EU PROJECTS‎ > ‎Craft West‎ > ‎



3.1 Introduction to the process

The Arkè process for the consolidation of waterlogged wood is based on three main steps:

✔  an impregnation treatment of the wood sample with aqueous solutions of starch and eventually other polysaccharides, to reintroduce in the wood structure new components (similar to the already washed away cellulose) with a binding function;
✔  a quick thermal treatment at high temperature and high pressure, that will allow the polymerisation of the polysaccharides inside the wood structure and will remove the still existing microbiological part (eggs and grubs of insects, bacteria, fungi, etc.);
✔  a desiccation treatment mild and efficacious, able to eliminate all the free water from the structure avoiding its collapse and favouring the consolidation of the new introduced binders.

This treatment has been set up initially by the partners Contento Trade e Gradient LMTAI and it has been perfected during this research thanks to many lab tests on small samples obtained from wood finds without artistic value.
Now we’re going to analyse more in detail the research activities on waterlogged wood carried out in the WEST project, in order to optimise the Arkè process.

3.2 Waterlogged wood samples characterization

3.2.1 Origins of the samples

The samples of waterlogged archaeological wood comes from different countries. Some of them are coming from the Museum and Centre Camlle Jullian of Marseille (France). They were found in the wrecks of Madrague de Giens (659), of Dramont (A, C abd E) and of the Cavaliere. The identification analysis of the species showed that the ships were built with different wood species as the pine (maritime, pinea, sylvestris, halepensis, leucodermis), the quercus (called also oak) and the Abies alba (called also silver fir).
Other samples were furnished by the Spanish museum (Culture department of Catalunya). In that lot, there were some samples deriving from some archaeological objects as the one called Culip (number III, IV, VI) date 1300 AD and samples coming from the ship La Draga Neolithica dated 5000 BC. The two wood species La Draga Neolithica were the softest among all we studied and tested.

Other samples finally came from an Italian wreck, la Julia Felix from Grado and a big sample from Ercolano.
To facilitate the study operation we chose some codification for recognizing the provenience and the number of samples for each wreck. Some codes are only the abbreviation names of the objects the samples are coming from, others have a number that the sender gave before sending the object to us. Samples dimensions

In the different studies related to the treatment of the waterlogged wood objects with different synthetic resins (PPG; PEG; etc), natural resins (animal glue), sugars (mannitol, sorbitol, saccarose), etc. and different dimension of samples have been used.

Young and Sims (1981) had already indicated that the samples coming from the part of the wood near the tracheide or of the fibre can be strongly impregnated. This is one of the obtained conclusions by the authors on the microscopic observation in a study on PEG fixed by cobalt thiocyanate.
Consequently the author recommend to use the wood samples having an axial length much bigger than the tracheide dimension. The fibre and tracheide dimension vary with the space and type of wood. The fibres can vary between 0.18 and 3 mm of length, while the tracheides of the soft wood can reach the 9.5 mm of length.

In the Table 1 are grouped some dimensional values found in the state of art related to the treatment processes of the waterlogged woods. It’s possible to say that there is a notable difference on the sample size It’s true also that often the samples dimension depend on the disposability of wood. On the contrary of tests performed with the fresh wood, in which the samples dimension were standard, it’s impossible to do the same with waterlogged wood samples. The treatment experiences for conservation have been realized on samples directly taken from the ship wrecks.





Cook & Grattan (1984)

10 - 20

20 - 100

20 - 100

Nishiura & Imazu (1990)




Scano & al. (1994)




Kazanskaya & al.(1990)




Grattan & al. (1980)










Hoffmann, P. (1984)




Watson, J. (1984)


ø 110


Table 1: Dimension values for samples from literature (in mm).

 We remark that for the fresh wood samples of small dimensions have been used for testing the different methods. Hereinafter in the Table 2 are reported some dimension found in the state of art for fresh wood samples. 





Seborg et al.(1953)











10 - 76

Bonneau (1991)


25 – 43 – 50


Hernandez (1994)


27 - 54


Table 2: Dimension values of the fresh wood sample from literature (in mm).

A comparison between those two tables show there isn’t notable difference between the waterlogged wood samples and the fresh wood.
Often the archaeological objects have different dimensions. More, the big wrecks of the ships are realized with different types and species of wood. And this causes another difficulty in the wood sampling. Generally it’ll be difficult to sacrifice an object because it has not the right dimensions, as the expense for a ship excavation is so high that nothing is wasted.

