ANOXIA - TREATMENT BY OXYGEN DEPRIVATION :
OPTIMIZING TREATMENT TIME OF MUSEUM OBJECTS
Michèle Gunn1*, Houri Ziaeepour2, Fabrice Merizzi3, Christiane Naffah4
1 Musee Quai Branly, 55 quai Branly, 75007 Paris, France.
2 Mullard Space Science Laboratory, University College London, Holmbury, St.
Mary, Dorking, RH5 6NT, Surrey, UK.
3Department of Islamic art, Louvre Museum, 34 quai Louvre 75001 Paris,
4 Centre de Recherche et de Restauration des Musées de France
14 Quai François Mitterrand 75001 Paris, France.
ANOXIA, treatment by oxygen deprivation is largely used for decontamination and
disinfestation of cellulose and protein-based organic materials. More specifically this method
is applied to more than one hundred thousand of objects destinated for a new museum in
Paris, "Musee du Quai Branly". We describe the anoxia installation in this museum and
report the result of a study regarding the efficiency of this method and the optimum treatment
time, crucial for treating a large collection. We show that the standard 21 days of exposure is
not always the optimal choice. Temperature plays a crucial role for hastening the death of
insects found within objects. At a temperature of 25°C, it is entirely possible to reduce
exposure times to 10 or 15 days for the insect species commonly found in museums. The
oxygen drop times is between 1 and 2 days for most objects, depending on type and porosity
of materials. This corresponds to a treatment time between 15 and 16 days. The effect of
humidity is less clear. It can increase the necessary treatment time both for larvae and for
Correspondence should be addressed to M.G. (firstname.lastname@example.org)
A large part of the collection of the future Musée du quai Branly - currently under
construction near the Eiffel Tower in Paris - is composed of cellulose and protein-based
organic materials. Such materials are favourable media for the development of micro-
organisms and insects, leading to their degradation.
This collection is presently being treated in a series of steps which include cleaning,
the taking of photographs, packaging and biological decontamination in the Le Berlier
building which has been especially equipped for this purpose.
The collection of the Musée du quai Branly, numbering about 275000 objects1,
comprises on the one hand collections from the Musée National des Arts d’Afrique et
d’Océanie (MNAAO) and the Musée de l’Homme (MH), and on the other hand has been
enriched by new acquisitions.
Studies of the general state of conservation of these collections in their original
institutions by experts, demonstrated the existence of infestation by Anobiidae, Dermestidae
and Tineidae, to name just a few. Infestation was found to be more or less serious depending
on the institution and departments in question. Given that it is difficult to reconstruct an
accurate case history of the infestation and the steps that have been taken to counter it, it was
decided to proceed with treatment of all objects containing organic materials, without
This prudent choice was made in view of the fact that the treated objects were not
destined to return to the site from which they came, but were going to be housed in a new
museum: a « complete overhaul » of the objets in order to reduce the level infestation to zero
is advisable under such circumstances. Furthermore, the objects are treated by oxygen
deprivation (anoxia), which minimises the risk of chemical degradation, althought some
discolorations of some pigments have been reported (TOSHICO, K., 1980); this cannot be said of
classical fumigation treatments even though the treatment times are much shorter in the latter
Heritage institutions currently employ oxygen deprivation treatment times (Tt) of 21 days.
This duration appears to have been adopted in the light of the results of experiments carried
out on a particularly resistant species, the rice weevil, an important pest in the food
industry : 500 hours (21 days) at 26°C, 12% relative humidity, in a nitrogen atmosphere
containing 1% oxygen. The exposure time is extended to 1000 hours (6 weeks) if the
temperature is lowered to 20°C (SELWITZ, C. et al, 1998).
Of more relevance in the museum field, the old house borer, Hylotrupes bajulus, is also a
species resistant to treatment by oxygen deprivation. Its favourite medium is resinous wood.
Eradication of this insect necessitated 20 days in somewhat different conditions: 20°C and
40% relative humidity. The duration can be reduced to 10 days if the temperature is raised to
30°C (VALENTIN, N., 1993).
It has gradually become standard practice to use a treatment time of 21 days. The
recommended conditions are in general as follows: less than 0.1% oxygen, a temperature
above 20°C and relative humidity of 50%.
The exact number will be established at the end of the collection treatment programme
There are a very large number of objects to be treated (more than 80% of the collection).
The deadline for completion of the collection treatment programme leads to constraints, in
view of which time is of the essence. It is therefore appropriate to analyse the time given to
each stage of the object treatment process in order that none be wasted.
If a reduction in the duration of oxygen deprivation treatment turns out to be possible, this
would enable a good speed to be maintained during the progress of the collection treatment
Consequently the key conclusion awaited from this study is the answer to the following
question: is the anoxia treatment efficient for an exposure time less than 21 days ?
