Sunteți pe pagina 1din 5

Archaeotechnology

The Practice and


Characterization
of Historic
Fire Gilding
Techniques
Kilian Anheuser

Fire gilding was the predominant technique for the gilding


of metalwork from 300 B.C. in China and 200 A.D.
in Europe until the invention of electroplating in the
19th century. This article investigates its metallurgical
aspects based on studies of original objects, gilding
replication experiments, and literary evidence.
Authors Note: All compositions are in weight percent. The
artifact appearing on this page is a 6th7th century AngloSaxon cruciform brooch of fire-gilded low-tin bronze. A firegilded copper Chinese garment hook with a turquoise inlay
from c. 200 B.C. is shown on page 60; three Chinese garment
hooks of fire-gilded and silvered copper with turquoise inlay
appear on page 61.

INTRODUCTION
Throughout history, gold plating silver and copper-alloy jewelry, statues,
figurines, coins, and other metal items
served a variety of purposes. In addition
to its decorative effect, which often appeared in combination with other techniques (e.g., inlay or relief decoration),
gold plating improved corrosion resistance and created the deliberate impression of a completely gold object. The
earliest gilded-metal objects, circa the
third millennium B.C., were produced by
covering the artifacts with gold foil and
then mechanically attaching it (e.g., by
folding the foil around the object) by
riveting or, later, hammering the edges
of the gold foil into grooves cut into the
substrate.1 Gold leaf was subsequently
used, which was produced by hammering the gold to a thin gage and then
attaching it with an adhesive. Pliny (c.
70) mentioned the use of egg-white for
this purpose,2 while resin or drying oil
were possible alternatives.

Silver artifacts were often gilded by


diffusion bonding, whereby gold foil was
burnished onto the hot substrate. The
interdiffusion of gold and silver resulted
in a strong metallurgical bond between
the two metals. Diffusion bonding was
rarely used on copper alloys as surface
oxidation prevented efficient bonding
when copper was heated in air.
Fire gilding, also known as amalgam
or mercury gilding, was developed during the 3rd century B.C. in China and the
1st century B.C. in Europe as an alternative diffusion-bonding method that was
also suitable for the gilding of copperalloy objects. By the 2nd and 3rd centuries, it had become the predominant technique for the gilding of metals in Europe
and the Middle East, a position it retained until the invention of electroplating in the mid-19th century.
Fire gilding was carried out by grinding gold leaf in mercury to create a paste
of gold amalgam. This paste was then
applied to a copper-alloy or silver object
and was subsequently heated to produce a firmly bonded yet porous gilding
layer. This layer was burnished to give a
smooth and reflective surface. A possible variation of this technique was the
application of gold foil onto a surface
coated with mercury and the formation

of the gold amalgam in-situ with the


same subsequent heating and burnishing. Fire gilding on silver or copper retained 825% of the mercury in the gold,
a substantial amount that can usually be
identified by straightforward and nondestructive x-ray fluorescence analysis.3,4
This article reports the results of a series
of fire gilding replication experiments in
which practical aspects and metallurgical questions associated with the technique were investigated.
THE PRACTICE OF FIRE
GILDING
When pieces of gold leaf are brought
together with mercury, the two metals
react immediately to form the gray gold amalgam (Au2Hg). Excess mercury
is used, and a paste of fine particles of
Au2Hg in mercury is obtained by grinding gold and mercury in a mortar. The
consistency of the paste can be altered by
adding extra mercury or removing mercury by squeezing the metal through a
linen cloth, leaving a paste with an overall composition of 8090% mercury and
1020% gold. The object is then covered
with a thin layer of gold amalgam paste.
Clean silver can be easily coated with
this paste, but the much lower solid solubility of mercury in copper makes wet-

Over the last 30 years, there has been a discernible increase in the number of scholars who have focused their research on early industrial organizations, a field of study that
has come to be known as Archaeotechnology. Archaeologists have conducted fieldwork geared to the study of ancient technologies in a cultural context and have drawn on
the laboratory analyses developed by materials scientists as one portion of their interpretive program. Papers for this bimonthly department are solicited and reviewed by Robert
M. Ehrenreich of the National Materials Advisory Board of the National Research Council.