In the different experiences realized we have mainly been inspired to the values found in the literature concerning the sampling. So we shall say that the size of the samples used for the tests is related to the availability of samples of different wood species. The reason is that some samples they sent us were big, some others were small. The efforts devoted to the obtaining of uniform wood size imply a loss of material. This is due to the wood fragility. The wood tends to subdivide in lamina and breaks easily during the cutting and during the permanence in water. Preparation and methodology

The samples obtained cutting the wood with an electric saw have often a parallelepiped shape. The dimension on the three axes are: a (longitudinal or axial) * t (tangential) * r (radial) and are generally ranging between (20-40) * (20-30) * (20-40) mm. Other bigger samples have dimension (a = 90, t = 60, r = 70). The dimension depend on one side on the quantity of wood at disposition, on the other hand on the mechanical resistance tests to be performed to evaluate the efficacy of the treatment. After the cutting, the samples are simply washed under tap water with precaution. Because of their age, this operation is done manually. The samples are then put in the distilled (or deionised) water. They stay in the water, in a polystyrene bag, for minimum two-three days (the most part for at least two weeks) before being impregnated.

This cleaning is not sufficient to wash the impurities that are inside the pores at a depth of 3-4 cm. In fact often some pores full of sand are found inside the wood after the cutting of the samples. So any external cleaning can’t completely eliminate the impurities without deteriorating the wood. But it’s important to notice that the infiltrated impurities, as sand and substances left by insects larva, seem to bring a contribution to the reinforcement of the mechanical structure of the wood.

3.2.2 Characterisation of the wood samples

Many tests demonstrated that the knowledge of degradation state of the wood constitute an important stage for the choosing of the concentration of the consolidating material. Some authors insist on the realization of chemical analysis to evaluate the lignin and carbohydrates content (Hoffmann & Jones (1990), Hoffmann (1982), McLeod (1990), De Long (1972, 1977). Other authors have demonstrated that the determination of the volumic mass and of the maximum water content of the waterlogged wood are sufficient to determine its degradation stage (Grattan & Mathias (986), Kazanskaya (1990). For this last point, Kazanskaya et al (1984), have largely contributed through a physical and chemical study on different archaeological wood species by infrared spectroscopy. The obtained correlation allow us to assert that the maximum water content (MWC) is a sufficient parameter to characterize the degradation stage of the wood. Chemical analysis

Some analysis have been performed to determine the chemical constitution of the different samples of waterlogged wood. Two centres have performed these analysis: the Institut du Pin de Bordeaux (France) and the Centro Restauri (Italy). The chemical analysis on the waterlogged samples of wood have been performed according to the standard method TAPPI. Only the ash content has not been tested in our study. Humidity analysis

The maximum water content (MWC) in different samples of waterlogged wood has been determined. Two methods have been used to assure the goodness of the results.
The first determination method, more used, consists on the heating in an oven the samples at 105°C for 10-12 hours. For the fresh wood the AFNOR norm foresees a minimum time of 8 hours for the samples of the cut wood of shape of cube of 1 cm and heated at 105°C. These dimensions are difficult to be respected with the waterlogged wood. Anyway we have tried to respect the dimension with the first samples.

The second, more rapid, is the method for the determination of the water content consist in heating the wood with the infra red rays. We have used a plant called Desiccator LP17-Mettler PM100 coupled with a printer Mettler Tolledo and on another one, Sartorius GmBh Gotingen, Type YTC01L N°-37030104. These instruments have some advantage compared to the oven. It’s more rapid buy mainly the infra red ray penetrated more in depth for the small elements: more, the measurement is completely automatic. The disadvantage is that it can’t be applied to one sample alone, like the procedure by oven, but a considerable number of samples together shall be tested.

It’s important to underline that some norms don’t indicate very well how to use the infra red analyser. More, specific norms to test the waterlogged wood are not ready jet. Both the two methods are destructive for the wood. Determination of the volumic mass

The determination of the volumic mass of a porous material is still under discussion because of difficulties in the methods of calculating this value. Also for wood material a completely satisfying methods has not been developed. The anisotropy of the wood in the three directions (axial, radial and tangential) is also a cause of this difficulty. This anisotropy is more accentuated in the waterlogged wood because of the heterogeneity of their degradation. We determined the volumic mass of the samples with two methods cited by the literature. Analytical method 1

This first method is only based on the use of a relation found for the fresh wood (Smith 1954) and adapted to the waterlogged wood by Tarkov and Feist (1966) that showed that it’s possible to find directly the real volumic mass of the waterlogged wood starting from thee maximum water content.
Other authors (Cook and Grattan, 1990) confirmed the efficacy of the proposed relation, introducing, in the formula, the hypothesis that the density of the dry material of the wall cell of the wood is 1.50. This value is frequently cited in literature. It has been deduced by the results obtained for the most part of the fresh wood species. Analytical method 2