Each anoxic treatment installation has its own characteristics. Thus, since the installation
we have used, named EPMQB (named according the name of the musée, Etablissement
Public Musée du quai Branly2), was specially designed for the Musée du quai Branly
treatment site, and was of a new and as yet untried form in the field of heritage and
conservation, it was necessary to carry out a study in order to optimize the conditions of
treatment for objects, in particular as regards oxygen drop times and exposure times.
The feasibility of treating infested museum objects by oxygen deprivation, either through
the use of oxygen scavengers, or in a controlled atmosphere of an inert gas such as nitrogen
(N2) or argon (Ar)) or carbon dioxide (CO2) is now well established. Resistant species such as
H. bajulus or A. punctatum (cellulose) can be totally eradicated, and this is also possible in
the case where insects are at the egg or larval stage which renders them more resistant to
treatment (Rust, M. et al, 1996 ; Selwitz, C. ; Maekawa, S., 1998). Many studies have already
been carried out by teams in the USA (Getty Conservation Institute) and Australia
(Australian Museum) for exemple. These studies have enabled the evaluation of influence of
different parameters, such as the level of oxygen (O2), temperature and relative humidity, on
the exposure times needed to achieve 100% mortality whatever the life cycle stage of the
Therefore, the goal of the study is to determine the efficiency of the EPMQB equipment
and the effectiveness of the traitement in the case of insects buried deep within an object.
This phase of the study should enable the degree to which oxygen is removed from the inside
of treated objects to be evaluated.
We report and discuss results obtained in the following areas :
1) the exposure time, Te, in the new EPMQB installation, leading to 100% mortality
irrespective of life cycle stage of the insects present in infested objects. The treatment
conditions are based on previous literature reports. They must be optimized from a mortality
viewpoint whilst avoiding endangering at the same time the physical structure of the treated
objects: an atmosphere with highly reduced oxygen content is used, between 1000vpm and
30vpm, a temperature of 25°C± 1°C and hygrometry of 50%± 5%.
2) the oxygen drop time, Ti33, defined as being the time taken to lower the oxygen
content in the treatment unit to the required level (0.1 %), and to study the effect of the
degree of loading with museum objects.
3) the oxygen desorption time of the objects Td. The Td value depends intrinsically on
the nature of the materials and the volume of the objects to be treated and the volume of the
Public Institution Musée du quai Branly
Ti, i for inert
anoxia chamber. The Td varies as a function of the permeability of the materials to gases, i.e.
nitrogen and oxygen in this case.
A - PROGRESS IN OXYGEN DEPRIVATION TREATMENT: MAIN RESULTS
OBTAINED BY OTHER INSTITUTIONS
I-TESTS ON INSECTS: EXPOSURE TIMES AND CLIMATIC CONDITIONS OF
The experiments carried out cover a very wide range of insects at all life cycle stages, i.e.
eggs, larvae, nymphs and adult insects. Atmospheres were modified through the use of the
three most frequently used gases: carbon dioxide (CO2), nitrogen (N2) and argon (Ar). The
anoxia treatment was performed in bubble chambers (VALENTIN, N.,1994; RUST, K. et al., 1996).
In some cases in order to simulate their being buried, insects were prepared in glass tubes
closed by a system allowing gaseous exchange. The tubes were subsequently left fixed within
blocks of wood. Other experiments were with sections of pine wood and with books of
dimensions which were artificially infested with the insects to be studied. The main results
show that :
- the most resistant life cycle stages of insects are eggs and larvae;
- not all the insects react in the same way, the old house borer is the most resistant;
- the treatment is more effective with argon than with nitrogen;
- temperature is an important factor whatever the other conditions.
The exposure time is reduced when the temperature increases; for example in the case of
H. bajulus (old house borer), when the temperature increases from 20°C to 40°C, the
exposure time is reduced from 21 days to 2 days. The studies carried out in the museum
environment show exposure times much lower than the standard of 21 days when the
climatic conditions are chosen appropriately, including for the most resistant species; for
- for the Anobiidae
or 5 days with 50% relative humidity, 30°C et 0.03% oxygen. One exception to this
range has been reported; this is Lasioderma serricorne (cigarette beetle), which
required 8 days at 25°C with 50 % relative humidity, or 9 days at 20°C and 40 %
relative humidity. (VALENTIN, N., 1993);
(e.g. : furniture beetle), complete elimination was achieved after 3
- for Hylotrupes bajulus (old house borer) of the Cerambycidae family, which has
shown itself to be rather more resistant, the exposure time was 10 days at 30°C with
40 % relative humidity or 20 days at 20 °C. (VALENTIN, N., 1993);
-for the Tineidae (e.g. clothes moths), 4 days at 25.5°C, 55% relative humidity and an
oxygen level below 0.1 % have been reported (RUST et al, 1996).