58

JOM November 1997

(~ Au3Hg)
(Au2Hg)

Temperature C

1997 November JOM

evaporated. The vapor pressure of merting of a copper or copper-alloy surface


cury at room temperature is too low,
more difficult. Not only is thorough
however, and no diffusion bond would
mechanical cleaning required, but the
be formed between the substrate and the
thin copper-oxide layer that is formed
gold. Thus, heating was an essential step,
on the surface by contact with air has to
and cold amalgam gilding would never
be removed by spreading the amalgam
have been possible.11
paste in combination with a corrosive
agent, such as dilute nitric acid. The use
CHARACTERIZATION STUDIES
of an agent was recommended in a 16th
century fire-gilding procedure by the
X-ray diffraction studies of replicated
Italian goldsmith and sculptor Benfire-gilding samples showed that the
venuto Cellini.5 Before the invention of
color change from gray to yellow indi2 m
mineral acids in the late Middle Ages,
cated the occurrence of a solid-state reFigure 2. A secondary electron image of the
solutions of alum, rock salt, and vinegar
action after the evaporation of all of the
surface of amalgam gilding after heating and
were used for this purpose.6 Modern
metallic mercury, in which the -gold
before burnishing. The twisted amalgam paramalgam (Au2Hg) lost further mercury
procedures recommend treating the copticle deformed in the grind process during
per with a solution of mercuric nitrate in
and transformed into the -gold amalpreparation.
nitric acid prior to the application of
gam (approximately Au3 Hg). A sample
gold amalgam to amalgamate the surate a smooth and brilliant surface (Figthat was heated just beyond the point of
face by electrochemical replacement.7
ure 3). Traditionally, the burnished surcolor change contained 25% mercury
face is then gently etched with pastes
(determined by electron microprobe
Rubbing cinnabar (red mercuric sulfide)
containing chlorides, nitrates, or sulfates
analysis on a cross section), which correover copper or silver produced the same
to improve the appearance (mise en
sponded to the mercury-rich border of
effect, and this may well have been how
couleur),8 although this is not an essenthe stability range of -gold amalgam in
the amalgamation of metals was first
the Au-Hg phase diagram.12 Further heatinvented.
tial part of the process.
The amalgam-paste-covered object is
The metalsmith must be careful to
ing created mixtures of + -phase or
then heated. After a few minutes at
the -phase alone. The -phase
250300C, which is well below the
was not observed. These results
Atomic Percent Mercury
boiling point of mercury (357C),
are in accordance with x-ray dif0 10 20 30 40 50 60 70 80 90 100
the gold amalgam changes color
fraction analyses of original fire
1,100
1,064.43 C
from gray to dull yellow, indicatgilding on a 15th century Chinese
900
ing that the object has been suffioffering stand 13 and Byzantine
L
ciently heated and is ready for burchurch
silver.14
700
nishing. The gilding has a porous
The analysis of original fire
500
419C
granular structure at this stage,
gilding from a variety of regions
20.1
388C
having lost more than two thirds of
and periods always showed a thick300 (Au)
its weight through the evaporation
ness of the plating between 210
122C
100
of mercury (Figure 1). Because of
m, which was much thicker than
39.0C
38.836C
its unevenness and porosity, the
the leaf gilding applied cold with