This method is based on the fact that all the waterlogged wood are “completely” full of water. A system to demonstrate this is i.e. the fact that all the waterlogged samples showed that, even cut in small pieces of some mm, when they are suddenly let tumble in the water, they fell down directly on the bottom of the container, while a trunk of fresh wood of dozens of tons floats. It will be more indicated to measure this parameter with a method based on a determination in water than within another liquid that could prevent the reuse of wood sample for other kind of tests. Consequently the adopted method is based on the Archimede’s principle. For the measurements we have used a Sartorius balance with centimetres, a test tube made of borosilicate glass type Simax of 250 ml, of internal diameter of 35 mm graduated at 2 mm and deionised water. The used method is the following.

✔  The wood samples have been measured by a decimal calibre and weighted. The test tube has been weighted empty and then filled with water till a determined level (150-200ml). The volume of the water is obtained by reading and also the mass of the ensemble.
✔  The sample of waterlogged wood is introduced and the volume and the mass are again measured to be used for the calculation of the volumic mass and of the bulk density.
✔  The sample volume is determined also with a second measure. This last method consists on the calculation done starting from the before measured dimension on the wood samples.

The deviation registered in the values obtained by calculations and those obtained by readings have been quite small (less than 5%). This is the evidence of the good affinity existing between the two different approaches for the volume determination.
The mass and the volume of the waterlogged wood sample allowed to determine the apparent volumic mass. The water mass used and the corresponding volume were measured in order to determine the volumic mass of the water, directly during the test. These two parameters have been useful to calculate the apparent density of the wood samples. So we have decided to establish a relation between the apparent density and the searched conditional density.

The maximum water content, finally, was useful to calculate the correction factor for the determination of the conditional density of the dry wood. It has been calculated starting from the above mentioned relation, expressing as follow a proportionality between two types of densities:


Rgc: conditional density of the dry wood
Rga: apparent density of the waterlogged wood
MWC: maximum water content of the wood (expressed in percentage)

We note that the apparent wood density, measured when it is in waterlogged conditions, is higher than the real density of the same wood without water. It’s higher than the density of the water used for the measurement. So the waterlogged wood sample, even if is small, deposit itself on the bottom of the test tube full of water during the measurement. It’s important to underline that this water saturation is the indispensable condition for the application of the method. This saturation doesn’t constitute an hypothesis but depend from a typical characteristic of ancient waterlogged wood.

More, the method using the Archimedes principle is not useful for the determination of the volume of the fresh samples of wood. In fact the fresh state of the wood (green wood) implies some holes full of gas (CO2 and N2, etc). The gas presence prevents the wood sample from a total immersion in the water. The values of the such obtained conditional volumic mass have been compared to the ones found with the first method. The obtained results show that the two methods give values of the same size order, even similar, with a small difference for the less degraded wood. This allow us to say that the two methods are comparables. The second method has an advantage: it’s not necessary to formulate the hypothesis of the constancy of the volumic mass of the substances of the wall cell, that we consider equal to 1.50.

It’s sufficient that the measurements of the samples and the readings of the water volumes are correctly performed to get the searched density. Analysis of the structure by SEM

There are different methods for the determination of the wood structure. The main instrument for all of them is the microscopic examination (Green, H.V. 1965). It’s an important medium to obtain some information on the level of degradation, also when the volumic mass and the wood species are unknown. In our research , the analysis with electronic microscope (SEM) has been performed on a SEM-EDS plant of the type JEOL JSM-5410LV/EDS ISIS LINK 300. From some samples of wood impregnated with starch we have taken some small samples of a section of about 5*5 mm2.

We have performed some analysis of the microscopic structure on some samples:

✔   Before the wood impregnation;
✔   After impregnation to understand the degree of penetration of the starch;
✔   After the final thermal treatment to study the effects of it on the wood;
✔   Finally, an observation after the drying of the treated wood.

The plant JSM-5410LV has also been equipped with a Spectrometer type at Energy Dispersion (EDS), to have a qualitative analysis of the samples, and a quantitative analysis of the constituents. The spectrometer of type EDS is not exactly the best instruments for a rigorous quantitative analysis, but it gave a global overview on the chemical elementary constitution of the samples.
In this research micro structural analysis has played only a complementary role, because the mainly used characterization method has been the MWC parameter (cheaper) but these analysis have been used in order to help the researchers in the interpretation of the obtained results during the work plan development.