More recently in 2000, a Japanese team proposed a practical protocol for anoxia treatment
according to the type of insect. They advise 25°C or 30°C for one to three weeks. When the
temperature is 20°C exposure time should be extended to 10 weeks with an oxygen level of
0.2%. In evidence, in this case the high level of the oxygen makes the treatment time longer
than in the previous experiments (KIGAWA, R. et al, 2000).
The experimental parameters for treatment, according to the infestation, are actually
well known. The key questions which now remain to be answered are whether or not the
experimental conditions are really achieved within the treated objects.
II-DEMONSTRATION OF OXYGEN DESORPTION OF TREATED OBJECTS
The results outlined above were obtained by simulating the burial of insects within
objects in order to mimic as closely as possible a real situation. However, it is difficult from a
technical point of view to directly evaluate the desorption time Td of oxygen from objects,
since to do this would in principle necessitate the positioning of detectors within the objects.
Td depends on the porosity of the objects studied, and on the permeability of the materials to
the gas used. The duration of treatment therefore depends on the desorption time. Various
methods were used to evaluate desorption such as calculation of the time needed for the
oxygen level to reach its equilibrium value, and comparison of the treatment times of infested
objects with the treatment times of reference samples.
II-1 calculation of the time needed for equilibrium to be reached.
Simulation was carried out with fresh, non-infested wood from different sources: poplar,
oak, walnut, having fixed dimensions. The wood samples were bare or covered by a thick
protective coat. They were enclosed in a pocket of volume 32 litres (0.032m3) until
equilibrium was reached. The initial experimental conditions were:
23°C, 0.1% to 0.2% oxygen. Equilibrium was reached with 0.4% oxygen. Oxygen desorption
was found to be more difficult with painted wood than bare wood. The difficulty increase in
the way poplar, oak, walnut, with a maximum desorption time of 120 hours (5 days).
The same experimentrepeated for infested walnut wood showed a shorter desorption time:
equilibrium was reached more quickly, in one hour for the bare wood and in 4 hours for the
painted wood. This result is a consequence of the greater porosity of infested wood in
comparison to healthy wood on account of the tunnels hollowed out by the infesting insects
(SELWITZ, C., 1998).
II-2 Comparison of treatment time of infested objects with treatment time of reference
samples (test samples)
Treatments carried out on museum objects using readily available test samples as
references and performed in a controlled argon atmosphere showed similar exposure times
for the objects and the reference samples, the differences in exposure times being one day or
This result was not always observed. Wooden objects (pianos, sculpture and panels of
wood) infested by Anobium punctatum (furniture beetle) had to be treated over 10 to 14 days,
compared to only 4 for the reference samples. Similarly, 7 days were needed for a textile
sample of dimensions 135x87x43cm infested by Attagenus megatoma (black carpet beetle) as
compared to 2 days for the corresponding reference sample. The treatment conditions were as
follows : a temperature above 20°C, relative humidity of 40 % to 50 % and an oxygen level
of 0.02-0.04% (VALENTINN, N., 1993).
Other experiments performed on powderpost beetle and termites buried in the heart of
the wood and sealed and in other cases non-sealed (more accessible) did not show any
difference in exposure times (RUST, KENNEDY, 1993).
III- DETERMINATION OF TREATMENT TIMES FOR WOODEN OBJECTS
The results below (table 1) were obtained for large objects infested by A. punctatum
(furniture beetle) and H. bajulus (old house borer).
Table 1 : Time and treatment conditions for museum objects using Argon.
According to Nieves Valentin, 1993
punctatum 25 400.03 14
175x64x35H. bajulus20 50 0.0415
punctatum 25 45 0.04 10
75x45x15H. bajulus20 450.0310
This table shows exposure times Te between 10 days and 15 days.
B -THE STUDY CARRIED OUT AT THE MUSEE DU QUAI BRANLY
The bibliography cited above shows that the most resistant insect developmental
stages are eggs and larvae; adults and nymphs are the first to be affected by oxygen
deprivation. Hylotrupes bajulus (common name: old house borer), has turned out to be the
species most resistant to oxygen deprivation treatment. This insect is however not very
commonly encountered in museum objects. It is generally found instead in resinous wood
structures of buildings. It was nonetheless chosen for the present study on account of its
resistance. This insect was also readily available, being bred prior to this study at the Centre
Technique du Bois et de l’Ameublement, CTBA (Technical Center for Wood and
At the same time as carrying out experiments on the insects, other experiments were
performed in order to evaluate oxygen desorption times from the materials used, the degree to
which the oxygen content within the materials was lowered and the effect of the type of
objects loaded on the oxygen drop time.