100
(Hg)
1
surface has a matte appearance.
an adhesive (typical thickness of
Twisted tubular amalgam particles,
0.10.4 m). 3,4 This is explained by
0 10 20 30 40 50 60 70 80 90 100
formed in the grinding process
the
fact that fire gilding is a diffuAu
Hg
Weight Percent Mercury
during preparation of the Au2Hg/
sion-bonding process; the role of
the mercury is to ensure a close
Hg paste, survived on the gilded
contact of substrate and plating. Because
avoid overheating the object because of
surface of the experimental samples preof the diffusion of copper or silver from
coppers tendency to oxidize in air at
pared for this study (Figure 2). This and
the substrate into the gold, plating with
high temperatures. Due to the porosity
the fact that the coating did not run
a thickness of less than 12 m would
of the gold amalgam, a black copperwhen samples were fired in an upright
show visible discoloration. On the other
oxide layer also forms beneath the gold,
position indicated that the gold amalhand, a thickness of more than 10 m not
the thickness of which increases with
gam did not become liquid at any point.
only meant unnecessary expense, but, as
temperature and length of time of the
Thus, the mercury must sublime directly
the experiments showed, often remained
heating. This layer eventually causes the
from Au2Hg.
uneven after burnishing.
plating to flake off. In the case of silver,
After heating, the object is burnished
To determine the influence of variawhich can be brought to a red heat in air
with an agate or steel burnishing tool to
tions in firing temperature on the rewithout oxidizing, the temperature is
compress the porous structure and crelimited by the rate of interdiffusion between the gold plating and the silver
substrate. If silver is overheated, the gold
will discolor or even disappear into the
substrate altogether, which restricts the
maximum firing temperature to approximately 350C. The best results were obtained in the laboratory when samples
were heated just beyond the color change
(i.e., 250350C) for heating times not
exceeding 1015 minutes.
Several authors1,9,10 have argued that
amalgam gilding could also be per20 m
20 m
formed without heating by using merFigure 1. A secondary electron image of the
Figure 3. A secondary electron image of the
cury as an adhesive and leaving the obsurface of an amalgam gilding replication
surface of an amalgam gilding replication
ject in open air until the excess mercury
sample before burnishing.
sample after burnishing.
59

the -phase + liquid field


this peritectic reaction
would be inhibited, and
-phase would develop
under non-equilibrium
conditions. Twilley concluded that the gilding
on the particular object
he investigated was performed at 400500C,
which he thought to be
an abnormally low temperature.
Bunker et al. based
their conclusions on
quantitative electron
microprobe analyses of gilding samples
from 3rd1st century B.C. Chinese bronze
vessels and Ordos bronze plaques. They
measured residual mercury contents of
917% in the gold (i.e., values mostly
within the solid-solution region of the
phase diagram). They also concluded
from the phase diagram that the gold
amalgam layer was initially melted when
the object was heated. They further ar-

sidual mercury content, fire-gilding


samples were heated for ten minutes at
temperatures between 150C and 565C.
Using electron-microprobe analysis of
cross sections, the mercury content starts
out at 8090%. There is a rapid loss of
mercury below 250C through the evaporation of excess mercury, followed by
the decomposition of Au2 Hg and Au3Hg.
The evaporation of mercury from -gold
amalgam proceeds at a much
slower rate at temperatures above
300C and requires more severe
heating. A sample that was heated
for 60 minutes at 605C had a residual mercury content close to the
detection limit of the electron microprobe (approximately 0.05%
mercury), but the copper substrate
became badly oxidized under these
conditions. The duration of the
heating and the thickness of the
gilding also influenced the final
mercury content. The effect was
particularly pronounced at temperatures below 300C. There was
no significant difference between
samples heated for 10, 30, or 60 minutes
at 360C. In practice, however, time and
thickness were less important than temperature because the thickness of highquality fire gilding was usually in the
narrow range of 210 m and heating
times of more than 15 minutes led to a
deterioration of the gilding from the oxidation of the copper substrate.
THEORETICAL
CONSIDERATIONS
Twilley13 and Bunker et al.15 discussed
fire gilding on the basis of the goldmercury phase diagram. In x-ray diffraction experiments, Twilley observed
the presence of only - and -gold amalgam in Chinese fire gilding. He argued
that under equilibrium conditions the
gold amalgam would have melted and
that the -phase would only be formed
above 419C through solidification. If
the temperature was decreased below
419C, -phase would be generated from
-phase in a peritectic reaction. Moreover, he assumed that because of the
great compositional disparity between
the liquidus and solidus boundaries of
60

