Working Group 17 
NIJMMARY 
1\ woven silk picture transferred 
tlfl image to the glass against 
which it had been pressed for 
I rlllrteen years. The image was 
I'(rrmed from salt, which occupied 
I he areas where the silk had not 
louched the glass. Salt impreg-
IIl1tion increases the water 
I'lilsorption of silk at a relative 
humidity below the value at which 
pure salt deliquesces. A salt 
;Ioilltion had formed in the fabric 
II I a moderate relative humidity 
IIl1d then migrated to the glass 
where the salt precipitated 
Ilocause the relative humidity 
WIlS below the deliquescence point 
,,1' pure salt. The process has 
IHlen replicated. 
Lighting and Climate Control 19525 
THE SPONTANEOUS TRANSFER TO GLASS OF AN IMAGE OF JOAN OF ARC 
Tim Padfield and David Erhardt 
Conservation Analytical Laboratory 
Smithsonian Institution 
Washington DC 20560 USA 
It is not unusual to find on opening the glazed frame of a picture that an 
image of the original has been imprinted on the inner surface of the glass. 
Here we propose an explanation for one example of this phenomenon. 
gog 
The silk picture shown in Figure 1 was woven by the Jacquard process in 
France around the beginning of this century. It now belongs to the National 
Museum of American History in Washington DC (1). This picture and six 
others from the same source were framed pressed against glass. They hung 
for seven years in a curator's office, after which they were put into a 
store room for a further seven years. When the frames were dismantled each 
glass was seen to have acquired a copy of the original picture. 
The image, shown in Figure 2, results from the scattering of light by fine 
crystals of salt, sodium chloride, which form an adherent film on the inner 
surface of the glass . The salt is mixed with a small amount of an organic 
substance with surfactant properties. The dark areas are clear glass. The 
white areas of salt on the glass match the white areas on the picture almost 
exactly. There is, however, a notable exception : the sleeve of the left arm 
(Figure 3, top). Here a dark mark on the glass (Figure 3, middle) 
corresponds to a glossy white area on the woven image. Figure 3, bottom, is 
a relief image of the cloth which shows that all the dark areas of the 
picture, but only a few of the white areas, including the highlight on the 
sleeve, are raised. The salt image does not match t he pattern of the colour 
but instead follows the relief of the original. The salt is found on the 
glass only where the cloth did not touch it. 
Figure 1: The woven silk picture 
of Joan of Arc. It measures 520 
by 320 mm. 
Figure 2: The image on the glass 
which was pressed against the 
picture. The pale areas are 
formed by light scattered from 
fine salt crystals. 
Figure 3 (left): An enlarged portion of 
Figure 1 (top) with the salt image (middle) 
and the surface relief of the silk (bottom). 
!jIll 
, ',~n" t. ~t.;£. j. 
Figure 6: A portion, 10 mm 
across, of a companion picture to 
Joan of Arc (top), and the 
corresponding salt image. 
I 22"C 
Figure 7: The process of salt 
transfer caused by a temperature 
gradient at high ambient relative 
humidity. 
ICOM( :II M~ill l '" 1fIIIlN" ' IIVA' I" ON,I!)H7 
Figure 4: The crystallization 
pattern of an evaporating salt 
solution with surfactant. 
VOL. III 
Figure 5: The pattern of salt 
formation on glass resulting 
from migration and evaporation 
of solution from saturated silk. 
The image on the glass is consistent with crystallization from a salt 
solution containing a trace of surfactant [2]. Figure 4 shows how crystals 
form over the glass of a beaker containing such a solution. Note how there 
is a gap between the surface of the solution and the lowest crystals on the 
glass. This gap is shown more clearly in Figure 5. Here a piece of silk, 
saturated with salt solution, is placed against a piece of glass. As 
evaporation proceeds crystals form on the glass at a small distance from the 
point of contact of glass with silk. Figure 6 shows a portion of a 
companion picture to Joan of Arc and the corresponding salt image. The 
clear area on the glass around the point of contact with the silk is clearly 
shown. 
Salt is widespread in the picture assembly. It is in the silk cloth, in the 
linen behind the silk and in the card backboard. The salt could have 
transferred to the glass during a period of high relative humidity (RH). 
Above 76% RH pure salt deliquesces: the crystals absorb water vapour from 
the air and dissolve to form a saturated solution. 