I-1 The installation used4 : EPMQB ANOXIA SYSTEM
I-1-1 Description of the installation
The anoxia system equipment has five parts
1) A Pressure Swing Adsorption (PSA) nitrogen production unit (TechnicAir) composed
of a an air compressor, b an air dryer and submicronic filter system, c a compressed
air reservoir (1m3), d a nitrogen/oxygen separation subunit made of two receptacles
containing activated charcoal molecular sieves (CMS), e a vessel containing distilled
water for humidifying the nitrogen, f an oil-water separator to avoid discharging
insoluble compounds into the general waste water sewage system.
2) A nitrogen storage unit with four nitrogen reservoirs each having a capacity of 3m3
Nitrogen production unit Reservoirs of nitrogen
3) A treatment unit with three rigid containers A, B, C having a volume of 1x25 m3 (B)
and 2x35 m3 (A, C).
Three rigid containers A, B, C
View of the loading in the container
4) A remote control system enabling control and monitoring of treatment containing: an
electric command desk housing a TSX 37 system sold under the name of
Télémécanique-Schneider, a PC computer connected to a printer, which enables the
reading and recording of the treatment parameters throughout the treatment cycle.
This installation was built by the company Mallet (division of CATS)
5) An oxygen level control unit including: A Xentra oxygen trace analyser for the
containers, two OLDHAM oxygen detectors for the area.
I-1-2 System operation
The technique chosen for oxygen deprivation is a dynamic system based on a continuous
flow of nitrogen through the enclosed treatment units. The nitrogen used is prepared from the
air in the room. The nitrogen is separated from the oxygen by the molecular sieve system.
The nitrogen is stored in a series of reservoirs. Oxygen is desorbed from the molecular sieves
under the pressure of nitrogen.
Treatment protocols are entered by keyboard and recorded in a computer file reserved for
these data: degree of humidity: 50%, exposure time:14 days, level of oxygen: 1000 vpm
(0.1%). These values can be changed. The temperature of the enclosed treatment units is the
same as that of the surrounding area, e.g. 25°C.The humidification of the enclosed treatment
units is performed through the humidification of the nitrogen.
There are three steps to the treatment cycle:
1) a purging phase of the enclosed units called "INJECTION GAZ 1" and "INJECTION
GAZ 2", which reduce the oxygen level to below 0.1%;
2) a treatment phase with an oxygen level below 0.1%, called "CONTACT GAZ";
3) a “rinsing” out phase : the oxygen level is raised to 20% by sucking in air from the
room, called "RINÇAGE", followed by the end of the treatment cycle, called "FIN
The oxygen drop times depend on the volume of the
enclosed treatment units and on the nature of the objects
loaded within them.
As a matter of definition, the exposure time Te, is considered to start once the oxygen level
has decreased to 0.1% at beginning of the phase called "CONTACT GAZ".
All the parameters are recorded and stored in files which can be accessed using Excel.
Averages are calculated on the basis of these data. At the end of a treatment cycle, a report is
produced. The latter displays the dates and times of the following events: beginning of the
treatment, moment when oxygen level reaches 0.1% (called "CONTACT GAZ"), end of the
Curves of parameters throughout the cycle of treatment
I-2 Looking for the proper exposure time: experiments on insects
Four series of experiments were performed with Hylotrupes bajulus at egg and larval
stages with the aim of determining the lowest exposure time. Each series was composed of
three experiments with identical treatment times. The eggs are placed on blotting paper and
conditioned in Petri dishes. Larvae are placed in small wooden blocks and enclosed in Petri
Each experiment was carried out in the following manner : three egg test samples and
three larvae test samples were placed in the oxygen deprivation sealed unit. One reference
egg test sample and one reference larvae test sample were left in the atmosphere of the area
outside the enclosed treatment units. The first series were treated using an exposure time Te
of 10 days. This is the shortest time given in the existing literature. After the experiments, the
test samples were put back in steamroom for 15 days. The % mortality arising from the
oxygen deprivation treatment can then be calculated after counting dead eggs and larvae and
also survivors. Results are shown in table 2
Egg test sample larvae test sample
Table 2. Mortality rates for exposure times of 10, 7 and 5 days.
H. bajulus (old house borer)
Experiment Exposure time
Test samples in nitrogen
Mortality rate %
Series 35 64
14 no hatching100
In the case of series 4, samples conditioned in Petri boxes were enclosed in sealed plastic
boxes to mimic the deep burying of insects.
The lowest exposure time at 25°C, 50% RH, with an oxygen level below 0.1% is 7 days
I-3 Oxygen drop times Ti and desorption time Td in the treatment unit
These values were calculated in the following manner, based on the automatic treatment
Ti = (time at which oxygen level of 0.1% is reached, called "CONTACT GAZ") - (time of
beginning of treatment).