tion of the gold amalgam retained their


shape after heating, showed that this
was not the case.
Bunker et al. and Twilley described
amalgam gilding as an equilibrium process, which it is not. Mixtures of gold
and mercury may require several months
until equilibrium is reached.16 Equilibrium conditions also imply a closed system that, at 500C, requires a mercury
vapor pressure of more than 800 kPa.
These high vapor pressures constituted
a serious problem when the Au-Hg phase
diagram was determined.16,17 Fire gilding, however, was always executed at
atmospheric pressure, nowhere near
equilibrium conditions. Au2Hg cannot
be melted at atmospheric pressure because it disintegrates into the atmosphere
with the loss of mercury.
In fact, the liquidus temperature in the
phase diagram is irrelevant for the firegilding process. Fire gilding is a solidstate reaction in which mercury is evaporated from a mixture of Au2Hg and mercury. When the metallic mercury has
disappeared, the Au 2Hg loses further mercury while forming first and finally -gold amalgam. It is
characteristic of a nonequilibrium
process that the extent of mercury
evaporation should vary with the
duration of heating and the thickness of the gilding, which was
borne out by the experimental evidence. Lechtman 18 and Oddy19
claimed that the evaporation of
mercury required that the object
be heated beyond the boiling point
of mercury. This reflected an underestimation of the vapor pressure of mercury and the ease with
which mercury could be lost from the
solid Au2Hg phase. The vapor pressure
of mercury is high enough from about
250C to allow the fire-gilding reaction
to take place.

gued that the absence of - and -phases


pointed at a solidification above 419C
in which only the solid solution of mercury would be formed in gold. A slightly
decreased mercury concentration toward
the surface led them to conclude that
VARIATIONS OF FIRE GILDING
firing temperatures were around or even
above 500C.
It is known from literary sources that
This study confirmed the observation
there were two different methods for
by Twilley that amalgam gilding conamalgam gilding an object: with a paste
sisted of - or -gold amalgam or a mixof gold amalgam (as described) or by
ture of both. The residual mercury conapplying gold leaf to a substrate coated
tents in fire gilding measured by Bunker
with mercury. In both cases, the applicaet al. were also in good accordance with
tion of the gold was followed by heating
the results obtained in
100
this study. However,

90

Bunker et al. and Twilley
80

were misled in estimat


70

Hg
60
ing the firing temperaAu

50
ture by the assumption

Ag
40
that gold amalgam
30
would melt in the pro
20

10

cess. Experimental evi

5
dence, such as the fact
20
25
0
5
10
15
that amalgam gilding did
Relative Distance (m)
not run over the surface
Figure 5. An electron-microprobe element concentration profile
in the furnace and that
across fire gilding gold amalgam paste on silver. The sample
twisted amalgam parwas heated for ten minutes at 305C. Note the narrow mercury
ticles from the preparapeak (<20% mercury) between the silver and gold.
wt. %

wt. %

100

Hg
90

80
Au

70

Ag

60

50

40

30

20

10

0
5
15
20
25
10
Relative Distance (m)
Figure 4. An electron-microprobe element concentration profile
across gold foil applied to a silver surface pretreated with
mercury. The sample was heated for ten minutes at 305C.
Note the 10 m thick silver amalgam layer (3040% mercury)
between the silver and gold.