3.2.3 Results of the characterization test on raw waterlogged wood samples Main categories of waterlogged wood samples 

The forty samples of waterlogged wood subdivided in several samples in order to be used in our research has been grouped in three main categories:

✔  The ultra-degraded woods; these are very soft, spongy to the touch;
✔  The degraded wood; these are soft but not spongy and represent the main part of the samples analysed;
✔  The mean degraded wood; these are relatively hard, slightly deteriorated.

We note that some of the analysed samples, for their irregular behaviour, can be identified as another wood category. This category will mainly be constituted by not homogeneous samples containing both soft and hard parts. This tendency can be explained by the fact that in some of the waterlogged wood samples, the level of degradation varies according to the axial length of the wood. But in some other samples the repartition of soft part and hard parts is completely irregular.
This behaviour has been encountered especially when the heart of the sample concerned has been attacked by xylophages insects or termites. Study on Maximum Water Content (MWC)

A first attempt of classification of the waterlogged samples has been realized on the basis of the maximum water Content (MWC). 
We have identified 3 categories taking into account that, for some species of fresh wood samples, the MWC value can range from 40-50% [Timell, (1975); Thomas, (1977)] till to 75% [Rosen, (1987); Baixeras, (1995)]. The selected classification is corresponding to the methods generally cited in literature [De Jong, (1977); Jones & Rule (1990)] and the differences are limited to the values chosen to limit each category.

The categories identified on the basis of MWC are the following:

✔   For values of MWC > 500%: Ultra - degraded wood. The water content of some samples can reach and also pass a value of 1000%. Typical examples of this kind of wood are the samples coming from the boat Grado Jula Felix Grande (GJF-G), samples coming from Spain, probably of fig, codified as ECBL and the samples from the boat La Draga Neolithica codified as DN5-BC. In this category are also included the samples coming from the boats Dramont et Cavalière;
✔   200 < MWC < 500 %: degraded wood;
✔   MWC < 200 %: mean degraded wood; the samples included in this category are few and could also be included in the pervious category, changing a little bit the MWC value chosen for the limit. The inclusion in this category is better motivated by the samples structure and by their relative solidity, that is considerably higher than the previous category.

We note that the classification based on MWC is relatively easy to exploit even it can bring some interpretation problem. Effectively the Maximum Water Content can not be the only element taken into account to evaluate its degradation level. We have seen in the literature that there are fresh woods that are classified as soft. The presence of this kind of essences among waterlogged wood samples can drive to irregular results in the consolidation process. Moreover, also the waterlogged samples included in the fourth category (not homogeneous), can not be classified using the MWC value only.

Thus we can conclude that the MWC value is an important parameter for waterlogged wood classification but to obtain satisfying results we need to combine it with other characterization tests.
For this reason, the volumic mass of the waterlogged samples concerned as been also determined. Study on conditional volumic mass and conditional density

The volumic mass has been proposed for the classification of waterlogged wood [Barbour, 1984]. In our research, we have demonstrated that this parameter is not sufficient to obtain the samples characterisation because of the not homogeneous level of degradation in the transversal and radial direction. Moreover several waterlogged wood samples, among those provided for this research, are characterized by a similar volumic mass but haven’t the same level of degradation.

Only comparing the considered samples with fresh wood samples of the same essence, without taking into account the aging and related parameters, a classification based on volumic mass can be useful.
In any case the classification obtained in this research on the basis of this parameter (Table shows results similar to many literature works carried out on this field [Amu, (1982); Barbour, (1984); Wieczorek & al. (1990)].

In fact we can conclude that the determination of the volumic mass according to the classic method (method 1) is not a sufficiently precise parameter to evaluate the degree of wood deterioration [Barbour, 1984]. We supposed that a consistent part of the error comes from the assumption of a fixed value for apparent wood density of any source (= 1,5 value calculated on the basis of the Smith relation). Thus we developed a new method (method 2) for the determination of the conditional density of waterlogged wood based on direct measures.

Well consisting with those available on the literature, the results obtained with this new method are more precise.
Finally we can observe that the combined use of the maximum water content and of the conditional density can improve the classification but can not eliminate at all the possible errors; these are mainly due to the strong correlation between these two parameters. The chemical analysis will thus allow to optimize the classification. 

Table Chemical analysis

In the Table are reported some results of the chemical analysis carried out on the waterlogged wood samples used in this research.


On a preliminary evaluation of the performed analysis, the extractable elements content should be related to the level of wood degradation. More is the wood degraded, lower will be its content of extractable elements. The literature is more insisting on the relevant decreasing of the cellulose content in the wood material, during its deterioration process.

