled smiths, who were able to empirically control their working temperatures. Too high temperatures might have spoiled the carefully cemented and tempered steel and we know from several analyses of ancient steel objects that this happened quite often (Tvlecote 1976, p. 56-57; 1986, p. 171-172; Giumlia-Mair 2004, p. 102-104). Even the sword blades were not always made of properly tempered steel in Roman times and not all blacksmiths were able to carry out all kinds of work on iron alloys. As Biringuccio very clearly describes (Biringuccio 1540, 136v. –137r.), this was still true even in the Renaissance !

22One of the workshop secrets mentioned in his text is the addition of sand or tuff or another "melting earth" on the piece to be worked, to produce some working slag and protect the surface from oxidation. Alum might have been used in a similar way, as degreasing agent and to reduce or eliminate the oxide during the welding process. As Biringuccio says «there are many (smiths) who burn the mass when they intend to boil it, and many who are so afraid of doing this, that they do not bring it to the right point as they should do, because when they work it very hard it scales away, it breaks off and it does not weld together» (transi. A. Giumlia-Mair).

23In Roman times alum was most probably employed with the aim to avoid the nuisances and accidents so vividly described by Biringuccio, i. e. to eliminate the oxide before the hammer-welding operations and as degreasing and pickling agent. Particularly in large workshops with a copious output the treatment of the working pieces with alum before welding would have speeded up the process in a noticeable way.