JOM November 1997

20 m
Figure 6. A secondary electron image of an
amalgam gilded surface on Cu-20Pb-10Sn
after 25 minutes of heating at 300C, showing
the gold-lead amalgam crystals in liquid mercury.

and burnishing. Early procedures for


the former variant can be found in the
late 3rd or 4th century Leiden Papyrus X
from Egypt6 and the 912th century collection of procedures known as Mappae
Clavicula.20 The latter was first attested by Pliny in c. 70.2 In the 16th
century, the Italian metallurgist
Biringuccio discussed both methods in his description of the firegilding process.21
In the case of fire gilding on silver, an initial coating of the substrate with mercury can be inferred
from the presence of a 1015 m
thick layer of silver amalgam with
2550% mercury underneath the
gilding. Mercury is absorbed into
the silver as soon as it is applied
because of the very high solid solubility of mercury in silver, with the
solubility limit being approximately 50%
at room temperature. This is shown by
an electron-microprobe element concentration profile of a section of an experimental gilding replication (Figure 4). A
similar silver amalgam layer was also
found on some original material, such as
a sample from a Japanese gilded silver
ear pendant.3 When more than one layer
of gold leaf is applied, the different layers blend into one and become indistinguishable in cross section. If the object
cannot be sampled, the spreading of pure
mercury before the application of gold
leaf can sometimes be recognized in those
places where the entire area originally
coated with mercury was afterward not
entirely covered with gold. In this case,
silver amalgam remains on the surface
adjacent to the gilded areas. An example
is the 4th century hunting plate from the
Sevso treasure, where silver amalgam
with 34% and 38% mercury was found
in two places to gilded areas.22
There is also a mercury enrichment at
the gold/silver interface when a paste of
gold amalgam is applied, but the zone is
only a few micrometers thick and has a
maximum mercury content of less than
25% (Figure 5). This occurs because there
is less mercury available and the resulting amalgam layer is thinner with a lower
mercury content. Also, the application
1997 November JOM

of gold amalgam paste on a silver object


can sometimes be recognized from
splashes of gold amalgam extending to
ungilded areas and from diffuse edges
of gilded areas and ornaments (e.g., the
gilded Sasanian silver plate).23 However,
sharp edges could be achieved by the
use of stopping-out pastes to mask areas
that were meant to remain ungilded.
Various medieval and post-medieval
procedures describe such pastes, which
were made of such disparate materials
as flour dust, sheep-bone ash, gesso and
glue, or powdered ceramics and egg
white.24
Because of the differences in the size
of copper and silver atoms and the resulting much lower solid solubility of
mercury in copper (<1% at room temperature), a copper amalgam layer
equivalent to the silver amalgam between substrate and plating is not formed

during fire gliding on copper or copper


alloys. Hence, it is only possible to distinguish between the two fire-gilding
methods on copper-alloy objects by the
presence of features like straight edges
of gold leaf on the surface or lumps of
gold amalgam that were left unburnished
for some reason.3
FIRE GILDING ON
LEADED BRONZE
Oddy et al. analyzed 44 leaf- and firegilded cast copper-alloy objects from the
Roman period.25 They observed that,
with only three exceptions, the lead content in fire-gilded objects was below 5%
and was usually significantly less when
compared to the 530% lead in the leafgilded bronzes. The tin content (18%)
was in the same range in both cases.
Similar observations were made for Japanese gilded bronzes 26 and for Romanesque fire-gilded and ungilded secular and ecclesiastical metalwork.27 Oddy
et al. explained these findings based on
a quotation from the 12th century metalworking manual of Theophilus (III, chapters 62 and 66).28 Theophilus stated that,
to avoid problems, copper had to be free
of lead to be fire gilded. Oddy et al.
attributed this to the higher solubility of
lead, tin, and zinc in mercury but did not
address the questions of what actually