Desorption times Td are determined by comparing the Ti observed when the containers are
loaded with objects to be treated with the Ti observed when the containers are empty
Td = Ti (loaded unit) - Ti (empty unit)
Table 3: Average desorption time for unit A, B and C
UnitAverage Ti of
Average Ti of empty
1day + 9 hrs ± 2hrs
23 hrs 10 hrs
C 35 m3
1day + 13 hrs ± 2 hrs
13 hrs ± 2 hrs
1day ± 1hr
11 hrs ± 1hr
The Ti and Td averages were calculated on the basis of six loadings. The load is analysed,
batch by batch and object by object, using TMS (The Museum System) files, where
information concerning the materials of which the objects are made, as well as their
dimensions and weight, are recorded.
It appears possible to infer from the composition of the different loads that a load made up
principally of wooden objects or wooden object and textile or skins, requires a priori a
longer oxygen drop time than a load of wood and vegetal pulp or vegetal fibres. However
differences in the space occupied by objects appeared to be the dominant factor. When the
volume of the closed treatment units is most effectively filled, the oxygen drop time is longer.
In contrast, it became clear that the sealed treatment units A and C, which each have a
volume of 35 m3, reach the stage called "CONTACT GAZ" (oxygen level reduced to 900vpm)
after a day and a half whereas the treatment unit B, having a volume of 25 m3, only required
II-Demonstration of the diffusion of nitrogen
II-1 Through the wrapping
Diffusion of nitrogen has been demonstrated by comparison of the oxygen level in a
cardboard box with the oxygen level in the sealed treatment unit.
Table 4 : Time taken for homogenization of atmospheres between cardboard packing boxes
and the sealed treatment units (here unit A)
after 15 minutes
after 40 minutes
after 5 hours
Treatment unit A 2113.62.6 4.5
Cardboard box 2120.6 12.8 4.5
The oxygen level is the same in the cardboard box and in the treatment unit about five hours
after the beginning of the treatment cycle.
The same phenomenon of equilibrium, here concerning oxygen concentration, occurs
between the treatment unit and the treated objects. Splits, cracks, tunnels dug out by insects,
and natural micropores in the wood can be seen as analogous to the opeining cut out in the
In order to study how oxygen concentrations vary within treated objects, a simulation
was carried out. Two wooden cases were used5, one in pine and the other in laminated wood,
which is less porous. Both had an interior volume of about 3 litres. Such a volume is a
prerequisite for the removal of air samples for later analysis. The pine case had a thickness of
7 cm all around the empty internal space. The laminated wood case had a thickness of 1 cm.
II-2 Demonstration of the degree of oxygen reduction within objects : experiments on gas
diffusion simulation with wooden cases
The degree of oxygen reduction within objects was studied through experiments carried
out in treatment unit A which is equipped with two probes for measuring oxygen levels.
Test 1: Carried out on a pine case of internal volume 3.1 litres and of thickness 7 cm.
The valve (V2) linking the case to the oxygen-analyser was disconnected at the beginning of
the experiment so as not to interfere with gaseous exchange (of nitrogen and oxygen). The
other valve (V1), which enables oxygen levels to be measured in the treatment unit, remained
connected to the oxygen-analyser.
Three days after the beginning of the treatment cycle, V2 was connected. The oxygen level
inside the box was measured after having calibrated the oxygen-analyser with the nitrogen
produced. The oxygen level was 0.0337% (337 vpm).
Five days after, the oxygen level was 0.0010 % (10 vpm).
Test 2: carried out on a pine case during a test of unit A (a) when not loaded with
objects, (b) when loaded with objects.
Oxygen levels in the box were measured every 24 hours for three days. The results are shown
in table 2.
The wooden cases were made by the company Hygiène Office
Table 5 : Change of oxygen level in the pine box compared to oxygen level in treatment unit
(a) Oxygen levels (vpm)
Unit treatment without load
(b) Oxygen levels (vpm)
Unit treatment with load and pine box
sealed with a plastic film
24 hours48 hours72 hours 86 hours
24 hours72 hours 86 hours
358 27431090 708370
Figure 1a: the evolution of oxygen drop in the pine box sealed with plastic film
starting from 21.5 %
D a t e s
Figure 1b: blow up of the low part of figure 1a
D a t e s
6 In the case of a pallet of objects sealed with a plastic film this value is 7000. Figure 3 compares oxygen levels
in the sealed pallet and the treatment unit. The great difference in the drop of oxygen tends to disappear as the
Figure2 compares oxygen levels in the sealed pallet and the treatment unit. The great
difference in the drop of oxygen tends to disappear as the treatment progresses.