A possible sequence of events within the frame of the picture of Joan of Arc 
is illustrated in Figure 7: as humid air diffuses slowly into the picture 
from the back, water absorption by the salt crystals buffers the relative 
humidity so that until all the salt has dissolved the relative humidity of 
the air within the enclosure cannot exceed 76%. Suppose now that a 
temperature gradient forms within the picture because its back is against a 
cool outside wall or because the front is warmed by a spotlight. The 
relative humidity at the back of the picture will be buffered to about 76% 
by the action of the salt, whose deliquescence point is not affected by 
temperature. However, the relative humidity at the warmer inner surface of 
the glass will be slightly below 76% because in an empty, closed space it is 
the water vapour concentration which tends to become uniform, not the 
relative humidity. Under these conditions the salt solution in the fibres 
will move by capillary flow among the fibres and then migrate from the point 
of contact across the glass. The salt solution begins to dry out as it 
traverses the glass, eventually depositing salt at a small distance from the 
point at which it emerged from the fibres. The evaporated water diffuses 
back through the picture and the cycle of deliquescence, capillary movement, 
evaporation and precipitation continues. The first deposit of salt crystals 
on the glass forms a capillary network allowing the solution to flow outward 
from the point of contact protected from evaporation. Fresh salt 
precipitates outside the existing fringe of salt. Over several years even a 
feeble movement of salt driven by intermittent imposition of a high relative 
humidity and a small temperature gradient may indeed have generated the 
image found on the glass. 
-
----~-
--------- -
---------~ -
[I Working Group 17 
Figure 8: The experimental silk 
cloth with the image formed on 
the glass (10 mm across) . 
I~igure 9: The faint salt image 
on glass formed over two weeks 
tn a constant environment of 65% 
RH and 22°C. The image is 3 mm 
llcross. The bright discs are 
Itnfocussed dust particles . 
Lighting and Climate Control 
Figure 8 shows the result of a laboratory simulation of the image transfer 
process proposed in Figure 7. A piece of silk cloth was made uneven with 
loops of silk thread. This cloth was put in a glazed frame with linen 
backcloth and card to imitate the construction of the original, except that 
an impermeable back was added. The silk alone was impregnated with salt. 
All the components were pre-conditioned to 76% RH by suspending them over 
saturated sodium chloride solution. After this the pieces were assembled 
gil 
and set on the laboratory bench. The glass was overlaid by a sheet of black 
paper which was warmed by a lamp. The back of the picture was maintained 
at the 22° C ambient temperature by fanning air over it. The slight increase 
in temperature of the silk would, if no salt were present, cause a small 
increase in the relative humidity of the air in the enclosure. In this 
case, however, the salt buffered the relative humidity by absorbing the 
water vapour released by the silk. After fourteen days the piece was 
disassembled and photographed. Figure 8 (bottom) shows the image on the 
glass. The similarity to the genuine effect is close, though the process 
has been arrested at an early stage in the spreading of the fine-grained 
salt layer out from the clear portion close to each point of contact. 
This salt image was formed quite rapidly at about 76% RH, the minimum value 
at which pure salt deliquesces. However, a relative humidity as high as 76% 
should never have occurred, at least not for long, in an air conditioned 
museum. We therefore set up an experiment to see if image transfer could 
occur at a lower relative humidity and at a uniform temperature . 
In this second experiment the test picture was conditioned to 65% RH and 
then assembled as before. After two weeks of incubation at 22° C salt 
transfer was again observed. The effect was weaker; indeed the salt was 
scarcely visible. The image shown in Figure 9 was photographed in a humidity 
chamber so that the fine salt deposit deliquesced to give liquid droplets 
which scattered light better . 
The silk was examined to determine the distribution of the salt, originally 
homogeneously distributed within the fabric. We found that it was now 
concentrated at the high points of the weave, where the cloth touched the 
glass. It is not obvious why this should be so. However, the authentic 
images, such as Figure 6, show a small pip of salt at the point of contact, 
surrounded by the clear halo. Whatever the process at work, there is no 
inconsistency between the original and the reproduction. 
How can salt move, apparently in solution in water, within cloth that is dry 
to the touch, at a relative humidity well below the deliquescence point? To 
gain more evidence we measured the water absorption of salt-impregnated silk 
as a function of relative humidity. The results were surprising. It seems 
that salt-impregnated silk does not behave like a simple mixture of salt and 
silk. The diagrams in Figure 10 show what happens. In the case of clean 
silk there is a continuous absorption of water as the relative humidity 
increases. The shape of the graph for pure salt is completely different: 
there is negligible absorption of water as the relative humidity rises until 
the critical value of 76% RH is reached. At this point the salt absorbs 
water from the air and dissolves in it to form a saturated solution. As the 
relative humidity rises further the salt solution continues to absorb water, 
theoretically reaching infinite dilution at 100% RH. One might expect to 
describe the behaviour of a mixture of salt and silk by adding together 
these two curves to produce the composite curve in box C. In fact, however, 
the step in the curve is smoothed out so that the contaminated silk absorbs 
more water than expected at a low relative humidity, as shown in box D. 