occurred if fire gilding was attempted


on leaded bronze, and why exactly it
should not be feasible. On the contrary,
Bunker et al.15 and Jett29 recently identified amalgam gilding on a number of
Chinese bronzes with 515% lead. Jett
also reported replication experiments in
which a golden fire-gilding layer was
formed on a copper alloy with 13% tin
and 9% lead.
For this study, fire-gilding experiments on bronze and leaded bronze were
undertaken to investigate what (if any)
practical problems occurred and what
were the underlying metallurgical reasons. It was found that unleaded bronze
(Cu-7Sn) could be gilded without any
difficulty, whereas attempts to fire gild
leaded bronze (Cu-10Pb-10Sn and Cu20Pb-10Sn) were unsuccessful. After the
application of the Au2Hg/Hg paste and
a first heating (which in the case of copper, silver, or tin bronze produced
a yellow surface of Au3Hg ready
for burnishing), the samples were
still covered with a mercury-rich
liquid phase in which crystals of
gold-lead amalgam appeared (Figure 6). Drops of molten lead-amalgam ran over the surface during
the heating and accumulated at
the bottom of the sample. In a second heating at a higher temperature (ten minutes at 400C), the
excess mercury evaporated and a
network of gray crystals of a ternary gold-lead amalgam formed
on the surface (Figure 7). The final
composition of the gold-lead amalgam
crystals was determined by electron-microprobe analysis and was found to vary
between 1045% lead and gold with no
preferred stoichometric composition. A
copper content of 2030% reflected the
heating conditions that allowed extensive diffusion of the base metal into the
plating.
The experiments showed that fire gilding can pose problems on leaded-copper
alloys. Lead has almost zero solid solubility in copper or bronze and segregates
as lead globules in the alloy. The mercury dissolves some of the lead near the

20 m
Figure 7. A secondary electron image of an
amalgam gilded surface on Cu-20Pb-10Sn
after ten minutes of heating at 400C, showing the network of gold-lead amalgam crystals with a few droplets of liquid mercury
(arrow).

61

surface and forms a lead amalgam that is


liquid at 300C and precipitates when
the object is cooled to room temperature.
If lead enters the plating, gray gold-lead
amalgam is generated rather than the
desired yellow -gold amalgam. The ternary gold-mercury-lead phase diagram,
which would allow a prediction of which
phases to expect, has not yet been established. This is all in accordance with
Jetts allegedly successful fire gilding on
leaded bronze. Jett obtained a gray surface first, and it took two further applications of gold amalgam to arrive at the
desired golden color. Because of its thickness, however, it became uneven and
could not be burnished to a smooth, flat
surface.
In the experiments reported here, 10%
lead in the alloy was sufficient to spoil
attempts at fire gilding; however, leaded
bronze was frequently used for firegilded objects in ancient China.15,30 This
could have been accomplished by one of
two methods. Because of its negligible
solid solubility in copper alloys, lead
tends to be rejected ahead of the advancing solid-liquid interface during casting.
The exact pattern of segregation is complex and depends on the way in which
the casting is fed, the cooling rates, the
direction of heat flow, and the orientation of the casting, which determines the
nature of any gravity segregation deriving from the relatively high specific gravity of lead. Thus, it is possible for lead to
be segregated away from the surface of a
casting and toward the center of the
cross section, especially where cooling is
very slow.31 Lead analyses based on a
small sample may, therefore, be very
misleading, but current developments
in the computer simulation of castings
may ultimately lead to the accurate modeling and prediction of lead segregation.32 Lead can also be removed chemically from copper-alloy surfaces. In the
12th century, Theophilus (III, chapter
68)28 advised readers on what to do if,
after casting, the copper-alloy object
turned out to be insufficiently pure for
gilding, and he recommended that a suitable surface could be produced by an
oxidative depletion process that entailed
the heating of the object in air until a
black oxide layer had formed. This layer
was then mechanically removed. The
reason this process worked was because
the base metals (i.e., lead, zinc, and tin)
were oxidized first, leaving a surface
layer of refined copper. Organic acids
could also be used for this purpose.
In the same context, Oddy explained
the composition of the fire-gilded horses
of San Marco (Cu-1Sn-1Pb), with the
necessity of using an alloy suitable for
amalgam gilding.33 The rather pure copper was very difficult to cast because of
its high melting point (>1,000C) and its
tendency to absorb water from the surrounding atmosphere, which would
62

have led to gas porosity in the solidifying metal. From the experiments reported here, it is evident that a low lead
content was certainly desirable, but the
reason why a bronze higher-tin was not
chosen for the horses remains obscure.
Large casts such as these would have
been difficult enough with a lower melting bronze.
CONCLUSIONS
It was previously assumed by most
authors that fire gilding was a hightemperature technique in which an object was heated at least to the boiling
point of mercury or even as high as
500C. This study has shown that the
gilding process was, in fact, carried out