Red curve : sealed pallet. Black curve: treatment unit
10102003 09:30:00 13102003 16:46:0014102003 09:30:00 17102003 15:53:00
% o x y ge n
Figure 2: Comparison between the palette and the treatment unit
Figure 3 : The drop of oxygen in the treatment unit A
Da t es
Figure 3 : The drop of oxygen in the treatment unit A
The oxygen level is measured at two point inside the treatment unit symmetric with
respect to the gas extractor. The small difference is due to asymmetric extraction of gas with
a rotating fan.
We interpret the slight increase of the oxygen level after around 2 hours to be due to
injection of oxygen from objects inside the treatment unit: due to extraction of gas, the
partial pressure of oxygen in the treatment unit becomes smaller than the oxygen pressure
inside the porous material of objects under treatment, consequently oxygen begins to be
extracted from objects and for a short time a increase in oxygen level is observed. However as
the volume of the oxygen inside the pores is very limited, the effect is small and will
disappear quickly and the exponential decrease of oxygen take over again.
This observation is very important because it proves in this process that not only nitrogen
diffuses in the porous material but also the latter loses its oxygen which helps to kill larvae
buried deep inside objects.
These figures show that the drop of oxygen depends not only on the time but also on
the porosity of materials and on the size of the objects. The different data observed as regards
desorption times, all suggest that oxygen desorption is rapid with the EPMQB anoxia
installation. Thus, eradication of H. bajulus, old house borer, requires a shorter exposure time
than those described in the literature: 7 days as compared to 10 days, in a reduced oxygen
atmosphere (less than 0.1%) with a temperature around 25°C and about 50% relative
Desorption times of 10 to 13 hours are achieved, for all the enclosed treatment units
used, for a target oxygen level of 0.09%. Desorption is concurrent with the oxygen drop in
the sealed treatment unit so that when the stage called "CONTACT GAZ" is reached, the same
conditions are reached within the most materials, as is confirmed by test 2 (a). For the less
porous ones, "CONTACT GAZ" is reached 1 or 2 days later. Homogenization between the
sealed treatment unit and the cardboard packing box is achieved after 4 hours. This means
that during the 9 remaining hours (13-4), evacuation of oxygen from the most inaccessible
spaces in the unit is taking place, e.g. in side the objects.
The tests on insects and the tests of gas diffusion cannot be performed on real objects.
Only simulations are possible. Such simulations must be as close as possible to a real
treatment situation or mimic the most difficult situations likely to be encountered. To this
end, we decided to leave closed Petri dishes in the treatment unit, placing them in cardboard
boxes under silk paper.
For the same reasons, the boxes used as models for objects to be treated during the
diffusion studies either had a thickness (7 cm) greater than generally found for objects to be
treated or a thickness close to that of treated objects but less permeable (laminated wood, or
wrapped with plastic film).
In these artificial but tough conditions, it appears reasonable to extrapolate the results,
in so far as the results obtained are reproducible and agree with those known in the literature.
These conditions appear to have been satisfied, our results being constant.
The conditions leading to the death of insects are now well known; they are
inescapable and in no doubt from a scientific point of view. According to the type of
equipment used, these mortal conditions are attained within treated objects after variable
amounts of time. Treatment stricto sensu only begins when those conditions are reached.
In the case of the EPMQB installation, the desorption phenomenon is observed both
by an increase in the oxygen level in the curve showing the evolution of oxygen levels over
time and in an indirect way by comparing the oxygen drop times of empty treatment units
with those of loaded treatment units. The longer oxygen drop times observed for loaded
treatment units are not simply a consequence of the modifications in gas circulation on
account of large obstacles (trolleys, boxes), but also result from the phenomenon of oxygen
We consider that in our system we have a binary mixture, nitrogen and oxygen.
Hydrosoluble pollutants such as CO2 and SO2 are essentially eliminated with the water in the
desiccator. Water vapour is constant and does not therefore influence gaseous O2/N2
exchange. The parameters of temperature and pressure are far removed from conditions
necessary for the liquefaction of gases. It is therefore reasonable to consider the O2/N2
mixture as a perfect gas. The mixture thus obeys the perfect gas law PV = nRT (P=pressure,
V=volume, n= number of moles , R= perfect gas constant, T= temperature in Kelvin). The
process, although dynamic, can be split up into a series of equilibrium states established
between the bulk volume of the sealed treatment units and the boxes (or objects). Following
Lechâtelier's principle, a system in equilibrium moves from one equilibrium state to another
equilibrium state in such a way as to compensate for any perturbation imposed upon it. Here,
the perturbation is the disappearance of oxygen from the sealed treatment unit.