Above 76% RH the water absorbing capacity of the mixture is less than 
expected. 
A 
relative humidify 
Figure 10: Comparison of the water absorption of silk and of salt : 
separate and mixed. 
g12 
Table 1: Water content of silk 
(percent of dry weight) at 25°C. 
silk plus 
RH clean silk 46% salt 
9 2.1 2.3 
26 4.7 5.1 
35 6.0 6.7 
45 7.5 8.5 
56 9.2 10.9 
67 11.4 15.2 
70 12.8 20.0 
72 13.5 30.7 
73 13.9 67.3 
67 13.5 21.0 
56 11.7 14.1 
45 9.9 10.8 
35 7.8 8.2 
26 6.5 6.6 
9 3. 3 3.3 
silk plus 
RH clean silk 3% salt 
15 4.4 4.3 
20 5.2 5.1 
26 6.0 5.9 
32 6.8 6.7 
38 7.8 7.6 
47 8.5 8.6 
54 9.4 9 . 5 
58 9.7 9.9 
64 10.4 11.1 
71 11.6 13.7 
78 13.5 17.8 
83 14.5 20.4 
93 20.0 33 . 5 
84 17.6 23.0 
78 15.8 19.6 
75 15.3 18.4 
72 14.7 17.3 
68 14.1 16 . 2 
65 13.8 15.5 
63 13.1 14.3 
58 12.5 13.2 
53 11.9 12.2 
40 9.9 9.8 
29 8.4 8.2 
20 6.6 6.5 
Table 2: Electrical resistance 
of silk containing 2.4% of salt. 
RH Mohm 
37 )7 
47 )7 
53 7 
57 6.8 
61 5.4 
63 4.5 
66 3.0 
73 1.6 
74 1.0 
75 0.72 
76 0.62 
36 )7 
60 6.5 
68 3.6 
73 1.6 
77 0.65 
79 0.52 
87 0.12 
80 0.36 
76 0.74 
73 0.64 
72 0.77 
RH Mohm 
37 )7 
60 5.4 
66 3.3 
71 2.3 
75 1.1 
80 0.46 
85 0.17 
89 0.084 
90 0.072 
84 0.21 
77 0.76 
74 0.74 
69 1. 25 
68 1.4 
63 2.6 
60 4.2 
59 4.8 
ICOM COMMITTEE FOR CONSERVATION, 1987 
25 
5 
Figure 11: Water absorption isotherms at 25°C of pure silk and of salt 
weighted silk. The isotherm of pure salt is expressed as grams of 
water per 3 grams of salt to match the data for 3% salt loaded silk. 
Some of our experimental results are shown in Figure 11 (all the measured 
values are given in Table 1). With very heavy salt contamination the extra 
water uptake becomes measurable at 25% RH. A three percent loading of salt 
causes separation of the two curves at about 55% RH. 
This extra water is not necessarily mobile. To confirm the existence of an 
ionic solution we measured the relative humidity dependence of the 
electrical resistance of silk containing 2.4% salt (Table 2). The results 
are plotted as conductivity in Figure 12. The electrical conductivity is 
enhanced in a manner very similar to the water absorption. There is no 
sudden change of conductivity at 76% RH. Electrical conductivity of the 
contaminated silk becomes measurable at a relative humidity well below the 
deliquescence point of salt while that of clean silk remains very small. 
It is quite clear from these measurements that salt and silk interact when 
mixed to form a system with a response to atmospheric moisture unlike that 
expected from adding the separate contributions of the components. 
c:. 
...:;:: 
E; 
~ 
.? 
:;;: 
~ ~ 
s:: 
1..3 
---. 
B ~ 
'", 
~ L.,J 
10 
clean silk 
QJ ~----------~------__ ~~ ____ ~ ____ ~ ________ ~ ____ -J 
50 (JO TV 80 
Humidify) % po 
Figure t 2: Tll{~ Clll~{jI: dcl\.l I ondllcllvlt.y of a silk strip impregnated 
with 2.4% HI1IL . IIItIHlfJlIl ' cI I\L IW"G. The strip was 16 mm wide by 85 mm 
long ~lnd wolp;lttld :1:10 S1.'iIlrIlII pI1/' Hquare metre. 