For craftspeople,
there was no need
to measure the
temperature: they
heated an object to
the point of color
change or slightly
beyond.
at a much lower temperature; it was a
solid-state reaction that did not involve
melting the gold amalgam. As a
nonequilibrium process, it also cannot
be described in terms of the phase diagram.
For craftspeople, there was no need to
measure the temperature: they heated
an object to the point of color change or
slightly beyond. They knew that overheating spoiled the gilding, however,
and restricted the gilding temperature
to the 250350C range, which left 8
25% mercury in the gold. Thus, fire gilding should be readily identifiable with xray fluorescence analysis of the surface.
If mercury is only detectable with much
more sensitive methods like atomic absorption spectroscopy or emission spectrometry, it is likely that it was only
present as a contaminant.
Aesthetically, the wide range of possible mercury contents in fire gilding (8
25% mercury) does not have any significant effect on the color of the gold. This
is quite surprising given the fact that the
gilding consists of one of two different
gold-mercury phases or a mixture of
botha solid solution of mercury in gold
with up to 15% mercury and Au3Hg
with 2025% mercury. Much more important to the final color is the degree of
interdiffusion with the substrate, copper or silver, that leads to a red or pale
discoloration if the gilding is too thin or
is overheated. Additional coloring of the
gilding with preparations of corrosive
salts was widely practiced by goldsmiths
and gilders.