When a substance disappears, the equilibrium shifted in such a way as to favour the
appearance of that substance:
Treatment unit N2+O2 N2 +O2 Box (or object)
When oxygen levels drop in the treatment unit, the equilibrium is shifted in direction 1 so as
to equalize oxygen concentrations.
Treatment unit N2 + O2 N2 + O2 Box (or object)
The treatment unit – cardboard box equilibrium also enables us to study the progress of the
levels of nitrogen and oxygen.
Table 6 : Partial pressures of gases
P is the total pressure in the treatment unit, and is close to atmospheric pressure. It is
recorded periodically by an automatic measuring device. The partial pressures are
calculated using the law of perfect gases. The partial pressure of oxygen in the cardboard
box is 4.5 times greater than the partial pressure of oxygen in the treatment unit. Oxygen
molecules thus naturally move in the direction cardboard box-treatment unit.
Modelling of nitrogen diffusion in porous objects enables the extrapolation of the findings
from this study to other materials and the prediction the oxygen desorption time according to
the material type (e.g. wood, feather, skin, etc.) and the size of the objects and the estimation
of the effectiveness of anoxia at the core of a treated object. We consider the dispersion
media to be constructed from a series of microscopic connected tubes through which the fluid
passes and disperses (BEAR, J. 1963). The duration for passing through each micro-tube is
random as well as the number of tubes visited by a tracer to arrive to a given depth in the
media. The distribution of passage time through N tubes for large N is Gaussian:
C0 is fluid density at contact surface; ?t is a delay time that fluid spends in a tube before it
begins its dispersion, τ is the characteristic dispersion time which depends on the microscopic
interactions and viscosity of the fluid. Defining an average tube length:
distribution (1) changes to:
For ?t = 0, (2) is simplified to:
When t/τ >> 0, (5) is a Gaussian with
All the microscopic phenomena are concentrated in the quantity:
which must be determined experimentally. For the simple geometry we have considered here
i.e. a media with infinite volume and infinite contact surface with fluid, ?2 is proportional to
x, the dispersion depth. For a finite object the dependence must be more complicated and
should also depend on the dimensions. Nonetheless, the test with a box (Fig. 3) shows that at
least for relatively large objects Gaussian distribution is very good fit. Dependence on x
however must be verified experimentally.
Fig. 4: Evolution of oxygen fraction inside the box.
This calculation will enable us to predict the oxygen desorption time within the objects
according to material (e.g. wood, feather, skin) and the size of objects and establish the
effectiveness of anoxia at the core of a treated object.
Another key question is whether the treatment conditions chosen enable the eradication
of any mutant insects present which may have become resistant to oxygen deprivation.
Several workers in the field have already looked into the question of whether insects can
adapt to modified atmospheres with very low oxygen content. They have shown that it is not
impossible for insects to survive without oxygen but this is very rare, and encountered above
all in the case of aquatic insects having water or ice habitats, for example certain small flies
living in water. The latter are able to resist more than 100 days in the absence of oxygen by
slowing down their respiration, increasing the amount of stored oxygen and also by a
physiological change which allows them to avoid drying out. In very humid conditions, even
in a reduced oxygen atmosphere, desiccation, the primary cause of death for insects, does not
occur (ZEBE, E. 1991).
However, in conditions which are far removed from aquatic conditions, such as in the food
industry, or in heritage institutions, such an adaptation is unlikely given that the relative
humidity is much lower. Resistance may arise in the case of insects which develop in
confined and very humid spaces where the oxygen concentration is low and the carbon
dioxide concentration is high, or with insects buried inside books or wood. A known example
is that of L. serricorne, the cigarette beetle. In such a case, it is sufficient to lengthen the
exposure time (SELWITZ, C., 1998).
A study of the resistance of Tribolium castaneum (Red flour beetle) to a reduced oxygen
atmosphere was carried out using the following conditions: 0.5% oxygen in nitrogen, or 20%
oxygen and 15% nitrogen in carbon dioxide with 95% relative humidity, these conditions
being maintained until only 30 to 50% of the insects were still alive. The surviving insects
were bred for 40 generations in the same conditions. This study showed that these insects
were only resistant towards the specific atmosphere to which they had been subjected. Insects
returned to normal air after 13 generations displayed continuing resistance for 8 generations.
The authors of this study considered that adaptation to atmospheres modified with carbon
dioxide is more likely to occur than in the case of modification with other gases. More
recently, in 2001, HOBACK and STANLEY studied several microhabitats where insects are
under hypoxia or anoxia such as stored grain or decaying wood. In such conditions insects
reduce their respiration rates when the oxygen level reached 10 %. Certain insects as C.
vomitoria larvae can survive at 1 % only 5-6 days. However, in general, insects cannot
withstand an oxygen level of 0.5 % for a long period of time.