III 
t 
Working Group 17 
I, 
, 
t, 
I' I 
\ 
I \ ( 
I \ \1 
I: :,\ ~ 
( I I 
I , ), 
", 
, 
I 
"'I"jure 13: Diagram of the salt 
11'lllIsfer process at constant and 
"n Iform temperature. 
Lighting and Climate Control 
These results suggest a mechanism for salt transfer that needs no external 
driving force such as a temperature gradient. This is summarised in 
Figure 13. The salt-contaminated silk absorbs more water from the 
atmosphere than pure silk. An electrolyte solution is formed within and 
among the fibres which is sufficiently mobile that it permeates by 
capillarity, or by surface diffusion, throughout the entire system. At the 
point of contact between fibre and glass the solution creeps over the glass. 
Once free from the fibre the normal deliquescence point of pure salt 
reasserts itself and the low relative humidity (that is below 76%) causes 
the solution to evaporate. The salt remains on the glass, the water vapour 
diffuses to the fibre and the cycle repeats itself. The end result is a 
transfer of salt from fibre to glass. 
The ability of salt to move in solution within a protective capillary 
network at quite a low relative humidity is demonstrated by the crystals 
forming in an ambient 50% RH outside the rim of the beaker shown in 
Figure 4. A salt image, though slow to form at a low relative humidity, is 
more likely to endure under these mild conditions because the salt separated 
out on the glass has no tendency to revert to a liquid form unless the local 
relative humidity exceeds 76%. This second, isothermal mechanism, though 
slow, is the more plausible explanation for the image of Joan of Arc. 
The two mechanisms for salt transfer both require that the glass surface be 
below 76% RH. They differ in the way a salt solution develops in the 
fibres. The first mechanism relies on a temperature gradient to produce a 
relative humidity of 76% in the fibres, a condition which causes even pure 
salt to deliquesce. The second, isothermal, mechanism depends on the 
formation of a solution in salted silk below 76% RH. This development of an 
ionic solution at a low relative humidity is not unique to silk. Cotton 
contaminated with salt also shows enhanced water absorption. This property 
may turn out to be a general attribute of hygroscopic solids contaminated 
with inorganic salts. 
The presence of an ionizing liquid has long been known to accelerate 
reactions on the surface of materials. For example, salt solutions enhance 
corrosion of metal surfaces [3]. We have found that contamination with a 
salt solution accelerates corrosion of marble by acetic acid vapour [4]. 
These reactions occur upon the surfaces of materials that are basically 
impermeable even though they may be porous on a gross scale. Here the 
normal deliquescence point of a salt applies. The same acceleration of 
degradation reactions may occur in hygroscopic organic materials, but at a 
relative humidity much lower than the deliquescence point of the pure salt. 
Contaminants which would normally be considered dry and inert at normal 
museum humidity may really be mobile and reactive, without showing symptoms 
such as sweating and darkening which occur when salts deliquesce in 
non - hygroscopic materials. 
Deliquescent inorganic salts are widespread in museums. They may be 
original components of objects, contaminants acquired from use, reaction 
products of gaseous air pollutants, deliberate additions during treatment or 
the unconsidered by-product of preservative treatment. Their presence must 
have an important effect on the permanence of materials. 
Acknowledgements 
Walter Hopwood, Martha Goodway, Joan Mishara and Harold Westley helped with 
the analytical work on the silk pictures. We are grateful for help from 
Rita Adrosko and Kathy Dirks of the National Museum of American History. 
NOTES & REFERENCES 
1. The silk picture of Joan of Arc was designed by G. Doyen and woven by 
Neyret Freres of St. Etienne in the early 20th century. There are seven 
pictures in the set owned by the Department of Social and Cultural 
History of the National Museum of American History. 
2. We used a saturated salt solution with 1% of sodium di-octyl sulfosuccinate 
for all the experimental work. A full account of the analytical and experimental 
work together with a more detailed commentary on the results is available as 
Conservation Analytical Laboratory report no 3349 (1987). 
3. U. R. Evans, The Corrosion and Oxidation of Metals: Scientific Principles 
and Practical Applications (London: Edward Arnold Ltd., 1960) pp. 486-501. 
4. Work in progress in our lab. Acetic acid vapour in a stream of air at 
70% RH reacts much faster with a piece of marble which is contaminated with a 
small amount of sodium bromide solution than with pure marble. Sodium 
bromide deliquesces at 54% RH. 
· tCOM COMMITTEE FOR CONSERVATION 
8th Triennial Meeting 
Sydney, Australia 
6-n September, 1987 
Preprints 
THE GEIIY 
CONSERVATION 
INSTITUTE 
Los Angeles 
1987