ACKNOWLEDGEMENTS
The author thanks Peter Northover for
valuable discussion. The experimental work
was carried out with a grant from the German Ministry for Research and Technology/
German Academic Exchange Service.
References
1. A. Oddy, Metal Plating and Patination, ed. S. La Niece and
P. Craddock (Oxford: Butterworth-Heinemann, 1993), pp.
171181.
2. Pliny, Naturalis Historia, 33, chapter 33 (c. 70).
3. K. Anheuser, D.Phil. thesis, University of Oxford (1996).
4. P. Northover and K. Anheuser, (paper presented at the
Am. Inst. for Conservation Gilded Metals Symposium, St.
Paul, MN, June 1995).
5. B. Cellini, Dell Oreficeria, chapter 26 (1568).
6. Pliny, Naturalis Historia, 33 (c. 70), chapter 6465. The use
of alum for the cleaning of silver is also attested by a
fire-gilding recipe in the late 3rd or 4th century Egyptian
Leiden Papyrus X, recipe 41. R. Halleux, ed., Les A Ichimistes
Grecs, Vol. 1: Leiden Papyrus X, Stockholm Papyrus, Fragments
(Paris: Les Belles Lettres, 1981).
7. O. Untracht, Jewelry Concepts and Technology (New York:
Doubleday, 1985).
8. M. Chapman, Ancient and Historic Metals, ed. D.A. Scott, J.
Podany, and B.B. Considine (Marina del Rey, CA: Getty
Conservation Institute, 1994), pp. 229238.
9. O. Vittori, Rivista di Archeologia, 2 (1978), pp. 7181; Gold
Bulletin 12 (1979), pp. 3539.
10. S. Boucher, Small Bronze Sculpture from the Ancient World,
ed. M. True and J. Podany (Malibu, CA: J. Paul Getty
Museum, 1990), pp. 161178.
11. K. Anheuser, Bulletin of the Metals Museum, 26 (1996), pp.
4751.
12. T.B. Massalski, ed., Binary Alloy Phase Diagrams, 2nd
edition (Materials Park, Ohio: ASM, 1990).
13. J. Twilley, The Ceramics Cultural Heritage, ed. P. Vicenzini
(Faenza: Techna, 1995), pp. 161173.
14. P. Dandridge, Metropolitan Museum, New York, personal communication, June 1995.
15. E.C. Bunker et al., Outils et ateliers dorfevres des temps
anciens, Antiquites Nationales Memoire 2, ed. Christiane Eluere
(St. Germain-en-Laye: Musee des Antiquites Nationales,
1993), pp. 5566.
16. S. Stenbeck, Zeitschrift fur anorganische und allgemeine
Chemie, 214 (1933), pp. 1626.
17. C. Rolfe and W. Hume-Rothery, Journal of the Less-Common
Metals, 13 (1967), pp. 110.
18. H. Lechtman, Science and Archaeology, ed. R.H. Brill
(Cambridge, MA: MIT Press, 1971), pp. 230.
19. A. Oddy, Aspects of Early Metallurgy, British Museum
occasional paper no. 17, ed. A. Oddy (London: British
Museum, 1980), pp. 129134.
20. T. Phillipps, Archaeologia, 32 (1847), pp. 183244. See also
C.S. Smith and J.G. Hawthorne, Transactions of the American
Philosophical Society, New Series, 64 (4) (1974), pp. 3128. (For
fire-gilding see recipes 55 and 219).
21. C.S. Smith and M. Teach Gnudi, eds., The Pirotechnia of
Vannoccio Biringaccio, Book IX, chapter 4 (New York: Dover,
1990). English translation only.
22. M.M. Mango and A. Bennett, Journal of Roman Archaeology Supplementary Series, 12 (1994).
23. A.C. Gunter and P. Jett, Ancient Iranian Metalwork in the
Arthur M. Sackler Gallery and the Freer Gallery of Art (Washington, D.C.: Freer & Sackler Galleries, 1992), p. 36.
24. M. Berthelot and C.-E. Ruelle, eds., Collections des Anciens
A Ichimistes Grecs, 3 volumes, 1888 (Osnabruck: Otto Zeller,
1967). For stopping-out pastes see book VI, recipes 21, 23,
and 29. Further recipes are given in Cellinis Dells Oreficeria,
chapter 33.
25. A. Oddy et al., Small Bronze Sculpture from the Ancient
World, ed. M. True and J. Podany (Malibu, CA: J. Paul Getty
Museum, 1990), pp. 103124.
26. S.F. Moran, Artibus Asiae, 31 (1969), pp. 5565.
27. A. Oddy, S. La Niece, and N. Stratford, Romanesque
Metalwork: Copper Alloys and Their Decoration (London: British Museum, 1986).
28. C.R. Dodwell, ed., Theophilus: de diaersis artibus, The
Various Arts, 1st Edition (London: Thomas Nelson, 1961).
29. P. Jett, Metal Plating and Patination, ed. S. La Niece and P.
Craddock (Oxford: Butterworth-Heinemann, 1993), pp. 193
200.
30. P. Jett and J.G. Douglas, Materials Research Society Symposium Proceedings, vol. 267 (Pittsburgh, PA: MRS, 1992), pp.
205223.
31. M.J. Hughes, J.P. Northover, and B.E.P. Staniaszek,
Oxford Journal of Archaeology, 1 (1982), pp. 359364.
32. M. Ratka and P. Sahm, (paper presented at the 13th
International Bronze Congress, Harvard, 28th May1st June
1996). and M. Ratka and P. Sahm, Der betende Knabe, Original
und Rekonstruktion, ed. G. Zimmer and N. Hacklander (Frankfurt: Peter Lang), pp. 67-80.
33. A. Oddy, MASCA Journal, 2 (1982), pp. 4547.

Killian Anheuser is currently a conservation scientist at


Staatliche Museen Berlin.
For more information, contact K. Anheuser,
Staatliche Museen Berlin, Schlosstr. la, D-14059,
Berlin, Germany, telephone 49-30-320-91-298;
fax +49-30-322-16-14.

JOM November 1997

S-ar putea să vă placă și