It would not be realistic to consider that the simple fact of subjecting aerobic insects to
an anoxic environment is sufficient to make them become anaerobic species.
The rice weevil, whose resistance seems to have given rise to the standardized
treatment time of three weeks, is an important pest in the agricultural industry. This insect is
not among those frequently encountered in museums. In museums, the insect population
is essentially composed of Anobiidae, Dermestidae, Tineidae, and other families such as the
Lyctidae and Lepismatidae (MAEKAWA, S., 1998, PINNINGER, D.). These insects are much less
resistant to oxygen deprivation treatment as is indicated by the results of studies involving
The conclusions we draw after the study carried out with the EPMQB installation are
based both on the results of teams abroad working in this field and on our observations at the
Musée du quai Branly.
Temperature plays a crucial role for hastening the death of insects found within
objects. Thus, at a temperature of 25°C, it is entirely possible to reduce exposure times to 10
or 15 days for the insect species commonly found in museums.
The role played by humidity is less clear-cut in spite of the fact that the principal
mechanism leading to insect mortality is desiccation, both for larvae and for adult insects.
In view of the results obtained with the EPMQB installation and also in view of those
recorded in the relevant literature for tests on reference samples and on real objects (such as
the results of the Getty Conservation Institute), and taking account of the experimental
conditions (temperature, relative humidity, oxygen levels) we can move as of now to an
exposure time (Te) of 14 days (2 weeks).
The oxygen drop times (Ti) being situated between 1 and 2 days for most objects, this
corresponds to a treatment time Tt between 15 and 16 days.
Tt = Ti + Te
In parallel with the treatment of objects, the rigorous hygiene monitoring programme
put in place on the collection treatment site will enable any new infestation to be detected.
For this monitoring programme, the Musée du Quai Branly has called in specialists in
tackling infestation. The hygiene monitoring programme involves the setting up of
pheromone and baited traps and these traps are renewed every three months.
A programme of harvesting and identification of insects has also been initiated. A list of all
the insects found dead or alive, in the treatment sites or on objects is kept by the
Cleaning/Dusting Department and transmitted to the Anoxia Department, who are
responsible for identification, performed in collaboration with the company Hygiène-Office
and the Laboratoire d'Entomologie du Muséum d'Histoire Naturelle6.
BANCK, H., J, ANNIS, P., C., "Suggested procedures for controlled atmosphere storage of dry
grain”. Commonwealth Scientific and industrial Research Organization ( CSIRO)
Division of Entomology Technical Paper, 13.
BEAR, J., “Hydrodynamic Dispersion”, 1963
KIGAWA, R., et al “Practical methods of low oxygen atmosphere and carbon dioxide
treatments for eradication of insect pests in Japan” in proceedings of 2001: Integrated Pest
Management for Collections . A Pest Odyssey. 1-3 October 2001, chapter thirteen.
HOBACK, W.W; STANLEY, D.W. “Insects in hypoxia”, J. of Insects Physiology, 2001, 47, 533-
PINNINGER, D., “New Pests for old : The changing status of museum insect pests in the UK” in
proceedings of 2001: Integrated Pest Management for Collections . A Pest Odyssey. 1-3
October 2001, chapter three.
RUST, M., VINOD, D., DRUZIK, J., PRESSEUR, F.,“ The feasability of using modified atmosphere to
control insect pests in museum”, Restaurator, 1996, 17, 43-60.
SELWITZ, C. , MAEKAWA, S., “Inert gases in control of museum insects pests”, Ed. The J. Paul
Getty Trust, 1998.
VALENTIN, N., 1993,“Comparative analysis of insect control by nitrogen, argon, and carbon
dioxide in museum, archive and herbarium collection”, International Biodeterioration et
Biodegradation, 1993, 32, 263-278.
VALENTIN, N., PREUSSER, F., “Insect control by inert gases in museums archives and archives”
Restaurator, 1990, 11, 22-33.
We thank Mr. Stéphane Martin, Director of the Musée du quai Branly, for entrusting us with
We also thank the company Hygiène Office for the wooden boxes used in the gas diffusion
experiments and the company Mallet for agreeing to modify the installation in line with our
needs for this research programme.
Entomology department of the Natural History Museum, Paris)
We thank Mr. David Pinninger for his assessment of this study, which enabled the Musée du
quai Branly to decide if and how it should be implemented.
We thank Mr. Jean-Pierre Mohen, Director of C2RMF7 for his advice on this project.
We also thank Mr. Jason Hart-Davis for help with the English of this report, and thanks also
go to the photographic team of the Musée du quai Branly, and in particular to Ms. Stéphanie
Centre de Recherche et de Restauration des Musées de France (French Museums Research and Restauration