The results obtained seems to confirm both these two assumptions, better as concerns the extractable elements (Figure 1) than cellulose (Figure 2), while this evolution is not useful in the case of lignin (Figure 3). We can also note the weak correlation between the maximum water content and the chemical composition. The chemical composition seems not only to reflect the degradation level expressed by MWC, but can also give other interesting information on the wood structure and on its potential durability. In any case, the complexity of these chemical analysis, corresponding to an high price, limited considerably the systematic use of this parameter in our characterization work.

 Figure 1: Correlation "maximum water content" - "extractable elements".

 Figure 2: Correlation "maximum water content" - "cellulose content".

 Figure 3: Correlation "maximum water content" - "lignine content".

We have to underline the great difference in the results of the analysis; the chemical composition of the samples is characterized by an «measure error» that is strongly variable, also according to the sampling point within the same sample; this fact justifies the measured great heterogeneity of results.

As conclusion, the differences in the evaluation of the level of degradation of waterlogged wood according to the chosen characterization method (chemical analysis, MWC, conditional density) are justified, from one side by the intrinsic heterogeneity of the wood degradation, on the other side by the few amount of wood material that is possible to destine to the wood characterization.

Similar results have been obtained using microscopic analysis confirming the complexity of the problem of waterlogged wood characterization. SEM analysis

Scanning electronic Microscopy SEM in a powerful mean to access the internal structure of a material. The different photos realized by SEM witness the great efficacy of the method. In spite of its performances, the problem posed by the water present in the wood structure is not easy to be solved, because of a technological aspect that can induce a great interpretation error.
The optimal conditions for a SEM analysis are linked to the material capacity of electrons emissions. Effectively, the basic principle used in the SEM is the analysis of the spectrum of the electrons emitted (forming the analytical spectrum) by the material, following to its radiation with a electronic excitation beam.

To obtain a stable electronic emission, the material to be analyzed should be dry and the operation should be realized under high vacuum. The impact of these condition on our samples can be strong.
A waterlogged wood sample, not impregnated, submitted rapidly to this high vacuum level (as requested by the analysis) can undoubtedly undergo to deformations. They are very little for some kind of wood but can arrive to shrinkage, to cracking, till to the separation of the layers of the cell walls for other kind of wood.

The photos presented here are taken from the area exposed to the electronic beam (about some microns), but the results interpretation will be used for the characterization the entire wood specimen.
This characterization problems regards especially the ultra degraded wood samples, very rich in water, and particularly the samples having a MWC value higher of 700 - 750. The wood samples codified as Cp III, Cp IV, Cp VI, DN5-BC, GJF, that are very soft on touch, are good examples of a critical material.


Figure 1: (left figure) SEM photo of a waterlogged wood sample: 15 E (Radial view presenting the pores of various size and the parenchyma).
Figure 2: (right figure). SEM photo of a waterlogged wood sample: 6 E (Axial view presenti the high degradation level of the specimen).

     Figure 3: Wood sample coming from Draga Neolithica of 5000 B.C. (Radial section).

On the photos of the Figure 3 and Figure 4, we can evaluate the external borders of different waterlogged wood cells that, appears regular as the fresh wood ones. We can thus conclude that, only examining SEM photos is not possible to determine if the wood is fresh or waterlogged. This observation is also valid as concerns the right area of Figure 5, where we can observe cell borders of variable thickness. The rapid application of high vacuum can have important effects both on waterlogged wood cell and fresh wood cells. The shrinkage that can arrive till to the separation of the layers of the cell walls is not an indication on the degradation level. Finally, the only element that allows to identify very well the wood degradation level is the total or partial disappearing of the layers that are forming the cell walls (i.e. the right area in Figure 3 or the left area in Figure 4).

As conclusion we can state that the SEM analysis can take part to the identification of the degradation level of a waterlogged wood sample in combination with other more useful analysis like MWC, conditional density, chemical analysis.


Figure 4: (left figure) SEM photo of a radial section of the wood sample Culip Cp III, IV et VI (dated 1300) exposing the fibers of this kind of wood.
Figure 5: (right figure) SEM photo of the radial section of the sample ECBL. Conclusion

As conclusion of this study on the characterization of the different waterlogged wood samples, we aim to underline the great diversity of their structures, resulting from the wide variety of wood species sampled and from the high heterogeneity of their degradation. Even within a single wood sample the degradation level is not homogeneous!

The aggregation of the results obtained, including those related to MWC, conditional density, chemical analysis and microscopic analysis, guarantees a good level of information on samples degradation. But we have still to solve the problem of the sample classification in order to define a consolidation process that meets the quality requirements of the archaeological sector.
A possible solution to this problem will be presented in the following chapters of this report.